ProgrammableMarketing/lunaprompts.com (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 · 9
  • Big bets · 3
  • Wildcards · 5
  • How we can help
Generated Nov 4, 2026 1 view
  1. Programmable
  2. Playbooks
  3. LunaPrompts
LunaPrompts logo

growth playbook · Nov 4, 2026

lunaprompts.com (opens in a new tab)

LunaPrompts growth playbook

Diagnosis

LunaPrompts has a sharp position (AI in the test, graded on craft) and strong assets: 13 sourced compare pages, a 2,747-model directory, named customer results and a free community.

The gaps are in getting found and converting: 14 key pages, including every feature page and the Greenhouse doc, serve the homepage's HTML and canonical, the homepage shows no customer proof, mobile speed fails Core Web Vitals, and the directory sends no traffic to the product. With a growth intern starting outreach, the first steps are fixing those pages and the domain's email policy, then building outbound from open AI engineering roles.

Growth Score

6/10

Biggest Opportunity

Make every page that already exists findable and pointed at a pilot, then feed the intern's outreach with companies hiring AI engineers now.

Biggest Risk

HackerRank, CodeSignal and Codility are adding AI-in-the-test features with free trials and enterprise security, so the window to own "AI-native screen" is short.

01

Start here

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

  1. 1

    Give the six feature pages, the Greenhouse page and the community their own HTML, so buyers and Google can find them

    View the page source of /features/integrity and confirm the title and canonical are the homepage's.

    High impact First result in 2-4 weeks
  2. 2

    Put the Amber and JupiterBrains results on the homepage, next to the demo button

    Recount total candidates screened across all customers.

    High impact First result in 2 weeks
  3. 3

    Bring mobile load under Google's limits: real visitors wait 4 seconds and taps lag

    Run the PageSpeed report and list the bundles the homepage loads.

    High impact First result in 1 month

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 · 2Big bets · 7Fill-ins · 8Reconsider · 6
Effort →

Showing 23 of 23 ideas

This week 6

Quick wins with a clear first action.

  • 01

    LunaPrompts is hiring a growth and GTM intern to generate leads, and the pages that intern will send buyers to are invisible to search. Fourteen sitemap URLs, including all six /features pages, the Greenhouse integration doc, /community, /challenges and /leaderboard, serve the homepage's HTML: the same title, description, H1 and a canonical tag pointing to the homepage. Google reads that as "these are copies of the homepage" and drops them, so a search for "AI engineer assessment with Greenhouse" or "prompt engineering challenge" cannot land on the page that answers it. Missing addresses such as /pricing also return 200 with the homepage, which hides broken links. The blog, compare and model pages already render correctly, so the fix is to add these routes to the same pre-render list.

    High impactLow effort First result in 2-4 weeks Founding engineer

    First action

    View the page source of /features/integrity and confirm the title and canonical are the homepage's.

    How to measure it

    Feature, docs and community pages indexed in Google

    Now
    –
    Target
    14 of 14
    By
    30 days

    Steps

    1. 1

      Confirm the problem

      Fetch each of the 14 URLs without JavaScript and record title and canonical.

      Founding engineer 1 hour

    2. 2

      Add routes to pre-render

      Same pipeline as /blog and /compare.

      Founding engineer 1 day

    3. 3

      Write page metadata

      Unique title, description and H1 for each feature page.

      Growth & GTM Intern 1 day

    4. 4

      Self-canonicals

      Each page points to itself, not to /.

      Founding engineer 1 hour

    5. 5

      Real 404s

      Unknown paths return status 404 with a helpful page.

      Founding engineer 2 hours

    6. 6

      Resubmit

      Sitemap and URL inspection in Search Console.

      Growth & GTM Intern 1 hour

    7. 7

      Outreach links

      One tracked link per feature for the intern's emails.

      Growth & GTM Intern 1 day

    8. 8

      Measure

      Impressions and clicks per page weekly.

      Growth & GTM Intern weekly

    First result
    2 to 4 weeks
    Cost
    Engineering time only
    Owner
    Founding engineer

    Smallest test

    Fix /features/integrity and /docs/integrations/greenhouse first and check them in URL inspection.

    Why it works

    Add the 14 routes to the pre-render step the blog and compare pages already use, each with its own title, meta description, H1, body copy and self-referencing canonical. Return a real 404 status for unknown paths. Resubmit the sitemap in Google Search Console and request indexing for the feature and Greenhouse pages. Then give the intern a tracked link for each feature page to use in outreach, so every email points at the exact capability the prospect cares about (integrity, ATS, analytics).

    Feature and integration pages are what evaluators search for after a demo and what outbound links to; they only work if they exist for crawlers and link previews. The comparison pages prove the team already knows how to do this.

    Ready to use

    ChecklistRoute fix list

    [ ] /features/assess
    [ ] /features/analytics
    [ ] /features/integrity
    [ ] /features/scale
    [ ] /features/candidate-discovery
    [ ] /features/ats-integrations
    [ ] /docs and /docs/integrations/greenhouse
    [ ] /community, /challenges, /leaderboard
    [ ] /enterprise (redirect 301 to / instead)
    [ ] /privacy and /terms
    [ ] Unknown paths return 404
    [ ] Sitemap resubmitted

    Risks and how to handle them

    • Pre-rendering adds build time

      Only 14 more routes; cache them.

    • Thin pages

      Write at least 300 words and a screenshot per feature.

    Evidence

    • Your websiteFourteen URLs in the sitemap return the homepage's HTML instead of their own: /features/assess, /features/analytics, /features/integrity, /features/scale, /features/candidate-discovery, /features/ats-integrations, /docs, /docs/integrations/greenhouse, /community, /challenges, /leaderboard, /enterprise, /privacy and /terms all come back with the homepage title ("LunaPrompts · Hire AI engineers who actually ship"), its description, its H1 and a canonical tag pointing to https://lunaprompts.com/. Addresses that do not exist, such as /pricing or /careers, also return status 200 with the same homepage HTML. Blog, compare and model pages are served with their own titles and canonicals.Source
    • HiringLunaPrompts is hiring a Growth & GTM Intern (remote) on Wellfound, with lead generation, market research, cold calling, B2B sales and outreach listed as the skills. The listing describes the company as a technical hiring and assessment platform helping companies evaluate engineering and AI talent at scale.Source
    • Your websiteCompare section: 13 pages "Reviewed 12 October 2026" with sources, a "Build your comparison" picker and an export table: LunaPrompts vs HackerRank, CodeSignal, Coderbyte and TestGorilla, plus third-party pairs such as HackerRank vs CodeSignal. The homepage names Codility as part of "the legacy assessment stack", but there is no Codility page.Source
    • Your websiteHomepage (checked November 4, 2026): "Hire AI engineers who actually ship." "LunaPrompts is the AI-native technical screen." Tests cover prompt engineering with a model picker, multi-file agent code in a Judge0 sandbox, algorithms in 9 languages, SQL on DuckDB, rubric-graded system design and multiple choice. Proof line: "Founder-led pilots. 1,000+ engineers screened." Calls to action: "Book a 20-minute founder-led demo" (one Calendly link) and "Try a sample assessment" (links to /challenges). No customer logos or testimonials on the page.Source
  • 02

    The homepage proof is one line, "1,000+ engineers screened", with no logos or quotes. Yet the blog has named results: Amber screened 2,000+ candidates and went from 45 days to 2 weeks, JupiterBrains cut screening effort by 72%. Amber's story alone is double the homepage number.

    High impactLow effort First result in 2 weeks Founder

    First action

    Recount total candidates screened across all customers.

    How to measure it

    Demo bookings per 100 homepage visits

    Now
    –
    Target
    +25%
    By
    60 days

    Steps

    1. 1

      Recount

      True candidates-screened total.

      Founder 1 hour

    2. 2

      Get approvals

      Logo and quote use from both customers.

      Founder 3 days

    3. 3

      Design strip

      Logos, numbers, quotes, links.

      Founding engineer 1 day

    4. 4

      Update the count

      Homepage, meta description and llms.txt.

      Founding engineer 1 hour

    5. 5

      Reuse

      Feature pages and outreach emails.

      Growth & GTM Intern 1 day

    6. 6

      Measure

      Demo clicks before and after.

      Growth & GTM Intern weekly

    First result
    2 weeks
    Cost
    None
    Owner
    Founder

    Smallest test

    Add the Amber numbers under the hero today.

    Why it works

    Add a results strip under the hero: two logos, the headline numbers and the named quotes, each linking to the full story. Update the screened count to the true total. Reuse the strip on feature pages and in the intern's emails.

    A named buyer with a number beats any feature list for a founder-led sale, and it costs a day.

    Risks and how to handle them

    • Customers decline logo use

      Use industry and numbers without the name.

    Evidence

    • Your websiteHomepage (checked November 4, 2026): "Hire AI engineers who actually ship." "LunaPrompts is the AI-native technical screen." Tests cover prompt engineering with a model picker, multi-file agent code in a Judge0 sandbox, algorithms in 9 languages, SQL on DuckDB, rubric-graded system design and multiple choice. Proof line: "Founder-led pilots. 1,000+ engineers screened." Calls to action: "Book a 20-minute founder-led demo" (one Calendly link) and "Try a sample assessment" (links to /challenges). No customer logos or testimonials on the page.Source
    • Your websiteCustomer stories on the blog: Amber, a student accommodation platform, "screened 2000+ candidates", cut time-to-hire "from 45 days to 2 weeks" and closed 5 positions from a top-25 shortlist, with a quote from its Talent Acquisition Lead; JupiterBrains cut "screening effort by 72% and time-to-hire by 52%", with a founder quote; a third story describes an anonymised "Global Fintech (2,400 engineers)" upskilling with "+68% accuracy", although its URL slug names a company. The homepage does not link to any of them.Source
  • 03

    Real-visitor data fails Core Web Vitals: content appears at 4.0 s and interactions take 362 ms. The lab test shows 323 KB of unused JavaScript and a 1.3 s server response. A product that sells engineering quality should not feel slow on a phone.

    High impactMedium effort First result in 1 month Founding engineer

    First action

    Run the PageSpeed report and list the bundles the homepage loads.

    How to measure it

    Field LCP and INP

    Now
    –
    Target
    Under 2.5 s and 200 ms
    By
    60 days

    Steps

    1. 1

      Profile

      Bundles loaded on /.

      Founding engineer 1 day

    2. 2

      Split code

      Lazy-load charts, motion and the database client.

      Founding engineer 3 days

    3. 3

      Defer tags

      Tag Manager and analytics after load.

      Founding engineer 1 day

    4. 4

      Images

      Modern formats and sizes.

      Founding engineer 1 day

    5. 5

      Cache

      Long cache headers on hashed assets.

      Founding engineer 1 hour

    6. 6

      Retest

      PageSpeed and field data monthly.

      Growth & GTM Intern monthly

    First result
    1 month (field data lags 28 days)
    Cost
    Engineering time
    Owner
    Founding engineer

    Smallest test

    Defer Tag Manager and retest.

    Why it works

    Split the bundle so the homepage does not load charts, motion and the database client; defer Tag Manager and analytics until after load; pre-render the hero; serve images in modern formats with sizes.

    Faster pages lift conversion and search ranking, and evaluators often open the link on a phone first.

    Risks and how to handle them

    • Analytics gaps

      Check events still fire after deferring.

    Evidence

    • Google PageSpeedMobile performance 48/100 (desktop 81). Real-visitor data fails Core Web Vitals on two measures: Largest Contentful Paint 4.0 s at the 75th percentile (target 2.5 s) and Interaction to Next Paint 362 ms (target 200 ms); server response 1.3 s. In the lab test the main content takes 9.3 s; 637 KB of JavaScript of which 323 KB is unused ("Reduce unused JavaScript" about 1.7 s), and Google Tag Manager blocks the main thread for about 0.4 s. Chrome UX Report shows LCP improving by 30% over recent months.Source
  • 04

    The growth intern will run outreach from lunaprompts.com, but the domain's DMARC policy is "p=none". Inbox providers increasingly expect enforcement from domains that send cold email, and a spoofed message in a recruiter's inbox would hurt a trust-based sale.

    Medium impactLow effort First result in 1 month Growth & GTM Intern

    First action

    Add a reporting address to the DMARC record.

    How to measure it

    Outreach reply rate

    Now
    –
    Target
    5%+
    By
    60 days

    Steps

    1. 1

      Add reporting

      rua address in DMARC.

      Founding engineer 10 minutes

    2. 2

      Review reports

      Two weeks of data.

      Growth & GTM Intern 2 weeks

    3. 3

      Fix senders

      Calendly, Workspace and any tools.

      Founding engineer 1 day

    4. 4

      Quarantine

      Move policy.

      Founding engineer 10 minutes

    5. 5

      Sending domain

      Separate domain or subdomain for cold email, warmed.

      Growth & GTM Intern 2 weeks

    6. 6

      Reject

      After clean reports.

      Founding engineer 1 month

    First result
    1 month
    Cost
    None
    Owner
    Growth & GTM Intern

    Smallest test

    Add the rua tag today.

    Why it works

    Read the DMARC reports for two weeks, fix any sender that fails, move to "p=quarantine" and then "p=reject". Send outreach from a subdomain or a separate sending domain so the main domain's reputation is protected.

    Deliverability is the ceiling on outbound; it is cheap to fix before volume starts.

    Risks and how to handle them

    • Legit mail blocked

      Move in stages and watch reports.

    Evidence

    • DNS checklunaprompts.com sends mail through Google Workspace with SPF ("~all") and DKIM, but its DMARC policy is "p=none", so mail that fails checks is still delivered and the domain gives inbox providers no enforcement signal.Source
    • HiringLunaPrompts is hiring a Growth & GTM Intern (remote) on Wellfound, with lead generation, market research, cold calling, B2B sales and outreach listed as the skills. The listing describes the company as a technical hiring and assessment platform helping companies evaluate engineering and AI talent at scale.Source
  • 05

    The homepage calls HackerRank, CodeSignal and Codility "the legacy assessment stack", and there are compare pages for four rivals, but none for Codility, which now markets tests "where candidates use Claude Code or Codex" with a free trial. That is exactly LunaPrompts' claim.

    Medium impactLow effort First result in 2 months Content writer

    First action

    Read Codility's AI assessment pages and pricing.

    How to measure it

    Demos from compare pages

    Now
    –
    Target
    +3 a month
    By
    90 days

    Steps

    1. 1

      Research

      Codility product and pricing pages.

      Content writer 1 day

    2. 2

      Write

      Same template and sources.

      Content writer 2 days

    3. 3

      Review

      Facts and fairness.

      Founder 1 hour

    4. 4

      Publish

      Add to picker and sitemap.

      Founding engineer 1 hour

    5. 5

      Promote

      Send to Codility users in outreach.

      Growth & GTM Intern ongoing

    6. 6

      Update

      Re-review quarterly.

      Content writer quarterly

    First result
    2 months
    Cost
    Writing time
    Owner
    Content writer

    Smallest test

    Draft the table section first.

    Why it works

    Write LunaPrompts vs Codility in the same sourced format, focusing on AI-in-the-test grading, agent code and pilot terms against a two-week trial. Add Codility to the comparison picker.

    Buyers who already use Codility search for alternatives; the page catches them and gives sales a competitor comparison.

    Risks and how to handle them

    • Competitor changes pricing

      Review dates are already shown; keep them current.

    Evidence

    • Your websiteCompare section: 13 pages "Reviewed 12 October 2026" with sources, a "Build your comparison" picker and an export table: LunaPrompts vs HackerRank, CodeSignal, Coderbyte and TestGorilla, plus third-party pairs such as HackerRank vs CodeSignal. The homepage names Codility as part of "the legacy assessment stack", but there is no Codility page.Source
    • Competitor researchRivals' own homepages: HackerRank, "the market-leading coding test and interview solution for hiring developers"; CodeSignal, an "Agentic Skills Intelligence & Validation Platform"; Codility, "Online coding tests and interviews where candidates use Claude Code or Codex", with "a 2-week free trial"; TestGorilla, "AI-Powered Talent Sourcing & Skills Assessments" with "the world's largest skills-tested talent pool".Source
    • Your websiteHomepage (checked November 4, 2026): "Hire AI engineers who actually ship." "LunaPrompts is the AI-native technical screen." Tests cover prompt engineering with a model picker, multi-file agent code in a Judge0 sandbox, algorithms in 9 languages, SQL on DuckDB, rubric-graded system design and multiple choice. Proof line: "Founder-led pilots. 1,000+ engineers screened." Calls to action: "Book a 20-minute founder-led demo" (one Calendly link) and "Try a sample assessment" (links to /challenges). No customer logos or testimonials on the page.Source
  • 06

    There is no pricing page, and /pricing shows the homepage. The pilot terms (one role, 10-30 candidates, 7 days, readout) and the honest list of what is not shipped (SSO, SOC 2) sit in the FAQ and llms.txt. Buyers want both before a call.

    Medium impactLow effort First result in 1 month Founder

    First action

    Copy the FAQ answers into a draft page.

    How to measure it

    Pilot-to-paid conversion

    Now
    –
    Target
    40%+
    By
    1 quarter

    Steps

    1. 1

      Draft

      Timeline, inclusions, terms.

      Founder 1 day

    2. 2

      Security section

      Status and dates.

      Founder 1 hour

    3. 3

      Build

      Pre-rendered page.

      Founding engineer 1 day

    4. 4

      Link

      Nav, footer, emails.

      Growth & GTM Intern 1 hour

    5. 5

      Track

      Visits and bookings.

      Growth & GTM Intern weekly

    6. 6

      Iterate

      Add buyer questions.

      Founder monthly

    First result
    1 month
    Cost
    None
    Owner
    Founder

    Smallest test

    Publish the 7-day timeline as its own page.

    Why it works

    Publish /pilot (or /pricing) with the 7-day timeline, what you send, what you get, subscription terms (unlimited tests and seats), the security status and the roadmap for SSO and SOC 2, with the Calendly button.

    Clear terms shorten the first call and qualify buyers; honesty on SOC 2 avoids late surprises.

    Risks and how to handle them

    • Prices invite comparison

      Show "from" or request a quote.

    Evidence

    • Your websitePilot and pricing (homepage FAQ and llms.txt): "Day 0: Pick a role", "Days 1-7: Run candidates. 10 to 30 invites", "Day 7: 20-minute readout ... side-by-side with your current screen." "Subscriptions include unlimited tests and authorized team seats, with no per-test or per-seat charge." There is no pricing page; llms.txt lists as not shipped: SAML SSO ("on roadmap"), SOC 2 ("not certified yet"), custom-domain white label and a public partner API.Source
    • Your websiteFourteen URLs in the sitemap return the homepage's HTML instead of their own: /features/assess, /features/analytics, /features/integrity, /features/scale, /features/candidate-discovery, /features/ats-integrations, /docs, /docs/integrations/greenhouse, /community, /challenges, /leaderboard, /enterprise, /privacy and /terms all come back with the homepage title ("LunaPrompts · Hire AI engineers who actually ship"), its description, its H1 and a canonical tag pointing to https://lunaprompts.com/. Addresses that do not exist, such as /pricing or /careers, also return status 200 with the same homepage HTML. Blog, compare and model pages are served with their own titles and canonicals.Source

This quarter 9

The core plays for the next three months.

  • 07

    The model directory is the biggest content asset (2,927 URLs), and it attracts exactly the AI engineers and leads LunaPrompts serves. But model pages link only to /models and the homepage: no challenge, practice or hiring link.

    High impactMedium effort First result in 1 month Founding engineer

    First action

    Pick the 20 most-visited model pages.

    How to measure it

    Challenge starts from model pages

    Now
    –
    Target
    500 a month
    By
    1 quarter

    Steps

    1. 1

      Find top pages

      Analytics by model page.

      Growth & GTM Intern 1 day

    2. 2

      Design CTAs

      Developer and buyer versions.

      Founding engineer 2 days

    3. 3

      Map challenges

      Model to challenge.

      Founding engineer 1 week

    4. 4

      Roll out

      All model pages.

      Founding engineer 1 week

    5. 5

      Track

      Clicks and signups.

      Growth & GTM Intern weekly

    6. 6

      Test copy

      Two CTA versions.

      Growth & GTM Intern 1 month

    First result
    1 month
    Cost
    Engineering time
    Owner
    Founding engineer

    Smallest test

    Add one CTA to the 20 top pages.

    Why it works

    Add two calls to action on every model page: "Test your prompting on this model" (a challenge that uses it in the model picker) and "Hiring engineers who build on this model? Run a pilot". Track both.

    Programmatic pages only grow the business when they route traffic; contextual offers convert far better than a footer.

    Risks and how to handle them

    • Directory feels salesy

      Keep CTAs small and relevant.

    Evidence

    • Your websiteModel directory: /models lists "All companies 2747" models from OpenAI (99), Anthropic (35), Google (120), Alibaba Cloud (140), IBM (108) and dozens more, with "documented limits, endpoint pricing and access, with sources and review dates"; 2,927 URLs in sitemap-models.xml. The /models page HTML is about 4.2 MB. Model pages link only to /models and the homepage, with no challenge, practice or demo link in their HTML.Source
    • Your websiteHomepage (checked November 4, 2026): "Hire AI engineers who actually ship." "LunaPrompts is the AI-native technical screen." Tests cover prompt engineering with a model picker, multi-file agent code in a Judge0 sandbox, algorithms in 9 languages, SQL on DuckDB, rubric-graded system design and multiple choice. Proof line: "Founder-led pilots. 1,000+ engineers screened." Calls to action: "Book a 20-minute founder-led demo" (one Calendly link) and "Try a sample assessment" (links to /challenges). No customer logos or testimonials on the page.Source
    • Your websiteBlog: 30 posts, the latest dated October 11, 2026, including "hiring field guides" (forward-deployed engineer hiring, AI cheating detection, system design assessments, ATS integration, unlimited-assessment pricing) and "Introducing Community Test Cases" (January 2026), where developers "earn credits" for test cases. Community: challenges, contests and a leaderboard "free to use", plus a Discord server.Source
  • 08

    The intern's job is lead generation and outreach, and LunaPrompts' pilot starts with "one open AI engineering req". Companies with an open AI engineer role on their job boards are the exact buyers, with a reason to talk this week.

    High impactMedium effort First result in 1 month Growth & GTM Intern

    First action

    List 50 companies with open AI engineer roles.

    How to measure it

    Pilots started from outbound

    Now
    –
    Target
    4 a month
    By
    1 quarter

    Steps

    1. 1

      Define the trigger

      Role titles and seniority.

      Founder 1 day

    2. 2

      Build the list

      50 companies weekly.

      Growth & GTM Intern weekly

    3. 3

      Find contacts

      Hiring manager and talent lead.

      Growth & GTM Intern weekly

    4. 4

      Write the sequence

      Three short emails per role.

      Founder 2 days

    5. 5

      Attach a sample

      Challenge from their job post.

      Founding engineer per lead

    6. 6

      Send

      From the warmed domain.

      Growth & GTM Intern daily

    7. 7

      Book

      Founder-led demos.

      Founder ongoing

    8. 8

      Review

      Reply and pilot rates weekly.

      Founder weekly

    First result
    1 month
    Cost
    Data tool and intern time
    Owner
    Growth & GTM Intern

    Smallest test

    Send 20 role-specific notes this week.

    Why it works

    Each week, pull open AI engineer, LLM engineer and forward-deployed engineer roles from public job boards, find the hiring manager or head of talent, and send a short note offering a free pilot for that specific role, with a sample challenge built from its job description.

    Timing beats targeting: an open req is a live pain, and a pilot for that role is an easy yes.

    Risks and how to handle them

    • Low reply rates

      Make every note role-specific.

    Builds on

    Evidence

    • HiringLunaPrompts is hiring a Growth & GTM Intern (remote) on Wellfound, with lead generation, market research, cold calling, B2B sales and outreach listed as the skills. The listing describes the company as a technical hiring and assessment platform helping companies evaluate engineering and AI talent at scale.Source
    • Your websitePilot and pricing (homepage FAQ and llms.txt): "Day 0: Pick a role", "Days 1-7: Run candidates. 10 to 30 invites", "Day 7: 20-minute readout ... side-by-side with your current screen." "Subscriptions include unlimited tests and authorized team seats, with no per-test or per-seat charge." There is no pricing page; llms.txt lists as not shipped: SAML SSO ("on roadmap"), SOC 2 ("not certified yet"), custom-domain white label and a public partner API.Source
    • Your websiteBlog: 30 posts, the latest dated October 11, 2026, including "hiring field guides" (forward-deployed engineer hiring, AI cheating detection, system design assessments, ATS integration, unlimited-assessment pricing) and "Introducing Community Test Cases" (January 2026), where developers "earn credits" for test cases. Community: challenges, contests and a leaderboard "free to use", plus a Discord server.Source
  • 09

    The pilot promises a "custom challenge mix assembled within 24 hours", and the blog has a job-description-to-assessment guide. Buyers can only reach that by booking a call. A page that turns a pasted job description into a sample assessment would let them see the product first.

    High impactMedium effort First result in 2 months Founding engineer

    First action

    Write five sample job descriptions and their mixes by hand.

    How to measure it

    Pilots started from /try

    Now
    –
    Target
    10 a month
    By
    1 quarter

    Steps

    1. 1

      Define output

      Mix and one sample task.

      Founder 1 week

    2. 2

      Build

      Paste, preview, send.

      Founding engineer 1 month

    3. 3

      Guardrails

      Rate limit and review.

      Founding engineer 1 week

    4. 4

      Launch

      Homepage and outreach.

      Growth & GTM Intern 1 week

    5. 5

      Follow up

      Founder emails every user.

      Founder daily

    6. 6

      Measure

      Starts to pilots.

      Growth & GTM Intern weekly

    First result
    2 months
    Cost
    Engineering time and model costs
    Owner
    Founding engineer

    Smallest test

    Offer the same thing by email first.

    Why it works

    Build /try: paste a job description, get a preview of the assessment mix and one sample task, then a button to send it to five candidates as a free pilot.

    Letting buyers see their own role assessed turns curiosity into a pilot without a calendar slot.

    Risks and how to handle them

    • Weak previews

      Hand-review the first 50.

    Evidence

    • Your websitePilot and pricing (homepage FAQ and llms.txt): "Day 0: Pick a role", "Days 1-7: Run candidates. 10 to 30 invites", "Day 7: 20-minute readout ... side-by-side with your current screen." "Subscriptions include unlimited tests and authorized team seats, with no per-test or per-seat charge." There is no pricing page; llms.txt lists as not shipped: SAML SSO ("on roadmap"), SOC 2 ("not certified yet"), custom-domain white label and a public partner API.Source
    • Your websiteBlog: 30 posts, the latest dated October 11, 2026, including "hiring field guides" (forward-deployed engineer hiring, AI cheating detection, system design assessments, ATS integration, unlimited-assessment pricing) and "Introducing Community Test Cases" (January 2026), where developers "earn credits" for test cases. Community: challenges, contests and a leaderboard "free to use", plus a Discord server.Source
    • Your websiteHomepage (checked November 4, 2026): "Hire AI engineers who actually ship." "LunaPrompts is the AI-native technical screen." Tests cover prompt engineering with a model picker, multi-file agent code in a Judge0 sandbox, algorithms in 9 languages, SQL on DuckDB, rubric-graded system design and multiple choice. Proof line: "Founder-led pilots. 1,000+ engineers screened." Calls to action: "Book a 20-minute founder-led demo" (one Calendly link) and "Try a sample assessment" (links to /challenges). No customer logos or testimonials on the page.Source
  • 10

    llms.txt lists the roles LunaPrompts is built for: AI Engineer, LLM Engineer, RAG Engineer, Agent Engineer, GenAI Engineer, Prompt Engineer and more. There is a forward-deployed engineer guide but no page per role.

    Medium impactMedium effort First result in 2 months Content writer

    First action

    Start with AI Engineer and RAG Engineer.

    How to measure it

    Role-page demos

    Now
    –
    Target
    5 a month
    By
    2 quarters

    Steps

    1. 1

      Pick roles

      Top eight.

      Founder 1 day

    2. 2

      Template

      Skills, task, scorecard.

      Content writer 2 days

    3. 3

      Write

      Two a week.

      Content writer 1 month

    4. 4

      Sample tasks

      Real challenges.

      Founding engineer 1 month

    5. 5

      Link

      From blog and outreach.

      Growth & GTM Intern ongoing

    6. 6

      Measure

      Traffic and demos.

      Growth & GTM Intern monthly

    First result
    2 months
    Cost
    Writing time
    Owner
    Content writer

    Smallest test

    Publish the AI Engineer page.

    Why it works

    Write one page per role with the skills to test, a sample task, a scorecard and an interview plan, each linking to a pilot for that role.

    Hiring managers search by role; role pages rank for intent and give outbound a link per role.

    Risks and how to handle them

    • Generic content

      Use real tasks and data.

    Evidence

    • Your websitePilot and pricing (homepage FAQ and llms.txt): "Day 0: Pick a role", "Days 1-7: Run candidates. 10 to 30 invites", "Day 7: 20-minute readout ... side-by-side with your current screen." "Subscriptions include unlimited tests and authorized team seats, with no per-test or per-seat charge." There is no pricing page; llms.txt lists as not shipped: SAML SSO ("on roadmap"), SOC 2 ("not certified yet"), custom-domain white label and a public partner API.Source
    • Your websiteBlog: 30 posts, the latest dated October 11, 2026, including "hiring field guides" (forward-deployed engineer hiring, AI cheating detection, system design assessments, ATS integration, unlimited-assessment pricing) and "Introducing Community Test Cases" (January 2026), where developers "earn credits" for test cases. Community: challenges, contests and a leaderboard "free to use", plus a Discord server.Source
  • 11

    LunaPrompts runs free challenges, contests and a leaderboard for AI engineers, and has a candidate-discovery feature. Companies want to meet the top scorers.

    Medium impactMedium effort First result in 2 months Growth & GTM Intern

    First action

    Ask two customers to sponsor a contest.

    How to measure it

    Paying sponsors

    Now
    –
    Target
    2 a month
    By
    2 quarters

    Steps

    1. 1

      Design

      Rules and opt-in.

      Founder 1 week

    2. 2

      Find sponsors

      Customers first.

      Growth & GTM Intern 2 weeks

    3. 3

      Build

      Sponsor branding.

      Founding engineer 2 weeks

    4. 4

      Promote

      Discord, Twitter, LinkedIn.

      Growth & GTM Intern monthly

    5. 5

      Introduce

      Top scorers to sponsor.

      Founder monthly

    6. 6

      Publish

      Winners and lessons.

      Content writer monthly

    First result
    2 months
    Cost
    Prizes
    Owner
    Growth & GTM Intern

    Smallest test

    Run one contest with one customer.

    Why it works

    Run sponsored monthly contests: a hiring company sets the challenge, top scorers opt in to be introduced, and the sponsor gets a pilot. Publish winners on the blog and Discord.

    Each contest is both developer content and a customer acquisition event.

    Risks and how to handle them

    • Privacy

      Explicit opt-in only.

    Builds on

    Evidence

    • Your websiteBlog: 30 posts, the latest dated October 11, 2026, including "hiring field guides" (forward-deployed engineer hiring, AI cheating detection, system design assessments, ATS integration, unlimited-assessment pricing) and "Introducing Community Test Cases" (January 2026), where developers "earn credits" for test cases. Community: challenges, contests and a leaderboard "free to use", plus a Discord server.Source
    • Your websiteFourteen URLs in the sitemap return the homepage's HTML instead of their own: /features/assess, /features/analytics, /features/integrity, /features/scale, /features/candidate-discovery, /features/ats-integrations, /docs, /docs/integrations/greenhouse, /community, /challenges, /leaderboard, /enterprise, /privacy and /terms all come back with the homepage title ("LunaPrompts · Hire AI engineers who actually ship"), its description, its H1 and a canonical tag pointing to https://lunaprompts.com/. Addresses that do not exist, such as /pricing or /careers, also return status 200 with the same homepage HTML. Blog, compare and model pages are served with their own titles and canonicals.Source
  • 12

    The site links Twitter, GitHub and Discord but no LinkedIn page, and has no newsletter. Heads of talent and engineering managers, the buyers, live on LinkedIn.

    Medium impactLow effort First result in 1 quarter Growth & GTM Intern

    First action

    Create the page and link it.

    How to measure it

    Demos from LinkedIn

    Now
    –
    Target
    4 a month
    By
    1 quarter

    Steps

    1. 1

      Create page

      Profile and link.

      Growth & GTM Intern 1 day

    2. 2

      Plan posts

      Three formats.

      Growth & GTM Intern 1 week

    3. 3

      Pull data

      Monthly insights.

      Founding engineer monthly

    4. 4

      Post

      Weekly.

      Founder weekly

    5. 5

      Newsletter

      Monthly digest.

      Growth & GTM Intern monthly

    6. 6

      Measure

      Followers and demos.

      Growth & GTM Intern monthly

    First result
    1 quarter
    Cost
    Time only
    Owner
    Growth & GTM Intern

    Smallest test

    Post the Amber result this week.

    Why it works

    Create the page, link it in the footer, and post weekly: one data point from assessments (how candidates prompt, common failure modes), one customer result, one guide. Add a monthly email for buyers.

    Founder-led sales runs on visibility with the buyer group; data from real assessments is content no rival can copy.

    Risks and how to handle them

    • Inconsistent posting

      Batch a month at a time.

    Evidence

    • Your websiteSocial links in the footer and page HTML: Twitter (@LunaPrompts), GitHub (luna-prompts) and Discord; no LinkedIn company page is linked anywhere, and the site has no newsletter sign-up. The domain was registered in February 2025.Source
    • Your websiteCustomer stories on the blog: Amber, a student accommodation platform, "screened 2000+ candidates", cut time-to-hire "from 45 days to 2 weeks" and closed 5 positions from a top-25 shortlist, with a quote from its Talent Acquisition Lead; JupiterBrains cut "screening effort by 72% and time-to-hire by 52%", with a founder quote; a third story describes an anonymised "Global Fintech (2,400 engineers)" upskilling with "+68% accuracy", although its URL slug names a company. The homepage does not link to any of them.Source
    • Your websiteBlog: 30 posts, the latest dated October 11, 2026, including "hiring field guides" (forward-deployed engineer hiring, AI cheating detection, system design assessments, ATS integration, unlimited-assessment pricing) and "Introducing Community Test Cases" (January 2026), where developers "earn credits" for test cases. Community: challenges, contests and a leaderboard "free to use", plus a Discord server.Source
  • 13

    The FAQ says Greenhouse has a documented LunaPrompts assessment workflow and Ashby has an integration. Talent teams browse those marketplaces when choosing assessments.

    High impactMedium effort First result in 2 quarters Founder

    First action

    Check current marketplace listing status.

    How to measure it

    Demos from marketplaces

    Now
    –
    Target
    3 a month
    By
    2 quarters

    Steps

    1. 1

      Check status

      Both marketplaces.

      Founder 1 day

    2. 2

      Prepare listing

      Copy, screenshots, offer.

      Growth & GTM Intern 1 week

    3. 3

      Submit

      Partner review.

      Founder 1 month

    4. 4

      Fix doc page

      Own URL.

      Founding engineer 1 day

    5. 5

      Co-market

      Webinar or post.

      Founder 1 quarter

    6. 6

      Track

      Source of demos.

      Growth & GTM Intern monthly

    First result
    2 quarters
    Cost
    Partner fees if any
    Owner
    Founder

    Smallest test

    Email both partner teams.

    Why it works

    Get listed (or upgraded) in both marketplaces with screenshots, a pilot offer and the customer results; fix the Greenhouse doc page so it has its own URL; co-market with a webinar.

    Marketplaces bring buyers at the moment of choosing, and integration lowers the switching cost.

    Risks and how to handle them

    • Slow partner reviews

      Start now.

    Builds on

    Evidence

    • Your websitePilot and pricing (homepage FAQ and llms.txt): "Day 0: Pick a role", "Days 1-7: Run candidates. 10 to 30 invites", "Day 7: 20-minute readout ... side-by-side with your current screen." "Subscriptions include unlimited tests and authorized team seats, with no per-test or per-seat charge." There is no pricing page; llms.txt lists as not shipped: SAML SSO ("on roadmap"), SOC 2 ("not certified yet"), custom-domain white label and a public partner API.Source
    • Your websiteFourteen URLs in the sitemap return the homepage's HTML instead of their own: /features/assess, /features/analytics, /features/integrity, /features/scale, /features/candidate-discovery, /features/ats-integrations, /docs, /docs/integrations/greenhouse, /community, /challenges, /leaderboard, /enterprise, /privacy and /terms all come back with the homepage title ("LunaPrompts · Hire AI engineers who actually ship"), its description, its H1 and a canonical tag pointing to https://lunaprompts.com/. Addresses that do not exist, such as /pricing or /careers, also return status 200 with the same homepage HTML. Blog, compare and model pages are served with their own titles and canonicals.Source
    • Your websiteCustomer stories on the blog: Amber, a student accommodation platform, "screened 2000+ candidates", cut time-to-hire "from 45 days to 2 weeks" and closed 5 positions from a top-25 shortlist, with a quote from its Talent Acquisition Lead; JupiterBrains cut "screening effort by 72% and time-to-hire by 52%", with a founder quote; a third story describes an anonymised "Global Fintech (2,400 engineers)" upskilling with "+68% accuracy", although its URL slug names a company. The homepage does not link to any of them.Source
  • 14

    Every pilot ends with a 20-minute readout side-by-side with the current screen. That meeting decides the deal.

    Medium impactMedium effort First result in 1 quarter Founder

    First action

    Write the template from the last three readouts.

    How to measure it

    Pilot conversion

    Now
    –
    Target
    +15 points
    By
    1 quarter

    Steps

    1. 1

      Template

      One page.

      Founder 1 week

    2. 2

      Automate data

      From dashboard.

      Founding engineer 2 weeks

    3. 3

      Proposal

      Same-day.

      Founder ongoing

    4. 4

      Follow up

      Day 3 and day 10.

      Growth & GTM Intern ongoing

    5. 5

      Track

      Win rates.

      Growth & GTM Intern monthly

    6. 6

      Improve

      Monthly review.

      Founder monthly

    First result
    1 quarter
    Cost
    None
    Owner
    Founder

    Smallest test

    Use it on the next pilot.

    Why it works

    Standardise the readout: a one-page report per pilot (signal found vs current screen, time saved, integrity flags), a recommended rollout, and a proposal sent the same day.

    A consistent readout makes win rates measurable and improvable.

    Risks and how to handle them

    • Small samples

      Show ranges, not false precision.

    Evidence

    • Your websitePilot and pricing (homepage FAQ and llms.txt): "Day 0: Pick a role", "Days 1-7: Run candidates. 10 to 30 invites", "Day 7: 20-minute readout ... side-by-side with your current screen." "Subscriptions include unlimited tests and authorized team seats, with no per-test or per-seat charge." There is no pricing page; llms.txt lists as not shipped: SAML SSO ("on roadmap"), SOC 2 ("not certified yet"), custom-domain white label and a public partner API.Source
    • Your websiteCustomer stories on the blog: Amber, a student accommodation platform, "screened 2000+ candidates", cut time-to-hire "from 45 days to 2 weeks" and closed 5 positions from a top-25 shortlist, with a quote from its Talent Acquisition Lead; JupiterBrains cut "screening effort by 72% and time-to-hire by 52%", with a founder quote; a third story describes an anonymised "Global Fintech (2,400 engineers)" upskilling with "+68% accuracy", although its URL slug names a company. The homepage does not link to any of them.Source
  • 15

    One case study is not about hiring at all: a 2,400-engineer fintech used LunaPrompts challenges, leaderboards and tracks to raise prompt accuracy by 68% in 90 days. That is a budget line (learning) separate from hiring.

    Medium impactMedium effort First result in 2 quarters Founder

    First action

    Ask two customers if they would run it internally.

    How to measure it

    Upskilling contracts

    Now
    –
    Target
    2
    By
    2 quarters

    Steps

    1. 1

      Define package

      Tracks and programme.

      Founder 2 weeks

    2. 2

      Build page

      Own URL.

      Founding engineer 1 week

    3. 3

      Price

      Per team.

      Founder 1 week

    4. 4

      Pitch customers

      Existing first.

      Founder 1 month

    5. 5

      Deliver

      First programme.

      Founding engineer 1 quarter

    6. 6

      Publish result

      Case study.

      Content writer end

    First result
    2 quarters
    Cost
    Delivery time
    Owner
    Founder

    Smallest test

    Offer it to one customer.

    Why it works

    Package "AI fluency for your engineering team": role tracks, team leaderboards and a 90-day programme, with its own page and pricing, sold to existing customers first.

    Selling to employees as well as candidates widens the account and smooths revenue between hiring cycles.

    Risks and how to handle them

    • Distraction from hiring

      Only sell to existing accounts at first.

    Evidence

    • Your websiteCustomer stories on the blog: Amber, a student accommodation platform, "screened 2000+ candidates", cut time-to-hire "from 45 days to 2 weeks" and closed 5 positions from a top-25 shortlist, with a quote from its Talent Acquisition Lead; JupiterBrains cut "screening effort by 72% and time-to-hire by 52%", with a founder quote; a third story describes an anonymised "Global Fintech (2,400 engineers)" upskilling with "+68% accuracy", although its URL slug names a company. The homepage does not link to any of them.Source
    • Your websiteBlog: 30 posts, the latest dated October 11, 2026, including "hiring field guides" (forward-deployed engineer hiring, AI cheating detection, system design assessments, ATS integration, unlimited-assessment pricing) and "Introducing Community Test Cases" (January 2026), where developers "earn credits" for test cases. Community: challenges, contests and a leaderboard "free to use", plus a Discord server.Source

Big bets 3

Larger plays that can change the trajectory.

  • 16

    LunaPrompts grades iteration, model choice and eval reasoning across 1,000+ engineers, and it knows which models candidates pick and how often they pass. No rival publishes that.

    High impactHigh effort First result in 2 quarters Founder

    First action

    Pull model-choice counts from the last six months.

    How to measure it

    Leads from the report

    Now
    –
    Target
    200
    By
    2 quarters

    Steps

    1. 1

      Check consent

      Terms allow aggregate use.

      Founder 1 week

    2. 2

      Pull data

      Anonymised aggregates.

      Founding engineer 2 weeks

    3. 3

      Analyse

      Five findings.

      Founder 2 weeks

    4. 4

      Design

      Report and charts.

      Content writer 2 weeks

    5. 5

      Launch

      LinkedIn, press, email.

      Growth & GTM Intern 1 week

    6. 6

      Follow up

      Outreach to downloaders.

      Growth & GTM Intern 1 month

    First result
    2 quarters
    Cost
    Design time
    Owner
    Founder

    Smallest test

    Publish one chart on LinkedIn.

    Why it works

    Release an anonymised twice-yearly report: model choice, iteration counts, common failures and what separates top scorers. Gate the full PDF for buyers, share charts on LinkedIn, and pitch it to the tech press.

    Original data earns links, press and trust, and positions LunaPrompts as the authority on AI-era hiring.

    Risks and how to handle them

    • Small samples

      Report ranges and sample sizes.

    Evidence

    • Your websiteHomepage (checked November 4, 2026): "Hire AI engineers who actually ship." "LunaPrompts is the AI-native technical screen." Tests cover prompt engineering with a model picker, multi-file agent code in a Judge0 sandbox, algorithms in 9 languages, SQL on DuckDB, rubric-graded system design and multiple choice. Proof line: "Founder-led pilots. 1,000+ engineers screened." Calls to action: "Book a 20-minute founder-led demo" (one Calendly link) and "Try a sample assessment" (links to /challenges). No customer logos or testimonials on the page.Source
    • Your websiteModel directory: /models lists "All companies 2747" models from OpenAI (99), Anthropic (35), Google (120), Alibaba Cloud (140), IBM (108) and dozens more, with "documented limits, endpoint pricing and access, with sources and review dates"; 2,927 URLs in sitemap-models.xml. The /models page HTML is about 4.2 MB. Model pages link only to /models and the homepage, with no challenge, practice or demo link in their HTML.Source
    • Your websiteBlog: 30 posts, the latest dated October 11, 2026, including "hiring field guides" (forward-deployed engineer hiring, AI cheating detection, system design assessments, ATS integration, unlimited-assessment pricing) and "Introducing Community Test Cases" (January 2026), where developers "earn credits" for test cases. Community: challenges, contests and a leaderboard "free to use", plus a Discord server.Source
  • 17

    llms.txt is honest that SAML SSO is "on roadmap" and SOC 2 is "not certified yet". Larger hiring teams ask for both before a pilot. Rivals like HackerRank and CodeSignal have them.

    High impactHigh effort First result in 3 quarters Founding engineer

    First action

    Get quotes from two compliance tools.

    How to measure it

    Mid-market pilots

    Now
    –
    Target
    3
    By
    3 quarters

    Steps

    1. 1

      Choose tool

      Compliance platform.

      Founder 2 weeks

    2. 2

      Policies

      Write and adopt.

      Founding engineer 1 month

    3. 3

      Ship SSO

      SAML.

      Founding engineer 1 month

    4. 4

      Audit

      Type I.

      Founder 1 quarter

    5. 5

      Trust page

      Status and dates.

      Founding engineer 1 week

    6. 6

      Sell

      Larger targets.

      Growth & GTM Intern ongoing

    First result
    3 quarters
    Cost
    Compliance and audit fees
    Owner
    Founding engineer

    Smallest test

    Publish a trust page with current status.

    Why it works

    Start SOC 2 Type I with a compliance tool, ship SAML SSO, and publish a trust page with dates. Use it in outreach to bigger companies.

    Security reviews are the most common blocker for mid-market deals; a dated plan keeps deals moving.

    Risks and how to handle them

    • Cost for a small team

      Type I first.

    Evidence

    • Your websitePilot and pricing (homepage FAQ and llms.txt): "Day 0: Pick a role", "Days 1-7: Run candidates. 10 to 30 invites", "Day 7: 20-minute readout ... side-by-side with your current screen." "Subscriptions include unlimited tests and authorized team seats, with no per-test or per-seat charge." There is no pricing page; llms.txt lists as not shipped: SAML SSO ("on roadmap"), SOC 2 ("not certified yet"), custom-domain white label and a public partner API.Source
    • Competitor researchRivals' own homepages: HackerRank, "the market-leading coding test and interview solution for hiring developers"; CodeSignal, an "Agentic Skills Intelligence & Validation Platform"; Codility, "Online coding tests and interviews where candidates use Claude Code or Codex", with "a 2-week free trial"; TestGorilla, "AI-Powered Talent Sourcing & Skills Assessments" with "the world's largest skills-tested talent pool".Source
  • 18

    The community already has scored challenges and a leaderboard. A certificate that candidates add to LinkedIn spreads the brand to every recruiter who sees it.

    Medium impactHigh effort First result in 2 quarters Founder

    First action

    Define the certificate's challenge set.

    How to measure it

    Certificates issued

    Now
    –
    Target
    1,000
    By
    3 quarters

    Steps

    1. 1

      Design

      Challenges and pass mark.

      Founder 1 month

    2. 2

      Build

      Verification pages.

      Founding engineer 1 month

    3. 3

      Proctoring

      Existing FaceMesh tools.

      Founding engineer 2 weeks

    4. 4

      Launch

      Discord and LinkedIn.

      Growth & GTM Intern 1 month

    5. 5

      Company access

      Discovery filter.

      Founding engineer 1 quarter

    6. 6

      Measure

      Shares and demos.

      Growth & GTM Intern monthly

    First result
    2 quarters
    Cost
    Engineering time
    Owner
    Founder

    Smallest test

    Issue badges for the next contest.

    Why it works

    Offer a proctored certificate built from the hardest challenges, with a public verification page. Let companies filter certified engineers through candidate discovery.

    Candidate-shared credentials are free distribution into the buyer's feed.

    Risks and how to handle them

    • Cheating

      Proctored and refreshed questions.

    Evidence

    • Your websiteBlog: 30 posts, the latest dated October 11, 2026, including "hiring field guides" (forward-deployed engineer hiring, AI cheating detection, system design assessments, ATS integration, unlimited-assessment pricing) and "Introducing Community Test Cases" (January 2026), where developers "earn credits" for test cases. Community: challenges, contests and a leaderboard "free to use", plus a Discord server.Source
    • Your websiteHomepage (checked November 4, 2026): "Hire AI engineers who actually ship." "LunaPrompts is the AI-native technical screen." Tests cover prompt engineering with a model picker, multi-file agent code in a Judge0 sandbox, algorithms in 9 languages, SQL on DuckDB, rubric-graded system design and multiple choice. Proof line: "Founder-led pilots. 1,000+ engineers screened." Calls to action: "Book a 20-minute founder-led demo" (one Calendly link) and "Try a sample assessment" (links to /challenges). No customer logos or testimonials on the page.Source

Wildcards 5

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

  • 19

    The upskilling story calls its customer "Global Fintech (2,400 engineers)", but its URL slug names a company. Either the customer agreed to be named, in which case say so, or the URL undoes the anonymity.

    Medium impactLow effort First result in 1 week Founding engineer

    First action

    Check the agreement with the customer.

    How to measure it

    Mismatch resolved

    Now
    –
    Target
    Yes
    By
    1 week

    Steps

    1. 1

      Check agreement

      Naming rights.

      Founder 1 day

    2. 2

      Decide

      Name or anonymise.

      Founder 1 day

    3. 3

      Change slug

      New URL.

      Founding engineer 1 hour

    4. 4

      Redirect

      301.

      Founding engineer 1 hour

    5. 5

      Update sitemap

      And feeds.

      Founding engineer 1 hour

    6. 6

      Check others

      All stories.

      Content writer 1 day

    First result
    1 week
    Cost
    None
    Owner
    Founding engineer

    Smallest test

    Decide this week.

    Why it works

    Confirm with the customer. If anonymous, change the slug to a neutral one, 301-redirect the old URL and update the sitemap. If named, add the name and logo.

    Buyers judge how you handle data by how you handle customer names.

    Risks and how to handle them

    • Lost link value

      The redirect keeps it.

    Evidence

    • Your websiteCustomer stories on the blog: Amber, a student accommodation platform, "screened 2000+ candidates", cut time-to-hire "from 45 days to 2 weeks" and closed 5 positions from a top-25 shortlist, with a quote from its Talent Acquisition Lead; JupiterBrains cut "screening effort by 72% and time-to-hire by 52%", with a founder quote; a third story describes an anonymised "Global Fintech (2,400 engineers)" upskilling with "+68% accuracy", although its URL slug names a company. The homepage does not link to any of them.Source
  • 20

    The directory already tracks limits, pricing and review dates for 2,747 models. Model prices and versions change monthly; a change log is newsworthy to every AI engineer.

    Medium impactLow effort First result in 1 month Content writer

    First action

    Diff last month's model records.

    How to measure it

    Monthly post readers

    Now
    –
    Target
    2,000
    By
    2 quarters

    Steps

    1. 1

      Diff records

      Monthly script.

      Founding engineer 1 week

    2. 2

      Write post

      Top changes.

      Content writer monthly

    3. 3

      Add challenge

      Newest model.

      Founding engineer monthly

    4. 4

      Promote

      Twitter, Discord, LinkedIn.

      Growth & GTM Intern monthly

    5. 5

      Email

      Subscribers.

      Growth & GTM Intern monthly

    6. 6

      Measure

      Readers and signups.

      Growth & GTM Intern monthly

    First result
    1 month
    Cost
    Writing time
    Owner
    Content writer

    Smallest test

    Publish one changes post.

    Why it works

    Publish a monthly post and email: new models, price changes, deprecations, with links to model pages and a challenge on the newest model.

    Recurring, data-driven posts earn links and return visits from the exact audience.

    Risks and how to handle them

    • Errors in data

      Keep sources visible.

    Evidence

    • Your websiteModel directory: /models lists "All companies 2747" models from OpenAI (99), Anthropic (35), Google (120), Alibaba Cloud (140), IBM (108) and dozens more, with "documented limits, endpoint pricing and access, with sources and review dates"; 2,927 URLs in sitemap-models.xml. The /models page HTML is about 4.2 MB. Model pages link only to /models and the homepage, with no challenge, practice or demo link in their HTML.Source
    • Your websiteBlog: 30 posts, the latest dated October 11, 2026, including "hiring field guides" (forward-deployed engineer hiring, AI cheating detection, system design assessments, ATS integration, unlimited-assessment pricing) and "Introducing Community Test Cases" (January 2026), where developers "earn credits" for test cases. Community: challenges, contests and a leaderboard "free to use", plus a Discord server.Source
  • 21

    The /models hub page is about 4.2 MB of HTML, which slows crawling and makes it heavy on phones.

    Medium impactLow effort First result in 1 month Growth & GTM Intern

    First action

    Measure the page in PageSpeed.

    How to measure it

    /models HTML size

    Now
    –
    Target
    Under 300 KB
    By
    1 month

    Steps

    1. 1

      Measure

      Size and speed.

      Founding engineer 1 hour

    2. 2

      Design

      Index plus pages.

      Founding engineer 1 day

    3. 3

      Paginate

      Company and type.

      Founding engineer 3 days

    4. 4

      Test

      Crawl and speed.

      Founding engineer 1 day

    5. 5

      Ship

      Deploy.

      Founding engineer 1 day

    6. 6

      Monitor

      Search Console.

      Growth & GTM Intern monthly

    First result
    1 month
    Cost
    Engineering time
    Owner
    Growth & GTM Intern

    Smallest test

    Paginate by company first.

    Why it works

    Paginate or load the list on scroll, keep the company index and top models in HTML, and link to company pages.

    Lighter hubs crawl and rank better and pass authority to model pages.

    Risks and how to handle them

    • Losing internal links

      Keep pagination crawlable.

    Evidence

    • Your websiteModel directory: /models lists "All companies 2747" models from OpenAI (99), Anthropic (35), Google (120), Alibaba Cloud (140), IBM (108) and dozens more, with "documented limits, endpoint pricing and access, with sources and review dates"; 2,927 URLs in sitemap-models.xml. The /models page HTML is about 4.2 MB. Model pages link only to /models and the homepage, with no challenge, practice or demo link in their HTML.Source
    • Google PageSpeedMobile performance 48/100 (desktop 81). Real-visitor data fails Core Web Vitals on two measures: Largest Contentful Paint 4.0 s at the 75th percentile (target 2.5 s) and Interaction to Next Paint 362 ms (target 200 ms); server response 1.3 s. In the lab test the main content takes 9.3 s; 637 KB of JavaScript of which 323 KB is unused ("Reduce unused JavaScript" about 1.7 s), and Google Tag Manager blocks the main thread for about 0.4 s. Chrome UX Report shows LCP improving by 30% over recent months.Source
  • 22

    Community members earn credits for test cases already. Engineers who used LunaPrompts in a hiring process know whether their own company hires AI engineers.

    Medium impactLow effort First result in 1 quarter Founder

    First action

    Announce it in Discord.

    How to measure it

    Pilots from referrals

    Now
    –
    Target
    2 a month
    By
    1 quarter

    Steps

    1. 1

      Design reward

      Credits or gift.

      Founder 1 day

    2. 2

      Build form

      Simple intro form.

      Founding engineer 1 day

    3. 3

      Promote

      Discord and post-test page.

      Growth & GTM Intern ongoing

    4. 4

      Follow up

      Founder call.

      Founder ongoing

    5. 5

      Pay out

      On pilot start.

      Founder ongoing

    6. 6

      Measure

      Intros to pilots.

      Growth & GTM Intern monthly

    First result
    1 quarter
    Cost
    Rewards
    Owner
    Founder

    Smallest test

    Announce in Discord.

    Why it works

    Offer credits or a gift for an introduction to a hiring manager that leads to a pilot, promoted in Discord and after each assessment.

    Warm intros from engineers reach buyers outbound cannot.

    Risks and how to handle them

    • Candidates feel used

      Make it optional and after the test.

    Evidence

    • Your websiteBlog: 30 posts, the latest dated October 11, 2026, including "hiring field guides" (forward-deployed engineer hiring, AI cheating detection, system design assessments, ATS integration, unlimited-assessment pricing) and "Introducing Community Test Cases" (January 2026), where developers "earn credits" for test cases. Community: challenges, contests and a leaderboard "free to use", plus a Discord server.Source
    • HiringLunaPrompts is hiring a Growth & GTM Intern (remote) on Wellfound, with lead generation, market research, cold calling, B2B sales and outreach listed as the skills. The listing describes the company as a technical hiring and assessment platform helping companies evaluate engineering and AI talent at scale.Source
  • 23

    The product's difference, AI in the test, is hard to grasp in text. Watching a candidate pick a model, iterate and fail a hidden test makes it obvious.

    Medium impactMedium effort First result in 1 month Content writer

    First action

    Ask three community members to record.

    How to measure it

    Demos after video views

    Now
    –
    Target
    +2 a month
    By
    1 quarter

    Steps

    1. 1

      Get consent

      Community volunteers.

      Growth & GTM Intern 1 week

    2. 2

      Record

      Five sessions.

      Founding engineer 1 week

    3. 3

      Narrate

      Founder.

      Founder 1 week

    4. 4

      Edit

      Two minutes each.

      Content writer 1 week

    5. 5

      Publish

      Site and LinkedIn.

      Growth & GTM Intern 1 week

    6. 6

      Measure

      Views and demos.

      Growth & GTM Intern monthly

    First result
    1 month
    Cost
    Editing time
    Owner
    Content writer

    Smallest test

    Record one session.

    Why it works

    Record five short anonymised sessions with consent, narrated by the founder, and use them on feature pages, compare pages and in outreach.

    Video shows the grading craft in two minutes and suits LinkedIn.

    Risks and how to handle them

    • Privacy

      Blur names and faces.

    Evidence

    • Your websiteHomepage (checked November 4, 2026): "Hire AI engineers who actually ship." "LunaPrompts is the AI-native technical screen." Tests cover prompt engineering with a model picker, multi-file agent code in a Judge0 sandbox, algorithms in 9 languages, SQL on DuckDB, rubric-graded system design and multiple choice. Proof line: "Founder-led pilots. 1,000+ engineers screened." Calls to action: "Book a 20-minute founder-led demo" (one Calendly link) and "Try a sample assessment" (links to /challenges). No customer logos or testimonials on the page.Source
    • Your websiteCompare section: 13 pages "Reviewed 12 October 2026" with sources, a "Build your comparison" picker and an export table: LunaPrompts vs HackerRank, CodeSignal, Coderbyte and TestGorilla, plus third-party pairs such as HackerRank vs CodeSignal. The homepage names Codility as part of "the legacy assessment stack", but there is no Codility page.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.

ProgrammableInfo

Turn the research into a list you can act on

These ideas start with finding the right companies. Info finds them and the right person at each, researches every one and drafts the outreach, then asks before it sends.

  • Build this list in Info for idea #8 (opens programmable.info in a new tab)
  • Build this list in Info for idea #12 (opens programmable.info in a new tab)
  • Build this list in Info for idea #22 (opens programmable.info in a new tab)

New accounts: 100,000 free credits with code BETA100K

Find the companies (opens programmable.info 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.

  • Autumn growth strategyOpen-source billing infrastructure built on Stripe: one API (attach, check, track) for usage-based pricing, credits, entitlements, limits…23 growth ideas
  • Gopuff growth strategyInstant delivery of groceries, snacks, drinks, alcohol, household goods and Starbucks drinks from its own micro-fulfillment centers, with…22 growth ideas
  • ALOHA growth strategyMakes organic, plant-based protein bars, mini bars, protein drinks and powders, sold online with autoship and in about 24,000 stores.20 growth ideas
  • F2 growth strategyAn AI platform for credit investing ("the agentic operating system for global credit"). It screens deals, spreads financials, works inside…22 growth ideas
  • DeepJudge growth strategyEnterprise AI search and workflows for legal teams: it indexes a firm's documents where they live (iManage, NetDocuments, SharePoint and…22 growth ideas
  • Jena Jewelry growth strategyA small Shopify shop selling handmade jewelry by Diana Echevarria (stone stretch bracelets, Morse code "Say It With Code" bracelets…21 growth ideas
  • Aspire Proptech growth strategyA Bangalore real estate marketing and sales partner: it sells new residential projects for developers and guides buyers through site…21 growth ideas
  • Multiplier growth strategyAn agent platform for fundamental equity investors. Multiplier runs in each fund's own cloud and connects AI agents to the fund's data…23 growth ideas

See every playbook and website review

Independent research. Programmable wrote this playbook from public sources, as they stood on November 4, 2026, and cites them throughout. LunaPrompts did not commission, review or endorse it, and Programmable is not affiliated with LunaPrompts. 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 LunaPrompts runs its business.

Something wrong, or do you represent LunaPrompts? 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

    Page rendering fixes, homepage proof, speed, DMARC, the Codility page and a pilot page; rename the case study URL.

  2. Days 31 to 6013 ideas

    Directory CTAs, open-role outbound, self-serve sample, role library, contests, LinkedIn, marketplaces, readouts and upskilling; the model changes post, hub fix, referrals and videos.

  3. Days 61 to 903 ideas

    Commit to the assessment-data report, SOC 2 and SSO, and the certificate.

The research behind the plan

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

Customers

What customers say

Customers value faster time-to-hire with a short, data-backed shortlist. The most cited problem: AI solves old coding tests in seconds.

  • Assessments that match real AI work
  • Hard to tell AI fluency from buzzwords
  • In their words: “AI-native”

Market

Competitive landscape

4 competitors compared: HackerRank, CodeSignal, Codility and TestGorilla.

  • Gap to exploit: Assessments graded on how candidates use AI
  • HackerRank: Built around attempt-based plans and classic coding tests
  • CodeSignal: Broad platform, less AI-engineering focus

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 firstNew accounts get 100,000 free enrichment credits with code BETA100K (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)
  • ProgrammableMediaAn AI note taker, interpreter and webinar host for your calls (opens programmable.media 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