ProgrammableMarketing/weka.io (opens in a new tab)
PDF

Views

  • Plan
  • Roadmap
  • Research

On this page

  • The brief
  • Start here
  • Idea portfolio
  • Effort vs impact
  • This week · 6
  • This quarter · 12
  • Big bets · 3
  • Wildcards · 2
  • How we can help
Generated Oct 22, 2026
  1. Programmable
  2. Playbooks
  3. WEKA
WEKA logo

growth playbook · Oct 22, 2026

weka.io (opens in a new tab)

WEKA growth playbook

Diagnosis

WEKA has moved its message from AI storage to inference economics, and its hiring shows how it plans to win: an inference product marketer to quantify cost per token, a competitive intelligence lead, an editorial manager focused on search and AI citations, a China-based technical marketer for APJ, partner marketing for MDF and neocloud co-marketing, and sales hires across the US, Europe and Asia.

The site has not caught up. The cost-per-token estimate needs a 30-minute call, the self-serve calculators answer older questions, the inference pages show zeros to crawlers, the homepage title still says AI storage, and the AI clouds that run WEKA are missing from the partners page.

This playbook fixes those first, then builds comparison pages, measured customer stories, localized benchmarks, OEM pages and a public cost-per-token benchmark.

Growth Score

7/10

Biggest Opportunity

Own the cost-per-token conversation: a calculator anyone can use, numbers machines can read, and a quarterly benchmark the market quotes.

Biggest Risk

VAST Data, DDN and Hammerspace all position as the AI data platform, so if WEKA's inference economics are not easy to see and check, buyers will compare on brand and price.

01

Start here

The three moves to make first. Each one is written up in full below.

  1. 1

    A self-serve cost-per-token calculator, so inference buyers do not need a 30-minute call to see the number

    Write the formula from the OCI benchmark and test it on three known customer setups.

    High impact First result in 1 month
  2. 2

    Put the real numbers in the HTML: the inference pages tell crawlers "0x more concurrent users"

    View the source of the AI Inference page and list every stat that reads zero.

    High impact First result in 2 weeks
  3. 3

    Rewrite the homepage title and description for the inference message

    Draft three title options and pick one.

    Medium impact First result in 2 weeks

02

Your idea portfolio

23 ideas, grouped by when to run them. Open any idea for how to run it: the steps, who to approach, copy you can use, and how to measure it.

  • 8 ideas are marked Unexpected: plays you are unlikely to have tried.

Every idea placed by the effort it takes against the impact it could have. Quick wins sit top left; select a number to open that idea. The filters below narrow the list by status, timing, category and what Programmable can build.

Effort vs impactEach number is an idea. Select one to open it.
Impact →
Quick wins · 3Big bets · 9Fill-ins · 8Reconsider · 3
Effort →

Showing 23 of 23 ideas

This week 6

Quick wins with a clear first action.

  • 01

    WEKA's new inference product marketer is asked to "quantify WEKA's economic value across GPU utilization, time-to-first-token, throughput, concurrency, tokens per GPU, and cost per token". The site already promises that number, but only after a form and a 30-minute call ("we'll send you an estimate of what your cost per token will be"). The two calculators that are self-serve answer older questions: one prices WEKApod storage for a SuperPOD from a GPU type and count, the other uses cloud list prices "as of May 30, 2023". Neither mentions tokens.

    High impactMedium effort First result in 1 month Senior Product Marketing Mana…

    First action

    Write the formula from the OCI benchmark and test it on three known customer setups.

    How to measure it

    Inference cost-analysis meetings

    Now
    –
    Target
    2x current
    By
    90 days

    Steps

    1. 1

      Agree the model

      Inputs, outputs and the benchmark ratios behind them.

      Senior Product Marketing Manager, AI Inference 1 week

    2. 2

      Check the method

      Engineering signs off the assumptions and ranges.

      Principal Product Manager, Augmented Memory Grid 3 days

    3. 3

      Design

      One screen: inputs left, cost per million tokens right.

      Web team 3 days

    4. 4

      Build

      On the AI Inference page, with results shown before any form.

      Web team 2 weeks

    5. 5

      Pass inputs to the form

      Prefill the 30-minute analysis request.

      Marketing Operations 2 days

    6. 6

      Retire the old pair

      Point the SuperPOD and 2023 cloud calculators to the new one.

      Web team 1 day

    7. 7

      Track

      Runs, inputs chosen and meetings booked.

      Marketing Operations weekly

    First result
    1 month
    Cost
    Engineering and design time
    Owner
    Senior Product Marketing Manager, AI Inference

    Smallest test

    Publish a static table of cost per million tokens for three model sizes, with and without Augmented Memory Grid.

    Why it works

    Build an inference economics calculator on /solutions/ai-inference-acceleration: inputs for model size, context length, concurrent users, GPU type and count, and current KV-cache setup; outputs tokens per GPU, time-to-first-token and cost per million tokens with and without Augmented Memory Grid, using the published OCI and CoreWeave benchmark ratios with the method shown. Show the result immediately, then offer the 30-minute analysis to check it against the visitor's own cluster, with their inputs passed to the form.

    Inference buyers already think in cost per token, and a number they can play with earns the meeting instead of being the price of it. Every run also tells the inference marketer which models, context lengths and GPU fleets prospects care about.

    Ready to use

    ChecklistCalculator inputs and outputs

    [ ] Model size (8B / 70B / 405B / custom)
    [ ] Context length and prompt reuse
    [ ] Concurrent users
    [ ] GPU type and count
    [ ] Current KV-cache tier (HBM only / DRAM / NVMe)
    [ ] Output: tokens per GPU, time-to-first-token, cost per million tokens
    [ ] Method note citing the OCI and CoreWeave benchmarks
    [ ] "Check it with an engineer" button prefilled with the inputs

    Risks and how to handle them

    • Numbers read as a promise

      Show ranges, the benchmark source and a "your results will vary" note.

    • Competitors copy the inputs

      The value is the method and benchmark data, which only WEKA has.

    Evidence

    • HiringSenior Product Marketing Manager, AI Inference (posted September 28, 2026): "Define and quantify WEKA's economic value across GPU utilization, time-to-first-token, throughput, concurrency, tokens per GPU, and cost per token. Build the economic case customers need to justify investment." The role also recruits "design partners and early adopters", owns the proposition for AI clouds that should make WEKA "the storage standard for the category", and prefers to "show specific numbers and named customers than make broad claims".Source
    • Your websiteAI Inference Cost Analysis page: "In a 30-minute call, we'll identify where your inference costs are highest"; "Once we get a snapshot of your infrastructure, we'll send you an estimate of what your cost per token will be on NeuralMesh"; proof points "Stability AI reduce their storage costs by 95% on a per TB basis" and CoreWeave benchmarks with "up to 4.2x" token throughput per GPU.Source
    • Your websiteThe two calculators on the site: the TCO Calculator ("See how much you can save by using WEKApod as the storage for your SuperPOD deployment") takes only a GPU type (H100/200, B200, GB200) and a GPU count and shows two totals, e.g. "$73,203,001 WEKA" against "$108,203,001 Other", and has no meta description; the Cloud Savings Calculator says "Results are based on publicly available list prices as of May 30, 2023" and asks for a form before showing savings. Neither mentions tokens, time-to-first-token or cost per token.Source
    • News coverageWEKA newsroom: NeuralMesh 6 (July 21, 2026) cites OCI benchmarks of "10x higher token throughput, 10x more concurrent users served, and 7x more tokens per GPU"; WEKA raised a $140 million Series E at a $1.6 billion valuation.Source
  • 02

    The editorial manager is hired so WEKA "earns accurate citations from LLMs", and robots.txt invites GPTBot, ClaudeBot and PerplexityBot in. But the stat blocks on the AI Inference and AI Token Efficiency pages are count-up animations that start at zero: the HTML those crawlers read says "0.0x higher sustained input-token throughput", "0x more concurrent users" and "0 seconds" for Cohere. The real figures (6.5x, 10x, 267%) only appear in script data.

    High impactLow effort First result in 2 weeks Programming & Editorial Manag…

    First action

    View the source of the AI Inference page and list every stat that reads zero.

    How to measure it

    Correct stat citations in AI answers

    Now
    –
    Target
    10 of 10 test prompts
    By
    60 days

    Steps

    1. 1

      Find the component

      Every page using the count-up stat block.

      Web team 1 day

    2. 2

      Fix rendering

      Server-render the final number.

      Web team 2 days

    3. 3

      Respect reduced motion

      No animation when users ask for none.

      Web team 1 day

    4. 4

      Re-crawl

      Request indexing for the inference pages.

      Search and GEO specialist 1 day

    5. 5

      Test answers

      Ten prompts about WEKA inference results across assistants.

      Programming & Editorial Manager 2 weeks

    6. 6

      Add to the publishing checklist

      Numbers must be readable without scripts.

      Programming & Editorial Manager 1 day

    First result
    2 weeks
    Cost
    $0
    Owner
    Programming & Editorial Manager

    Smallest test

    Fix the AI Inference page and compare its search snippet a week later.

    Why it works

    Render the final value in the HTML and animate from it (or animate only for people who have not asked for reduced motion). Check every page that uses the stat component, then test what Google's URL inspection and an AI assistant quote back for the inference pages.

    Search snippets and AI answers are built from the HTML, so the best proof WEKA has should be readable there. It is a component fix that corrects every page at once.

    Risks and how to handle them

    • Animation loses its effect

      Start from the real number and animate a highlight instead.

    Evidence

    • Your websiteInference pages, as delivered to browsers and crawlers before scripts run: the AI Inference page shows "0.0x higher sustained input-token throughput", "0x more concurrent users" and "0% More effective capacity per rack" (the real values, 6.5x, 10x and 267%, are only in the script data for a count-up animation); the AI Token Efficiency page shows "0x Higher Token Throughput", "0 seconds" for Cohere and "0.0 EB effective capacity per rack".Source
    • HiringProgramming & Editorial Manager (posted October 9, 2026) runs content across "the website, YouTube, our LinkedIn presence and newsletters, the Deep Geeks podcast". Goals include "WEKA's work ranks in search, earns accurate citations from LLMs", "Strong material should show up everywhere it can work: as an article, a clip, a newsletter feature, a resource page", and "aging work gets refreshed or gracefully retired"; the role pairs with a "search and GEO specialist" and tracks "organic and LLM-referred traffic, and citations".Source
    • Your websiteHomepage: the title is "WEKA | Accelerated AI Storage for Purpose-Built Compute" and the meta description "Accelerate your AI and HPC workloads ... software-defined storage system built for accelerated compute", while the page itself says "Faster AI output. Lower cost per token. More tokens per watt." and "The Inference Economy Runs on WEKA", and press releases call WEKA "the AI data and memory infrastructure company". The robots.txt is "Optimized for SEO and AI Discovery (GEO)" and explicitly allows GPTBot, ClaudeBot, PerplexityBot and others.Source
  • 03

    The homepage says "Faster AI output. Lower cost per token. More tokens per watt." and "The Inference Economy Runs on WEKA", and WEKA now calls itself "the AI data and memory infrastructure company". The title and description, which search results and link previews show, still say "Accelerated AI Storage for Purpose-Built Compute" and "AI and HPC workloads".

    Medium impactLow effort First result in 2 weeks Senior Director

    First action

    Draft three title options and pick one.

    How to measure it

    Search click-through

    Now
    –
    Target
    +10%
    By
    60 days

    Steps

    1. 1

      Rewrite homepage

      Title and description.

      Senior Director, Digital 1 hour

    2. 2

      Audit top pages

      Top 20 by traffic.

      Search and GEO specialist 1 day

    3. 3

      Fix missing descriptions

      Calculator and landing pages first.

      Programming & Editorial Manager 1 day

    4. 4

      Align social previews

      Open Graph text and images.

      Web team 2 hours

    5. 5

      Check previews

      LinkedIn and Slack link cards.

      Senior Director, Digital 1 hour

    6. 6

      Measure

      Search Console click-through.

      Search and GEO specialist monthly

    First result
    2 weeks
    Cost
    $0
    Owner
    Senior Director, Digital

    Smallest test

    Change the homepage title and description today.

    Why it works

    Rewrite the title and meta description around AI data and memory, cost per token and NeuralMesh, with one proof point. Do the same for the top 20 pages by traffic, starting with the TCO calculator, which has no description at all.

    Search results should promise what the page delivers, and AI summaries often lean on titles and descriptions.

    Risks and how to handle them

    • Losing "AI storage" rankings

      Keep "AI storage" in the title alongside the new message.

    Evidence

    • Your websiteHomepage: the title is "WEKA | Accelerated AI Storage for Purpose-Built Compute" and the meta description "Accelerate your AI and HPC workloads ... software-defined storage system built for accelerated compute", while the page itself says "Faster AI output. Lower cost per token. More tokens per watt." and "The Inference Economy Runs on WEKA", and press releases call WEKA "the AI data and memory infrastructure company". The robots.txt is "Optimized for SEO and AI Discovery (GEO)" and explicitly allows GPTBot, ClaudeBot, PerplexityBot and others.Source
    • Your websiteThe two calculators on the site: the TCO Calculator ("See how much you can save by using WEKApod as the storage for your SuperPOD deployment") takes only a GPU type (H100/200, B200, GB200) and a GPU count and shows two totals, e.g. "$73,203,001 WEKA" against "$108,203,001 Other", and has no meta description; the Cloud Savings Calculator says "Results are based on publicly available list prices as of May 30, 2023" and asks for a form before showing savings. Neither mentions tokens, time-to-first-token or cost per token.Source
    • News coverageWEKA newsroom: NeuralMesh 6 (July 21, 2026) cites OCI benchmarks of "10x higher token throughput, 10x more concurrent users served, and 7x more tokens per GPU"; WEKA raised a $140 million Series E at a $1.6 billion valuation.Source
  • 04

    Google's mobile test of weka.io downloads 7.2 MB, 5.3 MB of it from the Wistia player, and the main content takes 18.9 s to appear. Real Chrome users on the homepage wait 2.7 s for the largest element, which fails Core Web Vitals, while the rest of the site passes at 1.9 s.

    Medium impactLow effort First result in 1 month Web team

    First action

    Swap the homepage player for a click-to-load poster.

    How to measure it

    Homepage field LCP

    Now
    –
    Target
    under 2.5 s
    By
    60 days

    Steps

    1. 1

      Poster and facade

      Click-to-load video.

      Web team 2 days

    2. 2

      Defer tags

      Load after first paint.

      Marketing Operations 2 days

    3. 3

      Trim JavaScript

      Remove unused bundles.

      Web team 1 week

    4. 4

      Retest

      PageSpeed lab test.

      Web team 1 day

    5. 5

      Watch field data

      Chrome UX Report over 28 days.

      Web team monthly

    6. 6

      Apply elsewhere

      Other pages with embedded video.

      Web team 2 weeks

    First result
    1 month
    Cost
    Developer time
    Owner
    Web team

    Smallest test

    Test the facade on a copy of the homepage.

    Why it works

    Replace the embedded player with a poster image and a play button that loads Wistia on click, defer the tag manager and session tools until after first paint, and remove unused JavaScript (about 2 s in the test).

    The homepage is where paid and partner traffic lands, and a faster first view keeps more of it. WEKA sells speed, so its own page should feel fast.

    Risks and how to handle them

    • Fewer video plays

      Track plays before and after; a poster with a clear button usually holds them.

    Evidence

    • Google PageSpeedHomepage on mobile (October 21, 2026): performance 44/100, lab Largest Contentful Paint 18.9 s, 7.2 MB downloaded in 167 requests; 6.6 MB comes from 27 third parties, 5.3 MB of it from the Wistia video player; unused JavaScript costs about 2 s. Real Chrome users on the homepage see a 2.7 s LCP, which fails Core Web Vitals; across the whole site real users load in 1.9 s and interaction and layout stability pass.Source
  • 05

    The Cloud Savings Calculator says its results use cloud list prices "as of May 30, 2023" and hides the answer behind a form. Cloud storage prices and WEKA's own cloud products have moved since, and a dated, gated number invites doubt.

    Medium impactLow effort First result in 2 weeks Marketing Operations

    First action

    Check traffic and form fills on the calculator for the last 12 months.

    How to measure it

    Calculator-sourced meetings

    Now
    –
    Target
    +30%
    By
    90 days

    Steps

    1. 1

      Check usage

      Traffic and forms, 12 months.

      Marketing Operations 1 day

    2. 2

      Decide

      Refresh or redirect.

      Senior Director, Digital 1 day

    3. 3

      Update prices

      Current list prices, dated.

      Senior Product Marketing Manager, AI Inference 1 week

    4. 4

      Ungate the result

      Form after the answer.

      Web team 2 days

    5. 5

      Redirect if retired

      To the inference calculator.

      Web team 1 hour

    6. 6

      Date every tool

      Visible "updated" line.

      Programming & Editorial Manager 1 hour

    First result
    2 weeks
    Cost
    $0
    Owner
    Marketing Operations

    Smallest test

    Move the form below the result for two weeks.

    Why it works

    Either update the prices and show the savings before asking for details, or redirect the page to the new inference calculator and the public-cloud pages. Add an "updated" date to every calculator.

    A visitor who sees a 2023 date questions every other number on the site. Showing the result first also gives sales a warmer lead.

    Risks and how to handle them

    • Fewer form fills

      Fewer but warmer; compare meeting rates, not forms.

    Evidence

    • Your websiteThe two calculators on the site: the TCO Calculator ("See how much you can save by using WEKApod as the storage for your SuperPOD deployment") takes only a GPU type (H100/200, B200, GB200) and a GPU count and shows two totals, e.g. "$73,203,001 WEKA" against "$108,203,001 Other", and has no meta description; the Cloud Savings Calculator says "Results are based on publicly available list prices as of May 30, 2023" and asks for a form before showing savings. Neither mentions tokens, time-to-first-token or cost per token.Source
    • HiringSenior Product Marketing Manager, AI Inference (posted September 28, 2026): "Define and quantify WEKA's economic value across GPU utilization, time-to-first-token, throughput, concurrency, tokens per GPU, and cost per token. Build the economic case customers need to justify investment." The role also recruits "design partners and early adopters", owns the proposition for AI clouds that should make WEKA "the storage standard for the category", and prefers to "show specific numbers and named customers than make broad claims".Source
  • 06

    The new SDR role covers inbound leads "in a timely" way as well as outbound, and the cost-analysis page promises a cost-per-token estimate. A prospect who asks for that is the warmest lead on the site.

    High impactLow effort First result in 1 month Sales Development Representat…

    First action

    Measure today's time from cost-analysis form to first reply.

    How to measure it

    Request-to-meeting rate

    Now
    –
    Target
    +25%
    By
    60 days

    Steps

    1. 1

      Measure speed

      Current time to first reply.

      Sales Development Representative 1 day

    2. 2

      Set routing

      SDR plus inference specialist alert.

      Marketing Operations 2 days

    3. 3

      Template the reply

      First estimate and three questions.

      Senior Product Marketing Manager, AI Inference 2 days

    4. 4

      Book with an SE

      Same week.

      Sales Development Representative ongoing

    5. 5

      Review calls

      Weekly.

      Sales Development Manager weekly

    6. 6

      Track

      Meetings and pipeline by speed.

      Marketing Operations weekly

    First result
    1 month
    Cost
    SDR time
    Owner
    Sales Development Representative

    Smallest test

    Reply to this week's requests within a day and compare.

    Why it works

    Route cost-analysis forms to an SDR and an inference specialist at once, with a one-day response target. Open with a first estimate from the calculator using the details on the form, and book the 30-minute call with a solutions engineer.

    Speed and a specific first answer lift meeting rates for technical buyers, who are often comparing several vendors that week.

    Risks and how to handle them

    • Estimates before the call are wrong

      Label them as ranges and confirm on the call.

    Evidence

    • HiringSales hiring across regions: a Sales Development Representative in Tampa, "the primary engine of WEKA's outbound pipeline generation", who researches "trigger events" and will "Respond to inbound leads ... in a timely" way (posted October 12, 2026); Regional Sales Managers in New York, San Francisco/Seattle, Southern California, DACH, France, the Nordics, Northern Europe, Sydney, Tokyo and for the DoD; Senior Sales Engineers in Japan, South Korea and India; a Field CTO and a Global OEM/ODM Sales Manager in Taiwan.Source
    • Your websiteAI Inference Cost Analysis page: "In a 30-minute call, we'll identify where your inference costs are highest"; "Once we get a snapshot of your infrastructure, we'll send you an estimate of what your cost per token will be on NeuralMesh"; proof points "Stability AI reduce their storage costs by 95% on a per TB basis" and CoreWeave benchmarks with "up to 4.2x" token throughput per GPU.Source
    • HiringSenior Product Marketing Manager, AI Inference (posted September 28, 2026): "Define and quantify WEKA's economic value across GPU utilization, time-to-first-token, throughput, concurrency, tokens per GPU, and cost per token. Build the economic case customers need to justify investment." The role also recruits "design partners and early adopters", owns the proposition for AI clouds that should make WEKA "the storage standard for the category", and prefers to "show specific numbers and named customers than make broad claims".Source

This quarter 12

The core plays for the next three months.

  • 07

    The competitive intelligence hire will build "comparison frameworks" and competitor comparisons. WEKA's Learn hub already ranks for comparisons between other vendors ("isilon-vs-netapp", "vast-data-vs-netapp"), but the only WEKA head-to-head is against Amazon FSx for Lustre. Buyers comparing WEKA with VAST Data, DDN or Lustre find someone else's framing.

    High impactMedium effort First result in 1 quarter Senior Product Marketing Mana…

    First action

    Pick the three comparisons sales meets most in deals.

    How to measure it

    Comparison-page meetings

    Now
    –
    Target
    30 a quarter
    By
    2 quarters

    Steps

    1. 1

      Pick comparisons

      From deal data.

      Senior Product Marketing Manager, Competitive & Market Intelligence 1 week

    2. 2

      Gather facts

      Public benchmarks and docs only.

      Senior Product Marketing Manager, Competitive & Market Intelligence 2 weeks

    3. 3

      Write

      Fair tables by workload.

      Programming & Editorial Manager 2 weeks

    4. 4

      Legal review

      Claims and trademarks.

      Legal 1 week

    5. 5

      Publish

      Learn hub with dates.

      Web team 1 day

    6. 6

      Refresh

      Quarterly from the intelligence program.

      Senior Product Marketing Manager, Competitive & Market Intelligence quarterly

    First result
    1 quarter
    Cost
    Staff time only
    Owner
    Senior Product Marketing Manager, Competitive & Market Intelligence

    Smallest test

    Publish one comparison and track meetings.

    Why it works

    Turn the strongest competitor comparisons into fair public pages: NeuralMesh vs Lustre, vs parallel file systems, vs "AI operating system" platforms, by workload (training checkpoints, KV cache, multitenant AI clouds). Use published benchmarks, date each page, and add a migration path. Keep win themes internal.

    Comparison searches are late-stage and high intent, and the pages teach AI assistants how WEKA differs.

    Risks and how to handle them

    • Out-of-date claims

      Date and source every row; review quarterly.

    Evidence

    • HiringSenior Product Marketing Manager, Competitive & Market Intelligence (posted August 28, 2026): builds "competitor comparisons, playbooks, objection handling, comparison frameworks", will "Monitor Competitive Pricing & Packaging", and "Build the Field Intelligence Loop" from "competitive deal data, field feedback, and win/loss insights".Source
    • Your websiteSitemap (807 URLs, checked October 21, 2026): 40 customer stories, 90 Learn pages, 175 articles, 188 news items, 55 landing pages, 8 podcast episodes; every page is in English, with no hreflang. Learn pages compare other vendors with each other ("isilon-vs-netapp", "vast-data-vs-netapp") while the only WEKA head-to-head is against Amazon FSx for Lustre (a video and a solution brief); generic Learn pages include "what-is-aws", "what-is-azure" and "cpu-vs-gpu". Only Dell has its own partner page, although the newsroom covers Lenovo, HPE, Supermicro and Hitachi partnerships.Source
    • Competitor researchRivals' own homepages: VAST Data, "the first AI Operating System, unifying storage, database, and compute", with English and Japanese versions (hreflang ja-JP); DDN, "Data Intelligence Platform Built for AI"; Hammerspace, "The Data Platform for AI Anywhere".Source
  • 08

    WEKA wants to be "the storage standard" for AI clouds and runs partner marketing with "cloud and neocloud providers". Its own site has customer stories from Nebius, Together AI, TensorWave, Yotta, Firmus, IREN, Applied Digital, YTL AI Cloud, Neysa and Core42, yet the Cloud Partners page lists only AWS, Google, Microsoft, Oracle, 2nd Watch and DoiT.

    High impactLow effort First result in 1 month Partner Marketing Specialist

    First action

    Ask three AI-cloud customers whether they want to be listed.

    How to measure it

    Listed AI clouds

    Now
    –
    Target
    10
    By
    2 quarters

    Steps

    1. 1

      Ask providers

      Ten customers with stories.

      Partner Marketing Specialist 2 weeks

    2. 2

      Design the listing

      Region, GPUs, features, link.

      Web team 1 week

    3. 3

      Make the badge

      Brand guidelines.

      Creative team 1 week

    4. 4

      Publish

      On the cloud partners page.

      Web team 1 day

    5. 5

      Co-market

      MDF-funded launch posts.

      Partner Marketing Specialist 1 month

    6. 6

      Measure

      Clicks and referred deals.

      Marketing Operations monthly

    First result
    1 month
    Cost
    MDF and staff time
    Owner
    Partner Marketing Specialist

    Smallest test

    List the first three providers who agree.

    Why it works

    With each provider's agreement, add an "AI clouds running NeuralMesh" section: region, GPU types, what customers get (for example persistent KV cache), and a link to their offer. Give providers a "Runs on WEKA NeuralMesh" badge and a co-marketing kit paid from MDF.

    Buyers renting GPUs can choose a WEKA-backed cloud, providers get referrals, and every listing is another place WEKA is named as the standard.

    Risks and how to handle them

    • Providers compete with each other

      List all who qualify on the same terms.

    Evidence

    • Your websiteThe Cloud Partners page lists Amazon Web Services, Google, Microsoft, Oracle, 2nd Watch and DoiT. None of the AI clouds with WEKA customer stories appear there: Nebius, Together AI, TensorWave, Yotta, Firmus, IREN, Applied Digital, YTL AI Cloud, Neysa or Core42.Source
    • HiringPartner Marketing Specialist (posted October 14, 2026), reporting to the Senior Director of Campaigns and Growth Marketing: manages "MDF end-to-end", runs a "Quarterly Regional Partner Webinar" in each region, keeps the partner portal current with "campaigns-in-a-box", and connects "every MDF dollar, webinar, and campaign back to primarily marketing sourced" pipeline. Partners span "NVIDIA, cloud and neocloud providers, Tech Alliances, OEMs, and channel partners".Source
    • HiringSenior Product Marketing Manager, AI Inference (posted September 28, 2026): "Define and quantify WEKA's economic value across GPU utilization, time-to-first-token, throughput, concurrency, tokens per GPU, and cost per token. Build the economic case customers need to justify investment." The role also recruits "design partners and early adopters", owns the proposition for AI clouds that should make WEKA "the storage standard for the category", and prefers to "show specific numbers and named customers than make broad claims".Source
    • HiringPrincipal Product Manager, Augmented Memory Grid: works with "GPU cloud partners: Nebius, CoreWeave, TogetherAI, etc." to "validate benchmarks, and prioritize features that reduce cost-per-token". A Senior Customer Success Operations Manager will be "designing CS playbooks, building internal tools".Source
  • 09

    The inference marketer would "rather show specific numbers and named customers than make broad claims". Some stories do (Cohere's deploys went from 5 minutes to 15 seconds), but others, such as Together AI's, are a quote with no measured result.

    High impactMedium effort First result in 1 quarter Senior Product Marketing Mana…

    First action

    List customer stories with no number in them.

    How to measure it

    Inference stories with numbers

    Now
    –
    Target
    15
    By
    2 quarters

    Steps

    1. 1

      Audit stories

      Which lack a number.

      Senior Product Marketing Manager, AI Inference 1 week

    2. 2

      Ask customers

      One measured result each.

      Customer advocacy 1 month

    3. 3

      Check data

      With engineering.

      Principal Product Manager, Augmented Memory Grid 2 weeks

    4. 4

      Rewrite pages

      Number first.

      Programming & Editorial Manager 1 month

    5. 5

      Tag by metric

      For the calculator and decks.

      Web team 1 week

    6. 6

      Clip video

      Short customer clips.

      Video lead 1 month

    First result
    1 quarter
    Cost
    Staff time only
    Owner
    Senior Product Marketing Manager, AI Inference

    Smallest test

    Add a number to one story this month.

    Why it works

    Revisit the inference-heavy stories with each customer: one before-and-after number (time-to-first-token, tokens per GPU, deploy time, GPU utilization), the setup, and a short video clip. Tag stories by metric so the calculator and sales decks can pull them.

    Numbers from named customers are what technical buyers forward to their CFOs.

    Risks and how to handle them

    • Customers will not share numbers

      Offer ranges or relative gains.

    Evidence

    • HiringSenior Product Marketing Manager, AI Inference (posted September 28, 2026): "Define and quantify WEKA's economic value across GPU utilization, time-to-first-token, throughput, concurrency, tokens per GPU, and cost per token. Build the economic case customers need to justify investment." The role also recruits "design partners and early adopters", owns the proposition for AI clouds that should make WEKA "the storage standard for the category", and prefers to "show specific numbers and named customers than make broad claims".Source
    • Your websiteCustomer stories: Together AI's page is a quote (Augmented Memory Grid to "reduce the time involved in prompt caching", "more than 500,000 AI developers") with no measured result; elsewhere the site does use numbers, e.g. Cohere "Inference deployments took 5 minutes now occur in 15 seconds" and Firmus "6.5x Higher Token Throughput" on the homepage.Source
    • HiringPrincipal Product Manager, Augmented Memory Grid: works with "GPU cloud partners: Nebius, CoreWeave, TogetherAI, etc." to "validate benchmarks, and prioritize features that reduce cost-per-token". A Senior Customer Success Operations Manager will be "designing CS playbooks, building internal tools".Source
  • 10

    The APJ technical marketer is hired to "Own and grow WEKA's Chinese-language presence across WeChat, Zhihu, Xiaohongshu" and to build a "benchmark series by model family". Every page on weka.io is in English.

    High impactMedium effort First result in 1 quarter Senior Technical Product Mark…

    First action

    Pick the first three model families with the Field CTO.

    How to measure it

    APJ workshop sign-ups

    Now
    –
    Target
    200
    By
    2 quarters

    Steps

    1. 1

      Pick models

      Three families.

      Senior Technical Product Marketing Manager, APJ 1 week

    2. 2

      Run benchmarks

      With regional server partners.

      Field CTO, Taiwan 1 month

    3. 3

      Write in Mandarin

      Engineer-level detail.

      Senior Technical Product Marketing Manager, APJ 2 weeks

    4. 4

      Publish

      WeChat and Zhihu.

      Senior Technical Product Marketing Manager, APJ 1 week

    5. 5

      Run workshops

      With the Field CTO.

      Field CTO, Taiwan monthly

    6. 6

      Localize

      Japanese and Korean versions.

      Senior Technical Product Marketing Manager, APJ 1 quarter

    First result
    1 quarter
    Cost
    Agency and benchmark time
    Owner
    Senior Technical Product Marketing Manager, APJ

    Smallest test

    Publish one Mandarin benchmark post.

    Why it works

    Start with the models Chinese teams run most: one post per model family with inference results (tokens per GPU, time-to-first-token, long-context behavior) on the regional server vendors' hardware, written in Mandarin for engineers, then a workshop with the Field CTO. Reuse each benchmark in Japanese and Korean through agencies.

    Engineers in the region "trust peers and evidence, not campaigns", and benchmarks on familiar models are evidence they can check.

    Risks and how to handle them

    • Regional rules on content and data

      Have local counsel review the publishing setup.

    Evidence

    • HiringSenior Technical Product Marketing Manager, APJ (China-based, posted October 19, 2026): will "Localize HQ content", create a "benchmark series by model family, inference cost and TCO analyses", "Own and grow WEKA's Chinese-language presence across WeChat, Zhihu, Xiaohongshu", manage agencies for Japanese and Korean, and run architecture workshops with the Taipei-based Field CTO. "The region's most sophisticated AI teams trust peers and evidence, not campaigns."Source
    • Your websiteSitemap (807 URLs, checked October 21, 2026): 40 customer stories, 90 Learn pages, 175 articles, 188 news items, 55 landing pages, 8 podcast episodes; every page is in English, with no hreflang. Learn pages compare other vendors with each other ("isilon-vs-netapp", "vast-data-vs-netapp") while the only WEKA head-to-head is against Amazon FSx for Lustre (a video and a solution brief); generic Learn pages include "what-is-aws", "what-is-azure" and "cpu-vs-gpu". Only Dell has its own partner page, although the newsroom covers Lenovo, HPE, Supermicro and Hitachi partnerships.Source
    • HiringSales hiring across regions: a Sales Development Representative in Tampa, "the primary engine of WEKA's outbound pipeline generation", who researches "trigger events" and will "Respond to inbound leads ... in a timely" way (posted October 12, 2026); Regional Sales Managers in New York, San Francisco/Seattle, Southern California, DACH, France, the Nordics, Northern Europe, Sydney, Tokyo and for the DoD; Senior Sales Engineers in Japan, South Korea and India; a Field CTO and a Global OEM/ODM Sales Manager in Taiwan.Source
  • 11

    WEKA is hiring a Regional Sales Manager in Tokyo and senior sales engineers in Japan and South Korea, but weka.io has no translated pages. VAST Data runs a Japanese version of its site.

    Medium impactMedium effort First result in 1 quarter Senior Director

    First action

    Ask the Tokyo and Seoul teams which pages they send most.

    How to measure it

    Local-language meetings

    Now
    –
    Target
    20 a quarter
    By
    2 quarters

    Steps

    1. 1

      Pick pages

      Ten per language.

      Senior Director, Digital 1 week

    2. 2

      Translate

      Agency plus engineer review.

      Senior Technical Product Marketing Manager, APJ 1 month

    3. 3

      Add hreflang

      Language versions linked.

      Web team 1 week

    4. 4

      Route forms

      To regional sellers.

      Marketing Operations 1 week

    5. 5

      Promote

      Local events and LinkedIn.

      Field marketing, APJ ongoing

    6. 6

      Measure

      Traffic and meetings.

      Marketing Operations monthly

    First result
    1 quarter
    Cost
    Translation budget
    Owner
    Senior Director, Digital

    Smallest test

    Translate the homepage and one customer story into Japanese.

    Why it works

    Translate the ten pages local sellers send most (homepage, NeuralMesh, Augmented Memory Grid, AI Inference, AI cloud providers, two customer stories, contact) with hreflang, review them with local engineers, and route local-language forms to the regional team.

    Technical buyers share vendor pages internally, and a page in their language travels further.

    Risks and how to handle them

    • Translations go stale

      Translate only pages with an owner and a review date.

    Evidence

    • HiringSales hiring across regions: a Sales Development Representative in Tampa, "the primary engine of WEKA's outbound pipeline generation", who researches "trigger events" and will "Respond to inbound leads ... in a timely" way (posted October 12, 2026); Regional Sales Managers in New York, San Francisco/Seattle, Southern California, DACH, France, the Nordics, Northern Europe, Sydney, Tokyo and for the DoD; Senior Sales Engineers in Japan, South Korea and India; a Field CTO and a Global OEM/ODM Sales Manager in Taiwan.Source
    • Your websiteSitemap (807 URLs, checked October 21, 2026): 40 customer stories, 90 Learn pages, 175 articles, 188 news items, 55 landing pages, 8 podcast episodes; every page is in English, with no hreflang. Learn pages compare other vendors with each other ("isilon-vs-netapp", "vast-data-vs-netapp") while the only WEKA head-to-head is against Amazon FSx for Lustre (a video and a solution brief); generic Learn pages include "what-is-aws", "what-is-azure" and "cpu-vs-gpu". Only Dell has its own partner page, although the newsroom covers Lenovo, HPE, Supermicro and Hitachi partnerships.Source
    • Competitor researchRivals' own homepages: VAST Data, "the first AI Operating System, unifying storage, database, and compute", with English and Japanese versions (hreflang ja-JP); DDN, "Data Intelligence Platform Built for AI"; Hammerspace, "The Data Platform for AI Anywhere".Source
    • HiringSenior Technical Product Marketing Manager, APJ (China-based, posted October 19, 2026): will "Localize HQ content", create a "benchmark series by model family, inference cost and TCO analyses", "Own and grow WEKA's Chinese-language presence across WeChat, Zhihu, Xiaohongshu", manage agencies for Japanese and Korean, and run architecture workshops with the Taipei-based Field CTO. "The region's most sophisticated AI teams trust peers and evidence, not campaigns."Source
  • 12

    The OEM channel manager covers Dell, HPE, Lenovo, Supermicro, Cisco and Hitachi and must equip their sellers. Only Dell has a partner page on weka.io; Lenovo, HPE, Supermicro and Hitachi appear only in old news items.

    High impactMedium effort First result in 2 quarters Channel Sales Manager OEM

    First action

    Draft the Lenovo page from existing announcements.

    How to measure it

    OEM-sourced opportunities

    Now
    –
    Target
    +30%
    By
    3 quarters

    Steps

    1. 1

      Collect content

      Configurations and joint wins.

      Channel Sales Manager OEM 2 weeks

    2. 2

      Partner approval

      Brand and claims.

      Channel Sales Manager OEM 2 weeks

    3. 3

      Build pages

      One template.

      Web team 2 weeks

    4. 4

      Add to enablement

      Partner seller training.

      Channel Sales Manager OEM 1 month

    5. 5

      Add to portal

      Campaigns-in-a-box.

      Partner Marketing Specialist 1 week

    6. 6

      Measure

      Opportunities by partner.

      Marketing Operations monthly

    First result
    2 quarters
    Cost
    Staff time only
    Owner
    Channel Sales Manager OEM

    Smallest test

    Publish one OEM page.

    Why it works

    Build a page per OEM: certified configurations, reference architectures, joint customers, how to order through that partner, and a contact. Make each page the first link in partner seller enablement.

    Partner sellers search for "<their brand> + WEKA" before a call, and a clear page makes them more likely to bring WEKA in.

    Risks and how to handle them

    • Partner approvals slow

      Start with partners that already co-market.

    Evidence

    • HiringChannel Sales Manager OEM (posted September 9, 2026) owns partners "such as Dell, AMD, HPE, Cisco, Hitachi, and Lenovo" plus Supermicro, will "Develop joint go-to-market strategies" and "Design and deliver comprehensive enablement programs that equip sales and pre-sales teams across all partners". Two Partner ISV Alliances Managers are also open.Source
    • Your websiteSitemap (807 URLs, checked October 21, 2026): 40 customer stories, 90 Learn pages, 175 articles, 188 news items, 55 landing pages, 8 podcast episodes; every page is in English, with no hreflang. Learn pages compare other vendors with each other ("isilon-vs-netapp", "vast-data-vs-netapp") while the only WEKA head-to-head is against Amazon FSx for Lustre (a video and a solution brief); generic Learn pages include "what-is-aws", "what-is-azure" and "cpu-vs-gpu". Only Dell has its own partner page, although the newsroom covers Lenovo, HPE, Supermicro and Hitachi partnerships.Source
  • 13

    The partner marketing role runs a quarterly webinar per region and keeps the portal stocked with campaigns-in-a-box, and every MDF dollar must tie back to pipeline.

    Medium impactLow effort First result in 1 quarter Partner Marketing Specialist

    First action

    Package the next regional webinar this way.

    How to measure it

    Leads from partner reruns

    Now
    –
    Target
    300 a quarter
    By
    2 quarters

    Steps

    1. 1

      Plan the kit

      Recording, clips, email, page.

      Partner Marketing Specialist 1 week

    2. 2

      Produce

      After each webinar.

      Partner Marketing Specialist 2 weeks

    3. 3

      Publish to portal

      With tracked links.

      Partner Marketing Specialist 1 day

    4. 4

      Brief partners

      How to rerun it.

      Field marketing managers 1 week

    5. 5

      Fund reruns

      MDF claims tied to links.

      Partner Marketing Specialist ongoing

    6. 6

      Report

      Pipeline by partner.

      Marketing Operations quarterly

    First result
    1 quarter
    Cost
    MDF
    Owner
    Partner Marketing Specialist

    Smallest test

    Package one webinar for five partners.

    Why it works

    After each webinar, package the recording, three clips, a co-brandable email, a landing page template and tracked links. Partners rerun it to their customers with MDF support; leads route back with partner attribution.

    One production serves dozens of partners, and tracked links prove which MDF spend works.

    Risks and how to handle them

    • Partners do not use kits

      Start with five partners who ask for content.

    Evidence

    • HiringPartner Marketing Specialist (posted October 14, 2026), reporting to the Senior Director of Campaigns and Growth Marketing: manages "MDF end-to-end", runs a "Quarterly Regional Partner Webinar" in each region, keeps the partner portal current with "campaigns-in-a-box", and connects "every MDF dollar, webinar, and campaign back to primarily marketing sourced" pipeline. Partners span "NVIDIA, cloud and neocloud providers, Tech Alliances, OEMs, and channel partners".Source
  • 14

    The intelligence role must "Build the Field Intelligence Loop" from deal data and win/loss insights. The objections that lose deals are often the questions buyers also type into search.

    Medium impactMedium effort First result in 1 quarter Senior Product Marketing Mana…

    First action

    Interview five recent wins and five losses.

    How to measure it

    Competitive win rate

    Now
    –
    Target
    +5 points
    By
    3 quarters

    Steps

    1. 1

      Set up interviews

      Third party or PMM.

      Senior Product Marketing Manager, Competitive & Market Intelligence 2 weeks

    2. 2

      Tag reasons

      Shared taxonomy.

      Senior Product Marketing Manager, Competitive & Market Intelligence ongoing

    3. 3

      Pick top objection

      Monthly.

      Senior Product Marketing Manager, Competitive & Market Intelligence monthly

    4. 4

      Publish an answer

      Learn or FAQ page.

      Programming & Editorial Manager monthly

    5. 5

      Update competitor comparisons

      Same answer internally.

      Senior Product Marketing Manager, Competitive & Market Intelligence monthly

    6. 6

      Measure

      Win rate by theme.

      Marketing Operations quarterly

    First result
    1 quarter
    Cost
    Interview time
    Owner
    Senior Product Marketing Manager, Competitive & Market Intelligence

    Smallest test

    Run ten interviews.

    Why it works

    Interview buyers after each closed deal (won and lost), tag the reasons, and each month publish or update one public answer page (FAQ, comparison row or Learn page) for the most common objection.

    The site answers objections before the call, and sellers can send a public page instead of a slide.

    Risks and how to handle them

    • Buyers decline interviews

      Offer a short survey and a donation.

    Evidence

    • HiringSenior Product Marketing Manager, Competitive & Market Intelligence (posted August 28, 2026): builds "competitor comparisons, playbooks, objection handling, comparison frameworks", will "Monitor Competitive Pricing & Packaging", and "Build the Field Intelligence Loop" from "competitive deal data, field feedback, and win/loss insights".Source
    • Your websiteSitemap (807 URLs, checked October 21, 2026): 40 customer stories, 90 Learn pages, 175 articles, 188 news items, 55 landing pages, 8 podcast episodes; every page is in English, with no hreflang. Learn pages compare other vendors with each other ("isilon-vs-netapp", "vast-data-vs-netapp") while the only WEKA head-to-head is against Amazon FSx for Lustre (a video and a solution brief); generic Learn pages include "what-is-aws", "what-is-azure" and "cpu-vs-gpu". Only Dell has its own partner page, although the newsroom covers Lenovo, HPE, Supermicro and Hitachi partnerships.Source
  • 15

    The editorial manager is asked to make sure "aging work gets refreshed or gracefully retired". The 90-page Learn hub mixes strong inference topics (KV cache, prefill and decode) with generic ones such as "what-is-aws", "what-is-azure" and "cpu-vs-gpu" that attract readers who will never buy AI infrastructure.

    Medium impactLow effort First result in 1 quarter Programming & Editorial Manag…

    First action

    Export Learn page traffic and leads for 12 months.

    How to measure it

    Learn-sourced leads

    Now
    –
    Target
    +25%
    By
    2 quarters

    Steps

    1. 1

      Rank pages

      Traffic and leads.

      Search and GEO specialist 1 week

    2. 2

      Decide

      Keep, refresh, merge or redirect.

      Programming & Editorial Manager 1 week

    3. 3

      Refresh

      Top 20 buyer pages.

      Contract writers 1 month

    4. 4

      Redirect

      Generic pages.

      Web team 1 day

    5. 5

      Build the cluster

      Inference economics.

      Programming & Editorial Manager 1 month

    6. 6

      Repeat

      Twice a year.

      Programming & Editorial Manager 6 months

    First result
    1 quarter
    Cost
    Writing time
    Owner
    Programming & Editorial Manager

    Smallest test

    Redirect five generic pages and watch leads.

    Why it works

    Rank Learn pages by traffic and by leads; merge or redirect generic pages, refresh the ones that bring buyers, and add a cluster on inference economics that links to the calculator and the inference pages.

    A focused hub sends stronger topical signals to search and AI systems, and more of its readers become leads.

    Risks and how to handle them

    • Traffic drops

      Expect fewer visits and more leads; report both.

    Evidence

    • HiringProgramming & Editorial Manager (posted October 9, 2026) runs content across "the website, YouTube, our LinkedIn presence and newsletters, the Deep Geeks podcast". Goals include "WEKA's work ranks in search, earns accurate citations from LLMs", "Strong material should show up everywhere it can work: as an article, a clip, a newsletter feature, a resource page", and "aging work gets refreshed or gracefully retired"; the role pairs with a "search and GEO specialist" and tracks "organic and LLM-referred traffic, and citations".Source
    • Your websiteSitemap (807 URLs, checked October 21, 2026): 40 customer stories, 90 Learn pages, 175 articles, 188 news items, 55 landing pages, 8 podcast episodes; every page is in English, with no hreflang. Learn pages compare other vendors with each other ("isilon-vs-netapp", "vast-data-vs-netapp") while the only WEKA head-to-head is against Amazon FSx for Lustre (a video and a solution brief); generic Learn pages include "what-is-aws", "what-is-azure" and "cpu-vs-gpu". Only Dell has its own partner page, although the newsroom covers Lenovo, HPE, Supermicro and Hitachi partnerships.Source
    • Your websiteInference pages, as delivered to browsers and crawlers before scripts run: the AI Inference page shows "0.0x higher sustained input-token throughput", "0x more concurrent users" and "0% More effective capacity per rack" (the real values, 6.5x, 10x and 267%, are only in the script data for a count-up animation); the AI Token Efficiency page shows "0x Higher Token Throughput", "0 seconds" for Cohere and "0.0 EB effective capacity per rack".Source
  • 16

    The editorial role owns newsletters among WEKA's owned channels, but neither the homepage nor The Cache offers a sign-up. Readers of 175 articles and 8 podcast episodes have no way to hear about the next one except social.

    Medium impactLow effort First result in 1 month Programming & Editorial Manag…

    First action

    Add the sign-up box to The Cache.

    How to measure it

    Subscribers

    Now
    –
    Target
    5,000
    By
    2 quarters

    Steps

    1. 1

      Build the form

      Two fields.

      Web team 2 days

    2. 2

      Place it

      The Cache and articles.

      Web team 1 day

    3. 3

      Plan the digest

      Fortnightly format.

      Programming & Editorial Manager 1 week

    4. 4

      Send

      Every two weeks.

      Programming & Editorial Manager fortnightly

    5. 5

      Score

      Engagement into lead scoring.

      Marketing Operations 2 weeks

    6. 6

      Measure

      Growth and clicks.

      Programming & Editorial Manager monthly

    First result
    1 month
    Cost
    $0
    Owner
    Programming & Editorial Manager

    Smallest test

    Run the box on The Cache for a month.

    Why it works

    Add a short sign-up (email, role) to The Cache and the end of every article, send a fortnightly digest of the best piece, a clip and one benchmark, and connect subscribers to the marketing automation platform.

    An owned list compounds every article and keeps WEKA in front of buyers between buying cycles.

    Risks and how to handle them

    • Low sign-up rate

      Offer the benchmark report as the reason to subscribe.

    Evidence

    • Your websiteOwned channels: YouTube has about 3,340 subscribers and 212 videos; the Deep Geeks podcast has 8 episodes in the sitemap; no newsletter sign-up appears on the homepage or on The Cache, WEKA's content hub.Source
    • HiringProgramming & Editorial Manager (posted October 9, 2026) runs content across "the website, YouTube, our LinkedIn presence and newsletters, the Deep Geeks podcast". Goals include "WEKA's work ranks in search, earns accurate citations from LLMs", "Strong material should show up everywhere it can work: as an article, a clip, a newsletter feature, a resource page", and "aging work gets refreshed or gracefully retired"; the role pairs with a "search and GEO specialist" and tracks "organic and LLM-referred traffic, and citations".Source
  • 17

    The editorial brief says strong material should appear "as an article, a clip, a newsletter feature, a resource page". The YouTube channel has about 3,340 subscribers across 212 videos, and Deep Geeks has 8 episodes.

    Medium impactMedium effort First result in 1 quarter Programming & Editorial Manag…

    First action

    Pick next week's story and plan the five pieces.

    How to measure it

    Average views per story

    Now
    –
    Target
    2x
    By
    2 quarters

    Steps

    1. 1

      Pick story

      Weekly.

      Programming & Editorial Manager weekly

    2. 2

      Interview

      Engineer or customer.

      Programming & Editorial Manager weekly

    3. 3

      Produce formats

      Article, clips, post, newsletter.

      Contract writers weekly

    4. 4

      Publish

      Across channels.

      Organic social lead weekly

    5. 5

      Update Learn

      Link the story in.

      Programming & Editorial Manager weekly

    6. 6

      Review

      Formats that work.

      Programming & Editorial Manager monthly

    First result
    1 quarter
    Cost
    Production time
    Owner
    Programming & Editorial Manager

    Smallest test

    Run the routine for four weeks.

    Why it works

    Each week pick one engineer or customer story and ship it as an article, two short clips, a LinkedIn post from the engineer, a newsletter feature and an updated Learn page. Track which formats bring LLM-referred and organic visits.

    Repurposing multiplies reach for the same interview time, which is the scarce input.

    Risks and how to handle them

    • Engineer time

      Keep interviews to 30 minutes.

    Evidence

    • HiringProgramming & Editorial Manager (posted October 9, 2026) runs content across "the website, YouTube, our LinkedIn presence and newsletters, the Deep Geeks podcast". Goals include "WEKA's work ranks in search, earns accurate citations from LLMs", "Strong material should show up everywhere it can work: as an article, a clip, a newsletter feature, a resource page", and "aging work gets refreshed or gracefully retired"; the role pairs with a "search and GEO specialist" and tracks "organic and LLM-referred traffic, and citations".Source
    • Your websiteOwned channels: YouTube has about 3,340 subscribers and 212 videos; the Deep Geeks podcast has 8 episodes in the sitemap; no newsletter sign-up appears on the homepage or on The Cache, WEKA's content hub.Source
  • 18

    The inference marketer will "find, recruit, and support design partners and early adopters". There is no page that tells an AI team what being a design partner involves or how to apply.

    High impactMedium effort First result in 2 quarters Senior Product Marketing Mana…

    First action

    Write the qualification criteria.

    How to measure it

    Signed design partners

    Now
    –
    Target
    10
    By
    2 quarters

    Steps

    1. 1

      Define the offer

      Criteria, benefits, asks.

      Senior Product Marketing Manager, AI Inference 1 week

    2. 2

      Agree capacity

      Engineering support.

      Principal Product Manager, Augmented Memory Grid 1 week

    3. 3

      Publish page

      With an application form.

      Web team 1 week

    4. 4

      Promote

      To cost-analysis leads.

      Sales Development Representative ongoing

    5. 5

      Run pilots

      Measured results.

      Principal Product Manager, Augmented Memory Grid 1 quarter

    6. 6

      Publish results

      Stories with numbers.

      Senior Product Marketing Manager, AI Inference 1 quarter

    First result
    2 quarters
    Cost
    Engineering time
    Owner
    Senior Product Marketing Manager, AI Inference

    Smallest test

    Invite five cost-analysis leads.

    Why it works

    Publish a design-partner page: who qualifies (production inference, long context or agents), what they get (engineering access, a benchmark on their models, early features), what WEKA asks (a measured result and a story). Promote it to cost-analysis leads and Learn readers.

    A named program turns curious teams into committed pilots and produces the proof the inference business needs.

    Risks and how to handle them

    • Too many applicants

      Strict criteria and a waitlist.

    Evidence

    • HiringSenior Product Marketing Manager, AI Inference (posted September 28, 2026): "Define and quantify WEKA's economic value across GPU utilization, time-to-first-token, throughput, concurrency, tokens per GPU, and cost per token. Build the economic case customers need to justify investment." The role also recruits "design partners and early adopters", owns the proposition for AI clouds that should make WEKA "the storage standard for the category", and prefers to "show specific numbers and named customers than make broad claims".Source
    • HiringPrincipal Product Manager, Augmented Memory Grid: works with "GPU cloud partners: Nebius, CoreWeave, TogetherAI, etc." to "validate benchmarks, and prioritize features that reduce cost-per-token". A Senior Customer Success Operations Manager will be "designing CS playbooks, building internal tools".Source
    • Your websiteAI Inference Cost Analysis page: "In a 30-minute call, we'll identify where your inference costs are highest"; "Once we get a snapshot of your infrastructure, we'll send you an estimate of what your cost per token will be on NeuralMesh"; proof points "Stability AI reduce their storage costs by 95% on a per TB basis" and CoreWeave benchmarks with "up to 4.2x" token throughput per GPU.Source

Big bets 3

Larger plays that can change the trajectory.

  • 19

    WEKA's whole message is now inference economics, it has OCI and CoreWeave benchmarks, and the APJ role plans a "benchmark series by model family". No one publishes a neutral-looking, repeatable view of what memory and storage choices do to cost per token across popular models.

    High impactHigh effort First result in 3 quarters Senior Product Marketing Mana…

    First action

    Pick five models and two GPU types for the first edition.

    How to measure it

    Benchmark downloads

    Now
    –
    Target
    10,000
    By
    1 year

    Steps

    1. 1

      Define method

      Models, setups, metrics.

      Principal Product Manager, Augmented Memory Grid 1 month

    2. 2

      Run tests

      Lab and partner clouds.

      Principal Product Manager, Augmented Memory Grid 2 months

    3. 3

      Review

      Independent check of method.

      Senior Product Marketing Manager, AI Inference 1 month

    4. 4

      Publish

      Report, data and harness.

      Programming & Editorial Manager 1 month

    5. 5

      Localize

      Mandarin and Japanese.

      Senior Technical Product Marketing Manager, APJ 1 month

    6. 6

      Repeat

      Quarterly.

      Senior Product Marketing Manager, AI Inference quarterly

    First result
    3 quarters
    Cost
    Lab time and GPU hours
    Owner
    Senior Product Marketing Manager, AI Inference

    Smallest test

    Publish one chart for one model.

    Why it works

    Publish a quarterly benchmark: popular open models, context lengths and GPU types, with tokens per GPU and time-to-first-token for common KV-cache setups, the harness published so others can reproduce it, and results in English, Mandarin and Japanese.

    Original, repeatable data earns press, links and AI citations, and makes WEKA the reference for the metric it wants buyers to use.

    Risks and how to handle them

    • Seen as vendor marketing

      Publish the harness and invite others to rerun it.

    Evidence

    • News coverageWEKA newsroom: NeuralMesh 6 (July 21, 2026) cites OCI benchmarks of "10x higher token throughput, 10x more concurrent users served, and 7x more tokens per GPU"; WEKA raised a $140 million Series E at a $1.6 billion valuation.Source
    • HiringSenior Product Marketing Manager, AI Inference (posted September 28, 2026): "Define and quantify WEKA's economic value across GPU utilization, time-to-first-token, throughput, concurrency, tokens per GPU, and cost per token. Build the economic case customers need to justify investment." The role also recruits "design partners and early adopters", owns the proposition for AI clouds that should make WEKA "the storage standard for the category", and prefers to "show specific numbers and named customers than make broad claims".Source
    • HiringSenior Technical Product Marketing Manager, APJ (China-based, posted October 19, 2026): will "Localize HQ content", create a "benchmark series by model family, inference cost and TCO analyses", "Own and grow WEKA's Chinese-language presence across WeChat, Zhihu, Xiaohongshu", manage agencies for Japanese and Korean, and run architecture workshops with the Taipei-based Field CTO. "The region's most sophisticated AI teams trust peers and evidence, not campaigns."Source
    • HiringProgramming & Editorial Manager (posted October 9, 2026) runs content across "the website, YouTube, our LinkedIn presence and newsletters, the Deep Geeks podcast". Goals include "WEKA's work ranks in search, earns accurate citations from LLMs", "Strong material should show up everywhere it can work: as an article, a clip, a newsletter feature, a resource page", and "aging work gets refreshed or gracefully retired"; the role pairs with a "search and GEO specialist" and tracks "organic and LLM-referred traffic, and citations".Source
  • 20

    The CS operations hire will design playbooks and internal tools, and WEKA's value is measured in GPU productivity. Customers who see their own gains are the easiest to expand and to turn into stories.

    High impactHigh effort First result in 3 quarters Senior Customer Success Opera…

    First action

    Draft the report for three customers by hand.

    How to measure it

    Customers receiving reports

    Now
    –
    Target
    80%
    By
    3 quarters

    Steps

    1. 1

      Pick metrics

      Available from observability.

      Senior Customer Success Operations Manager 1 month

    2. 2

      Build the report

      Template and data pull.

      Senior Customer Success Operations Manager 1 quarter

    3. 3

      Pilot

      Ten customers.

      Customer Success 1 quarter

    4. 4

      Train CSMs

      Business reviews.

      Senior Customer Success Operations Manager 1 month

    5. 5

      Roll out

      All customers.

      Senior Customer Success Operations Manager 1 quarter

    6. 6

      Feed stories

      Best results to marketing.

      Senior Product Marketing Manager, AI Inference ongoing

    First result
    3 quarters
    Cost
    Tooling time
    Owner
    Senior Customer Success Operations Manager

    Smallest test

    Make one report by hand.

    Why it works

    Build a quarterly report per customer from NeuralMesh observability: GPU utilization, throughput, cache hit rates and estimated tokens per GPU against their baseline, with recommendations. Use it in business reviews, renewals and story requests.

    Showing measured value each quarter protects renewals in a market with strong rivals and surfaces expansion moments.

    Risks and how to handle them

    • Telemetry limits

      Use only what customers already share.

    Evidence

    • HiringPrincipal Product Manager, Augmented Memory Grid: works with "GPU cloud partners: Nebius, CoreWeave, TogetherAI, etc." to "validate benchmarks, and prioritize features that reduce cost-per-token". A Senior Customer Success Operations Manager will be "designing CS playbooks, building internal tools".Source
    • HiringCompany description in the job posts: "WEKA is a pre-IPO, growth-stage company"; "Our technology is trusted by 30% of the Fortune 50, as well as leading hyperscalers, neoclouds, AI builders, and research organizations"; NeuralMesh "combines high-performance data storage and extended GPU memory in a single software platform".Source
    • Your websiteCustomer stories: Together AI's page is a quote (Augmented Memory Grid to "reduce the time involved in prompt caching", "more than 500,000 AI developers") with no measured result; elsewhere the site does use numbers, e.g. Cohere "Inference deployments took 5 minutes now occur in 15 seconds" and Firmus "6.5x Higher Token Throughput" on the homepage.Source
  • 21

    WEKA is running a 90-minute virtual lab where attendees benchmark a six-node cluster and "crash a node". Hands-on labs are rare in storage marketing and convert engineers better than webinars.

    High impactHigh effort First result in 3 quarters Senior Director

    First action

    Record the November lab and survey attendees.

    How to measure it

    Lab-to-opportunity rate

    Now
    –
    Target
    15%
    By
    3 quarters

    Steps

    1. 1

      Debrief November

      Attendance and survey.

      Senior Director, Campaigns and Growth Marketing 1 week

    2. 2

      Plan themes

      Six months.

      Senior Product Marketing Manager, AI Inference 2 weeks

    3. 3

      Automate the lab

      Repeatable environment.

      Technical marketing 1 month

    4. 4

      Add regions

      APJ and EMEA slots.

      Senior Technical Product Marketing Manager, APJ 1 quarter

    5. 5

      Partner editions

      Co-hosted.

      Partner Marketing Specialist 1 quarter

    6. 6

      Measure

      Opportunities by attendee.

      Marketing Operations monthly

    First result
    3 quarters
    Cost
    Lab infrastructure
    Owner
    Senior Director, Campaigns and Growth Marketing

    Smallest test

    Run a second lab in December.

    Why it works

    Make it a monthly series with themes (KV cache for inference, checkpointing, multitenancy), regional time slots, a lab certificate, and a follow-up path to the cost analysis. Offer the lab to partners as a co-hosted event.

    Engineers who have broken and recovered a cluster themselves trust the product, and their activity tells sales who is serious.

    Risks and how to handle them

    • Lab costs

      Limit seats and reuse environments.

    Evidence

    • Your websiteEvents: "WEKA Lab: Break a Live Cluster", a 90-minute virtual lab on November 3 with "an isolated six-node cluster" where attendees "Benchmark storage, test three protocols, and crash a node"; also GTC Berlin 2026, the WEKA Executive Summit 2026 and SC26.Source
    • HiringPartner Marketing Specialist (posted October 14, 2026), reporting to the Senior Director of Campaigns and Growth Marketing: manages "MDF end-to-end", runs a "Quarterly Regional Partner Webinar" in each region, keeps the partner portal current with "campaigns-in-a-box", and connects "every MDF dollar, webinar, and campaign back to primarily marketing sourced" pipeline. Partners span "NVIDIA, cloud and neocloud providers, Tech Alliances, OEMs, and channel partners".Source
    • HiringSenior Technical Product Marketing Manager, APJ (China-based, posted October 19, 2026): will "Localize HQ content", create a "benchmark series by model family, inference cost and TCO analyses", "Own and grow WEKA's Chinese-language presence across WeChat, Zhihu, Xiaohongshu", manage agencies for Japanese and Korean, and run architecture workshops with the Taipei-based Field CTO. "The region's most sophisticated AI teams trust peers and evidence, not campaigns."Source

Wildcards 2

Unexpected experiments, each with a cheap way to test it.

  • 22

    The homepage loads 27 third parties and 6.6 MB of third-party code, across analytics, advertising, intent and session tools. Some tools overlap, and not every tool has a clear owner.

    Medium impactLow effort First result in 1 month Marketing Operations

    First action

    Export the tag manager container and list owners.

    How to measure it

    Third-party count

    Now
    –
    Target
    under 15
    By
    1 quarter

    Steps

    1. 1

      Inventory

      All tags and hosts.

      Marketing Operations 1 week

    2. 2

      Assign owners

      Purpose and contract.

      Marketing Operations 1 week

    3. 3

      Remove

      Unused and duplicate tools.

      Marketing Operations 1 week

    4. 4

      Confirm vendors

      Each tool under contract.

      Marketing Operations 1 week

    5. 5

      Check consent

      Tags wait for consent.

      Marketing Operations 1 week

    6. 6

      Review

      Quarterly.

      Marketing Operations quarterly

    First result
    1 month
    Cost
    $0
    Owner
    Marketing Operations

    Smallest test

    Remove one unused tool this week.

    Why it works

    List every tag with its owner and purpose, remove duplicates and unused tools, confirm each host belongs to a vendor under contract, and make sure tags respect the consent banner.

    Fewer tags mean a faster page, cleaner data and less risk.

    Risks and how to handle them

    • Breaking reports

      Remove tags one at a time and check dashboards.

    Evidence

    • Google PageSpeedHomepage on mobile (October 21, 2026): performance 44/100, lab Largest Contentful Paint 18.9 s, 7.2 MB downloaded in 167 requests; 6.6 MB comes from 27 third parties, 5.3 MB of it from the Wistia video player; unused JavaScript costs about 2 s. Real Chrome users on the homepage see a 2.7 s LCP, which fails Core Web Vitals; across the whole site real users load in 1.9 s and interaction and layout stability pass.Source
  • 23

    The inference role must make "an economic argument for a CFO in the afternoon". WEKA's pages are written for infrastructure engineers, but AI budgets are approved by finance.

    Medium impactLow effort First result in 1 quarter Senior Director

    First action

    Draft the page with one finance reviewer.

    How to measure it

    CFO page visits from deals

    Now
    –
    Target
    200 a quarter
    By
    2 quarters

    Steps

    1. 1

      Draft

      Plain language.

      Senior Product Marketing Manager, AI Inference 1 week

    2. 2

      Finance review

      One CFO advisor.

      Senior Product Marketing Manager, AI Inference 1 week

    3. 3

      Design

      Page and PDF.

      Web team 1 week

    4. 4

      Publish

      Ungated.

      Web team 1 day

    5. 5

      Enable sellers

      Forwarding note.

      Sales enablement 1 week

    6. 6

      Measure

      Visits from deals.

      Marketing Operations monthly

    First result
    1 quarter
    Cost
    Staff time only
    Owner
    Senior Director, Digital

    Smallest test

    Send the draft to three open deals.

    Why it works

    Publish a short page and a two-page PDF that explain cost per token, why memory drives it, and what the OCI and CoreWeave results mean in money, with no acronyms. Give sellers a version to forward.

    Deals stall at budget approval, and a page written for the approver moves them.

    Risks and how to handle them

    • Too simple for engineers

      Link to the technical pages.

    Evidence

    • HiringSenior Product Marketing Manager, AI Inference (posted September 28, 2026): "Define and quantify WEKA's economic value across GPU utilization, time-to-first-token, throughput, concurrency, tokens per GPU, and cost per token. Build the economic case customers need to justify investment." The role also recruits "design partners and early adopters", owns the proposition for AI clouds that should make WEKA "the storage standard for the category", and prefers to "show specific numbers and named customers than make broad claims".Source
    • News coverageWEKA newsroom: NeuralMesh 6 (July 21, 2026) cites OCI benchmarks of "10x higher token throughput, 10x more concurrent users served, and 7x more tokens per GPU"; WEKA raised a $140 million Series E at a $1.6 billion valuation.Source
    • Your websiteAI Inference Cost Analysis page: "In a 30-minute call, we'll identify where your inference costs are highest"; "Once we get a snapshot of your infrastructure, we'll send you an estimate of what your cost per token will be on NeuralMesh"; proof points "Stability AI reduce their storage costs by 95% on a per TB basis" and CoreWeave benchmarks with "up to 4.2x" token throughput per GPU.Source

03

How Programmable can help

Where an idea in this playbook is something Programmable builds or runs, it's listed here with what we'd do. Everything else is yours to run.

ProgrammableTalk

Fill the events in this plan

These ideas are events. Talk gets registrants for them and tells you who's coming and what their company does before the day.

Plan an event (opens programmable.talk in a new tab)
ProgrammableDigital

Show it, don't just describe it

These ideas need something to show: a demo, a video, an animation, graphics or a deck. Digital turns screenshots or the site into all of them, in the company's own brand.

Make a demo (opens programmable.digital in a new tab)
More example playbooks and reviews

More growth playbooks

Other companies we have researched, each read from what they are hiring for.

  • SARELLY growth strategyA Mexican makeup and fashion-accessories brand co-founded by creator Anna Sarelly: waterproof mascaras, lips, palettes and kits, plus bags…22 growth ideas
  • BioCentury growth strategyAn intelligence platform for biopharma: independent analysis and news, the BCIQ drug discovery and development database, executive and…22 growth ideas
  • Lavendo growth strategyA boutique recruiting firm that finds sales, go-to-market and technical talent for startups, with a candidate job board and career…21 growth ideas
  • Robin Radar Systems growth strategyA Dutch maker of 3D radar systems that detect and classify small flying objects: IRIS for drone detection (counter-UAS) and MAX for bird…22 growth ideas
  • Subsplash growth strategyAn all-in-one engagement platform for churches: custom apps, websites, online giving, media and live streaming, church management…22 growth ideas
  • Siena AI growth strategyAI agents for consumer brands that handle customer service, shopping, social comments and DMs, reviews and voice, built on a shared…21 growth ideas
  • Takealot Group growth strategySouth Africa's largest online retailer and marketplace (takealot.com), with Mr D for food, grocery and retail delivery, Takealot…22 growth ideas
  • Lightspeed Systems growth strategyCloud software for K-12 schools: web filtering, student safety monitoring, classroom management, digital hall passes, device management…21 growth ideas

See every playbook and website review

Independent research. Programmable wrote this playbook from public sources, as they stood on October 22, 2026, and cites them throughout. WEKA did not commission, review or endorse it, and Programmable is not affiliated with WEKA. Company and product names are trademarks of their owners, used only to identify them. The ideas are suggestions and opinion, not professional, legal or financial advice, and not statements about how WEKA runs its business.

Something wrong, or do you represent WEKA? Email [email protected] and we will correct it or take it down.

30/60/90 roadmap

What to focus on in each month, and the ideas that carry it.

  1. Days 1 to 307 ideas

    Cost-per-token calculator, crawler-readable stats, homepage title, video facade, the 2023 calculator and one-day cost-analysis replies; start the tag audit.

  2. Days 31 to 6013 ideas

    Comparison pages, AI-cloud listings, measured stories, Mandarin benchmarks, Japanese and Korean pages, OEM pages, partner kits, win-loss, Learn refresh, newsletter, repurposing and the design-partner program; the CFO page.

  3. Days 61 to 903 ideas

    Commit to the big bets: the public cost-per-token benchmark, customer tokens-per-GPU reports and a monthly hands-on lab.

The research behind the plan

3 research areas, each summarised in a few lines. Open any of them for the full detail.

Customers

What customers say

Customers value more tokens and users from the same GPUs. The most cited problem: GPU memory limits and KV-cache eviction.

  • Fast model loading and deployment
  • Rising cost per token as context grows
  • In their words: “cost per token”

Market

Competitive landscape

4 competitors compared: VAST Data, DDN, Hammerspace and Lustre and cloud file services.

  • Gap to exploit: A self-serve, checkable cost-per-token estimate
  • VAST Data: Broader platform may be heavier for focused inference needs
  • DDN: Legacy parallel file system image

Hiring signals

WEKA is hiring to win the inference market and to make its marketing a measured engine.

  • Senior Product Marketing Manager, AI Inference: Quantify cost per token; design partners. Ideas 1, 9, 18, 19.
  • Programming & Editorial Manager: Search, LLM citations and refreshing aging work. Ideas 2, 15, 16, 17.
  • Senior Product Marketing Manager, Competitive & Market Intelligence: Comparison frameworks and a win-loss loop. Ideas 7, 14.

From the team that wrote this playbook

All products
  • ProgrammableAgencyWorking AI systems in days, not quarters (opens programmable.agency in a new tab)
  • ProgrammableInfoAn agent that finds, researches and asks first (opens programmable.info in a new tab)
  • ProgrammableDigitalScreenshots in, an animated product demo out (opens programmable.digital in a new tab)
  • ProgrammableWatchKnow the day a supplier falls out of compliance (opens programmable.watch in a new tab)
  • ProgrammableTalkGet registrants for your events, and know who’s coming (opens programmable.talk in a new tab)
  • ProgrammablePageText what you need, your customer gets a page that works (opens programmable.page in a new tab)
  • ProgrammableCommunityForums with labelled AI members, never empty (opens programmable.community in a new tab)
ProgrammableMarketing
ExamplesLeaderboardsProductsTermsPrivacyCookiesDPALegalContact

© 2026 AI-Powered Growth Intelligence