Firecrawl growth playbook

When an account passes 800 credits in a month, show an in-app banner and send one email: "At your pace you'll use about 4,200 credits this month. Hobby covers 5,000 for $16." Use their own endpoint mix, including JSON extraction at 5 credits a page, so the recommendation is right.

The Growth Engineer is measured on conversion. A recommendation built from the user's own usage removes the work of reading the credit table, and the JSON add-on is the most common reason a free user runs out sooner than expected.

Subject: You've used 812 of your 1,000 free credits

Hi {first name},

At your pace this month you'll use about {projected} credits, mostly {top endpoint}.

{Plan} covers {plan credits} a month for ${price}, and you can cap pay-as-you-go so nothing surprises you.

Upgrade in one click: {link}

The Firecrawl team

• Projections are wrong for bursty usage

  Use the last 7 days rather than the month to date, and say "about".

• HiringGrowth Engineer, Product Growth: "Own the growth surfaces end to end: signup, activation, onboarding, and conversion"; "Treat conversion and activation as your number"; experience with playgrounds, docs surfaces and upgrade flows is a bonus.Source
• Your websitePricing: Free 1,000 credits a month (no card), Hobby $16/month yearly, Standard $83, Growth $333, Scale $599 (yearly), Enterprise custom. Credits roll over only on Scale (one month) and Enterprise. JSON extraction adds 4 credits a page; an error page such as a 403 costs 1 credit. The page shows no customer logos, only testimonials from X users.Source
• Your websiteHomepage: "Give your AI agents web data and beyond"; claims 150,000+ companies, 1.25M+ developers, 5 billion+ requests, 400,000+ MCP servers installed and 2.5M+ weekly SDK downloads; logos include Shopify, Lovable, Canva, Zapier, Apple, Replit, DoorDash, Gamma and Sierra.Source

Update the existing comparison page and the "Tavily alternatives" post: what the acquisition means (facts only), a table mapping Tavily search parameters to Firecrawl Search, before-and-after code in Python and Node, and the current Alexandria numbers.

The Head of Growth Marketing owns SEO/GEO and weekly signups. The page already ranks for "Firecrawl vs Tavily"; an ownership change is when switching feels cheapest, and a parameter map removes the real cost of switching.

• Reads as opportunistic

  Keep it factual and useful; the parameter map is the value.

• programmable.infoThe site already has 21 "Firecrawl vs" pages (Tavily, Exa, Parallel and others), about 280 Alexandria site pages with "Try in Agent", 232 glossary pages, 10 free tools and a /benchmarks section. The Firecrawl vs Tavily page says "90+ data providers" and does not mention the Nebius acquisition.Source
• Competitor researchAn independent 2026 benchmark found Brave, Firecrawl, Exa and Parallel within noise of each other on agentic search quality, with latency varying 20×. Parallel published a "Firecrawl alternatives" page on October 11, 2026. Nebius completed its acquisition of Tavily in 2026.See the research Source
• HiringHead of Growth Marketing: own weekly signup pacing across SEO/GEO, partnerships, brand and paid; "Take over our paid ads program and the agency running it: audit it in your first month and decide to scale, fix, or end it"; inherit product marketing, SEO, paid and YouTube. Eric Ciarla, the Chief Growth Officer, owns narrative, launches and open source.Source
• Your websiteAlexandria page: "The knowledge library for superintelligence"; 278 providers, 1,672 capabilities, 113M+ indexed sources; featured providers include Shopify, Benzinga, Apollo, Fiscal.ai and FullEnrich. Providers join a waitlist "to get paid"; no payout terms are published. The homepage still says 93 providers and 640 capabilities.Source

Publish a live board of the 50 sources agents request most that Alexandria cannot yet serve (by category, with request volume bands), and a "Claim this source" button that drops the owner straight into the provider waitlist with the demand number attached.

A rights owner who sees that agents asked for their data 40,000 times last month has a reason to reply. It turns a cold outbound list into inbound, and the board doubles as content about where agent demand is going.

Subject: AI agents asked for {source} 40,000+ times last month

Hi {first name},

Firecrawl serves web data to AI agents for 150,000+ companies. {Source} is one of the most requested sources our agents can't yet access properly: it's #{rank} on our public demand board ({link}).

Alexandria pays providers for attributed access when agents use their data. If you own the rights, a 20-minute call would tell you what that demand is worth.

{name}
Partnerships, Firecrawl Alexandria

• Publishing demand helps competitors prioritise

  Show bands and categories, not exact counts, and only the top 50.

• HiringPartnerships Manager, Alexandria: triage the provider waitlist, go outbound to "sources ranked by agent demand", negotiate rights and redistribution, and report catalog gaps; measured on "which sources come online, how fast, and on what terms". Providers get "paid, attributed access" to agent demand.Source
• Your websiteAlexandria page: "The knowledge library for superintelligence"; 278 providers, 1,672 capabilities, 113M+ indexed sources; featured providers include Shopify, Benzinga, Apollo, Fiscal.ai and FullEnrich. Providers join a waitlist "to get paid"; no payout terms are published. The homepage still says 93 providers and 640 capabilities.Source
• Search resultsSeries B announcement: Alexandria combines paid official data providers, a Research Index, a Developer Index, a Government Index and site connectors; Firecrawl already pays providers such as Wikimedia Enterprise and plans a self-service system for contributors to earn when agents use their knowledge.Source

Serve every public number (providers, capabilities, indexed sources, stars, requests served, weekly SDK downloads) from one stats endpoint updated daily, and render it everywhere, including launch assets and job posts.

The Member of Design Staff ships launch assets constantly; each one copies numbers that are already out of date. A single source keeps every surface current and turns growing numbers into a live proof point.

• A number drops and shows in public

  Show monotonic totals (requests served) and review any falling count before it renders.

• Your websiteAlexandria page: "The knowledge library for superintelligence"; 278 providers, 1,672 capabilities, 113M+ indexed sources; featured providers include Shopify, Benzinga, Apollo, Fiscal.ai and FullEnrich. Providers join a waitlist "to get paid"; no payout terms are published. The homepage still says 93 providers and 640 capabilities.Source
• programmable.infoThe site already has 21 "Firecrawl vs" pages (Tavily, Exa, Parallel and others), about 280 Alexandria site pages with "Try in Agent", 232 glossary pages, 10 free tools and a /benchmarks section. The Firecrawl vs Tavily page says "90+ data providers" and does not mention the Nebius acquisition.Source
• Your websiteHomepage: "Give your AI agents web data and beyond"; claims 150,000+ companies, 1.25M+ developers, 5 billion+ requests, 400,000+ MCP servers installed and 2.5M+ weekly SDK downloads; logos include Shopify, Lovable, Canva, Zapier, Apple, Replit, DoorDash, Gamma and Sierra.Source
• HiringMember of Design Staff: launch videos and social assets "for a constant launch cadence", concepting every launch with the Creative Lead; the work is "the first thing millions of developers see from us".Source

Move the strongest logos to the pricing page and replace the tweet wall with three short outcomes from named companies (for example AgentOps moving off Apify for a 50× speed-up), placed next to Growth, Scale and Enterprise.

Conversion is the Growth Engineer's number. On a usage-based pricing page the question is "do companies like mine trust this at volume", which tweets from individual builders do not answer.

• Customers decline to give numbers

  Use relative outcomes ("50× faster") or descriptors ("a public e-commerce platform").

• Your websitePricing: Free 1,000 credits a month (no card), Hobby $16/month yearly, Standard $83, Growth $333, Scale $599 (yearly), Enterprise custom. Credits roll over only on Scale (one month) and Enterprise. JSON extraction adds 4 credits a page; an error page such as a 403 costs 1 credit. The page shows no customer logos, only testimonials from X users.Source
• Your websiteHomepage: "Give your AI agents web data and beyond"; claims 150,000+ companies, 1.25M+ developers, 5 billion+ requests, 400,000+ MCP servers installed and 2.5M+ weekly SDK downloads; logos include Shopify, Lovable, Canva, Zapier, Apple, Replit, DoorDash, Gamma and Sierra.Source
• Your websiteTestimonial: Alex Reibman (AgentOps) "Moved our internal agent's web scraping tool from Apify to Firecrawl because it benchmarked 50x faster"; Steven Tey: "Open-source Clay alternative just dropped".Source
• HiringGrowth Engineer, Product Growth: "Own the growth surfaces end to end: signup, activation, onboarding, and conversion"; "Treat conversion and activation as your number"; experience with playgrounds, docs surfaces and upgrade flows is a bonus.Source
• programmable.infoIn the past year the starting price rose from $9 to $16 a month and the headline changed from "Turn websites into LLM-ready data" to "Give your AI agents web data and beyond". npm downloads are about 957k a week across 23 packages; YouTube has 61.7k subscribers and 111 videos.Source

Show a per-job breakdown in the dashboard and the API response: pages succeeded, pages that returned error status (and that they cost 1 credit), pages with no result (free), add-on credits, and a link to the runbook for each failure type.

The Support Engineer is measured on leverage rather than tickets. A receipt answers the most common billing question before it becomes a ticket, and fewer surprise charges protects net revenue retention.

• Showing error charges draws complaints

  The charge exists either way; explaining it lowers disputes.

• HiringSupport Engineer: first-line technical support for "high-value and enterprise accounts": failed scrapes, blocked sites and rate limits; keep the AI support agents effective and write runbooks and docs fixes; measured on "leverage", not tickets closed.Source
• Your websitePricing: Free 1,000 credits a month (no card), Hobby $16/month yearly, Standard $83, Growth $333, Scale $599 (yearly), Enterprise custom. Credits roll over only on Scale (one month) and Enterprise. JSON extraction adds 4 credits a page; an error page such as a 403 costs 1 credit. The page shows no customer logos, only testimonials from X users.Source
• HiringGrowth Engineer, Product Growth: "Own the growth surfaces end to end: signup, activation, onboarding, and conversion"; "Treat conversion and activation as your number"; experience with playgrounds, docs surfaces and upgrade flows is a bonus.Source

Rebuild paid reporting on cost per paid account and 90-day revenue, split by campaign. Then test three changes: stop bidding on "firecrawl" where organic already captures it, move budget to high-intent competitor and use-case terms ("tavily alternative", "scrape website to markdown"), and test sponsorships on developer YouTube channels and newsletters against search.

Firecrawl clearly values paid enough to keep an agency and make its review a first-month goal. A free tier this generous makes signup-based reporting misleading, and the audit decides the budget for the year.

[ ] 90 days of spend by campaign, joined to accounts that paid
[ ] Cost per signup vs cost per paid account, by campaign
[ ] Brand-term holdout in one region
[ ] Search terms report: which queries bought free accounts only
[ ] Landing page per campaign: does it match the query?
[ ] Agency fee as a share of spend
[ ] Recommendation: scale, fix or end, with the budget for next quarter

• Pausing brand terms lets competitors bid on "firecrawl"

  Hold out one region only and watch competitor impression share.

• HiringHead of Growth Marketing: own weekly signup pacing across SEO/GEO, partnerships, brand and paid; "Take over our paid ads program and the agency running it: audit it in your first month and decide to scale, fix, or end it"; inherit product marketing, SEO, paid and YouTube. Eric Ciarla, the Chief Growth Officer, owns narrative, launches and open source.Source
• Your websitePricing: Free 1,000 credits a month (no card), Hobby $16/month yearly, Standard $83, Growth $333, Scale $599 (yearly), Enterprise custom. Credits roll over only on Scale (one month) and Enterprise. JSON extraction adds 4 credits a page; an error page such as a 403 costs 1 credit. The page shows no customer logos, only testimonials from X users.Source
• programmable.infoIn the past year the starting price rose from $9 to $16 a month and the headline changed from "Turn websites into LLM-ready data" to "Give your AI agents web data and beyond". npm downloads are about 957k a week across 23 packages; YouTube has 61.7k subscribers and 111 videos.Source
• Competitor researchAn independent 2026 benchmark found Brave, Firecrawl, Exa and Parallel within noise of each other on agentic search quality, with latency varying 20×. Parallel published a "Firecrawl alternatives" page on October 11, 2026. Nebius completed its acquisition of Tavily in 2026.See the research Source

Publish success rate, P95 latency and token savings every month by category (e-commerce, news, docs, social, government), with the methodology, what regressed and what was fixed, and an open 1,000-URL test set others can run.

An independent benchmark already found the top agent search APIs within noise of each other, so reliability over time is the difference a buyer can verify. A monthly report gives sales, the CMO and support one number to point to.

• A bad month is public

  That honesty is the point; publish the fix with the regression.

• HiringSearch Engineer: reverse engineer how sites detect automated traffic, build diagnostics and monitoring, and "Own scrape success rate: Measure it, explain regressions, and ship the fixes."Source
• Your websiteHomepage benchmark: 96% coverage on a 1,000-URL benchmark against Puppeteer 79% and cURL 75%, a P95 latency of 3,387 ms and 93% fewer input tokens.Source
• HiringResearch Engineer, Benchmarks: weekly public benchmarks of data providers (for example Apollo vs FullEnrich vs DataLegion, Built In vs ZipRecruiter), public leaderboard pages with marketing, and results that change which provider Alexandria calls. "Our numbers have to survive the vendors who lose." The company enrichment benchmark (October 2) ranked Apollo.io first of six providers.Source
• Competitor researchAn independent 2026 benchmark found Brave, Firecrawl, Exa and Parallel within noise of each other on agentic search quality, with latency varying 20×. Parallel published a "Firecrawl alternatives" page on October 11, 2026. Nebius completed its acquisition of Tavily in 2026.See the research Source

Offer one month of rollover on yearly Standard and Growth plans, and make it the headline of the yearly toggle. Model the breakage first so the CFO knows the cost.

The CFO is hired to set credit, prepaid and pay-as-you-go policy; the Growth Engineer owns conversion. Rollover makes yearly plans safer to buy and removes a common downgrade trigger.

• Revenue recognition gets harder

  Cap rollover at one month and agree treatment with the auditor first.

• Your websitePricing: Free 1,000 credits a month (no card), Hobby $16/month yearly, Standard $83, Growth $333, Scale $599 (yearly), Enterprise custom. Credits roll over only on Scale (one month) and Enterprise. JSON extraction adds 4 credits a page; an error page such as a 403 costs 1 credit. The page shows no customer logos, only testimonials from X users.Source
• HiringChief Financial Officer: set policy for credits, prepaid and pay-as-you-go plans, enterprise contracts and discount authority; deploy the $75M Series B and prepare the next raise; treat tokens as a P&L line. Hiring "at a pace that will nearly double the team".Source
• HiringGrowth Engineer, Product Growth: "Own the growth surfaces end to end: signup, activation, onboarding, and conversion"; "Treat conversion and activation as your number"; experience with playgrounds, docs surfaces and upgrade flows is a bonus.Source

Score self-serve accounts daily on company domain, number of API keys, week-over-week credit growth and concurrency hits; send the top tier to a named contact with an offer of an Enterprise trial (SLA, zero data retention, SSO).

Concurrency limits (25 on Standard, 50 on Growth) are a technical ceiling a growing team hits before it thinks about Enterprise. Reaching them first turns a frustrating limit into a sales conversation.

• Developers dislike sales outreach

  Lead with the limit they hit and write as an engineer, which is what the BDR role asks for.

• HiringDeployment team of six: a Technical BDR to "open Firecrawl's outbound motion", reporting to the VP of Deployment, measured on "qualified pipeline created, not emails sent" and told to "reference their stack"; Forward Deployed Engineers measured "in adoption and expansion"; plus Sales, Success and Implementation roles.Source
• Your websitePricing: Free 1,000 credits a month (no card), Hobby $16/month yearly, Standard $83, Growth $333, Scale $599 (yearly), Enterprise custom. Credits roll over only on Scale (one month) and Enterprise. JSON extraction adds 4 credits a page; an error page such as a 403 costs 1 credit. The page shows no customer logos, only testimonials from X users.Source
• HiringSupport Engineer: first-line technical support for "high-value and enterprise accounts": failed scrapes, blocked sites and rate limits; keep the AI support agents effective and write runbooks and docs fixes; measured on "leverage", not tickets closed.Source
• Your websiteHomepage: "Give your AI agents web data and beyond"; claims 150,000+ companies, 1.25M+ developers, 5 billion+ requests, 400,000+ MCP servers installed and 2.5M+ weekly SDK downloads; logos include Shopify, Lovable, Canva, Zapier, Apple, Replit, DoorDash, Gamma and Sierra.Source

With each benchmark release, send winners an embeddable, dated badge linking to the leaderboard, and send every other vendor a private breakdown of where they lost, with an invitation to list in Alexandria or improve their result through it.

Vendors who win share the result with their own audience, which links back to Firecrawl. Vendors who lose are the providers the Partnerships Manager needs, and the breakdown gives that conversation a concrete start.

• Losing vendors dispute the method

  Publish the method and data, and accept re-runs with a fixed process.

• HiringResearch Engineer, Benchmarks: weekly public benchmarks of data providers (for example Apollo vs FullEnrich vs DataLegion, Built In vs ZipRecruiter), public leaderboard pages with marketing, and results that change which provider Alexandria calls. "Our numbers have to survive the vendors who lose." The company enrichment benchmark (October 2) ranked Apollo.io first of six providers.Source
• HiringPartnerships Manager, Alexandria: triage the provider waitlist, go outbound to "sources ranked by agent demand", negotiate rights and redistribution, and report catalog gaps; measured on "which sources come online, how fast, and on what terms". Providers get "paid, attributed access" to agent demand.Source
• Your websiteAlexandria page: "The knowledge library for superintelligence"; 278 providers, 1,672 capabilities, 113M+ indexed sources; featured providers include Shopify, Benzinga, Apollo, Fiscal.ai and FullEnrich. Providers join a waitlist "to get paid"; no payout terms are published. The homepage still says 93 providers and 640 capabilities.Source

Hold launches for five consecutive days twice a year: one product a day, one launch video, a live stream, a daily giveaway of credits, and a countdown page developers can subscribe to.

Batching launches gives the Creative Lead and design staff one story to tell instead of twenty, and a week of daily news gets coverage that single launches do not.

• Holding launches slows shipping

  Ship to beta users as usual and hold only the public announcement.

• HiringMember of Design Staff: launch videos and social assets "for a constant launch cadence", concepting every launch with the Creative Lead; the work is "the first thing millions of developers see from us".Source
• HiringGrowth team roles also include a Social Video Producer (fast, short-form social video, "a founder with a strong take") and an Open Source Engineer.Source
• programmable.infoIn the past year the starting price rose from $9 to $16 a month and the headline changed from "Turn websites into LLM-ready data" to "Give your AI agents web data and beyond". npm downloads are about 957k a week across 23 packages; YouTube has 61.7k subscribers and 111 videos.Source
• Your websiteBlog: more than 20 posts between October 1 and 13, mostly "best X" roundups and tutorials (plugins, APIs, connectors), with launches such as Universal Scrape (October 8) and the official ChatGPT plugin (October 6); benchmarks and customer stories are rare.Source

Tag accounts by origin (agent skill, MCP, website). For agent-created accounts, send the human a short email after real usage with what their agent did, their credit pace and one link to claim the dashboard and set a spend cap.

Firecrawl hired an Agent Experience Engineer because agents are now a channel, but paid conversion still depends on a human deciding to pay. Telling the owner what their agent has already done is a concrete reason to look, and a spend cap removes the fear that keeps them on Free.

Subject: Your agent has used Firecrawl 50 times

Hi,

An AI agent signed you up for Firecrawl on {date}. Since then it has scraped {pages} pages and run {searches} searches, mostly from {top domains}.

At this pace you'll use about {projected} credits this month (you have 1,000 free).

See everything it did and set a spend cap: {link}

The Firecrawl team

• Owners did not expect an email

  Send one, from the product, with a clear reason and an unsubscribe.

• Search resultsGitHub: about 192.5k stars and 10.2k forks; AGPL-3.0 with MIT SDKs; the README says the cloud "includes additional features", but the open-source vs cloud comparison is only an image. An agent-onboarding skill lets an AI agent sign a user up and get an API key.Source
• HiringAgent Experience Engineer: A/B tests across agent-facing surfaces (MCPs, CLIs, skills, APIs, documentation); form hypotheses "about how agents discover, interpret, and select tools", for "an audience that isn't human".Source
• Your websiteHomepage: "Give your AI agents web data and beyond"; claims 150,000+ companies, 1.25M+ developers, 5 billion+ requests, 400,000+ MCP servers installed and 2.5M+ weekly SDK downloads; logos include Shopify, Lovable, Canva, Zapier, Apple, Replit, DoorDash, Gamma and Sierra.Source
• HiringGrowth Engineer, Product Growth: "Own the growth surfaces end to end: signup, activation, onboarding, and conversion"; "Treat conversion and activation as your number"; experience with playgrounds, docs surfaces and upgrade flows is a bonus.Source

Publish one case study with a live provider (requests served, attribution, money earned in month one, with their approval), and a calculator on the provider page that estimates earnings from a source's demand band.

The Partnerships Manager is measured on which sources come online and how fast. Rights owners sign faster when they can see what peers earned; silence on payouts reads as risk.

• Published earnings set expectations too high

  Show ranges and state they depend on agent demand.

• #3 A public Alexandria "most requested sources" board so providers come to you

• Your websiteAlexandria page: "The knowledge library for superintelligence"; 278 providers, 1,672 capabilities, 113M+ indexed sources; featured providers include Shopify, Benzinga, Apollo, Fiscal.ai and FullEnrich. Providers join a waitlist "to get paid"; no payout terms are published. The homepage still says 93 providers and 640 capabilities.Source
• Search resultsSeries B announcement: Alexandria combines paid official data providers, a Research Index, a Developer Index, a Government Index and site connectors; Firecrawl already pays providers such as Wikimedia Enterprise and plans a self-service system for contributors to earn when agents use their knowledge.Source
• HiringPartnerships Manager, Alexandria: triage the provider waitlist, go outbound to "sources ranked by agent demand", negotiate rights and redistribution, and report catalog gaps; measured on "which sources come online, how fast, and on what terms". Providers get "paid, attributed access" to agent demand.Source
• HiringChief Financial Officer: set policy for credits, prepaid and pay-as-you-go plans, enterprise contracts and discount authority; deploy the $75M Series B and prepare the next raise; treat tokens as a P&L line. Hiring "at a pace that will nearly double the team".Source

Co-build five one-click templates with Lovable and Replit (a lead list builder, a price monitor, a docs chatbot, a news digest, a competitor tracker) that use Firecrawl with the user's own key, listed in each builder's template gallery.

Partnerships is one of the channels the Head of Growth Marketing owns. These customers are distribution channels with millions of builders. A template that works on first run beats any number of "best tools" posts, and each run creates an account.

• Partners prefer their own scraping

  Start with the partner whose users already use Firecrawl the most.

• Your websiteHomepage: "Give your AI agents web data and beyond"; claims 150,000+ companies, 1.25M+ developers, 5 billion+ requests, 400,000+ MCP servers installed and 2.5M+ weekly SDK downloads; logos include Shopify, Lovable, Canva, Zapier, Apple, Replit, DoorDash, Gamma and Sierra.Source
• HiringHead of Growth Marketing: own weekly signup pacing across SEO/GEO, partnerships, brand and paid; "Take over our paid ads program and the agency running it: audit it in your first month and decide to scale, fix, or end it"; inherit product marketing, SEO, paid and YouTube. Eric Ciarla, the Chief Growth Officer, owns narrative, launches and open source.Source
• Your websiteTestimonial: Alex Reibman (AgentOps) "Moved our internal agent's web scraping tool from Apify to Firecrawl because it benchmarked 50x faster"; Steven Tey: "Open-source Clay alternative just dropped".Source

Publish a diligence pack next to the trust center: the crawling and robots.txt policy, publisher opt-out process, data provenance for Alexandria sources, zero data retention terms, a standard DPA and pre-approved answers to the 30 most common security and legal questionnaire items.

The Legal Engineer is asked for playbooks sales can use without a lawyer, and the Deployment team is growing. A pack that answers the web-data questions up front shortens every enterprise cycle and frees counsel for the novel deals.

• Public positions become a target

  Publish the policy and keep detailed evidence under NDA.

• HiringLegal team of five: the General Counsel sets "what we crawl and how, how we treat robots.txt and publisher opt-outs, what data provenance we stand behind" and defends it "in customer diligence"; the Legal Engineer builds "playbooks the sales team can use without waiting on a lawyer" and proves systems match the terms "in a customer security review"; Commercial Counsel negotiates data-licensing deals.Source
• HiringDeployment team of six: a Technical BDR to "open Firecrawl's outbound motion", reporting to the VP of Deployment, measured on "qualified pipeline created, not emails sent" and told to "reference their stack"; Forward Deployed Engineers measured "in adoption and expansion"; plus Sales, Success and Implementation roles.Source
• Your websitePricing: Free 1,000 credits a month (no card), Hobby $16/month yearly, Standard $83, Growth $333, Scale $599 (yearly), Enterprise custom. Credits roll over only on Scale (one month) and Enterprise. JSON extraction adds 4 credits a page; an error page such as a 403 costs 1 credit. The page shows no customer logos, only testimonials from X users.Source

Let any rights holder list a source: verify ownership, set access rules, publish a revenue share, see requests and earnings live, and get paid monthly. Keep negotiated deals for large providers.

The long tail of sources agents request is too large to sign one by one. A clear, self-serve offer is what made app stores and ad networks scale, and every new source makes Alexandria harder to copy.

• Low-quality or unlicensed sources

  Rights attestation, quality checks and the right to delist.

• #3 A public Alexandria "most requested sources" board so providers come to you
• #14 A provider earnings case study and a "what your data could earn" calculator

• Search resultsSeries B announcement: Alexandria combines paid official data providers, a Research Index, a Developer Index, a Government Index and site connectors; Firecrawl already pays providers such as Wikimedia Enterprise and plans a self-service system for contributors to earn when agents use their knowledge.Source
• HiringPartnerships Manager, Alexandria: triage the provider waitlist, go outbound to "sources ranked by agent demand", negotiate rights and redistribution, and report catalog gaps; measured on "which sources come online, how fast, and on what terms". Providers get "paid, attributed access" to agent demand.Source
• Your websiteAlexandria page: "The knowledge library for superintelligence"; 278 providers, 1,672 capabilities, 113M+ indexed sources; featured providers include Shopify, Benzinga, Apollo, Fiscal.ai and FullEnrich. Providers join a waitlist "to get paid"; no payout terms are published. The homepage still says 93 providers and 640 capabilities.Source
• HiringChief Financial Officer: set policy for credits, prepaid and pay-as-you-go plans, enterprise contracts and discount authority; deploy the $75M Series B and prepare the next raise; treat tokens as a P&L line. Hiring "at a pace that will nearly double the team".Source

Publish private-offer-ready marketplace listings for Enterprise with the SOC 2 Type II report attached, and give sales a standard marketplace path for deals above a set size.

The CFO owns enterprise contracts and discount authority. Marketplace deals clear procurement faster and draw down budgets that are already approved.

• Fees cut margin

  Price the fee into private offers.

• Your websiteHomepage: "Give your AI agents web data and beyond"; claims 150,000+ companies, 1.25M+ developers, 5 billion+ requests, 400,000+ MCP servers installed and 2.5M+ weekly SDK downloads; logos include Shopify, Lovable, Canva, Zapier, Apple, Replit, DoorDash, Gamma and Sierra.Source
• Your websitePricing: Free 1,000 credits a month (no card), Hobby $16/month yearly, Standard $83, Growth $333, Scale $599 (yearly), Enterprise custom. Credits roll over only on Scale (one month) and Enterprise. JSON extraction adds 4 credits a page; an error page such as a 403 costs 1 credit. The page shows no customer logos, only testimonials from X users.Source
• HiringChief Financial Officer: set policy for credits, prepaid and pay-as-you-go plans, enterprise contracts and discount authority; deploy the $75M Series B and prepare the next raise; treat tokens as a P&L line. Hiring "at a pace that will nearly double the team".Source

Publish a versioned web access standard (what Firecrawl crawls, how it honours robots.txt and opt-outs, how it identifies itself, how Alexandria sources are licensed), a registry where any site can opt out or opt into paid access, and a quarterly independent audit.

Enterprise buyers and publishers both want to know a crawler behaves. A checkable standard answers diligence, gives the policy conversation something concrete, and turns opt-outs into a path into Alexandria's paid program.

• Commitments limit what the product can do

  Version the standard and change it in public with notice.

• #16 A self-serve "web data diligence pack" so enterprise deals don't wait on legal

• HiringLegal team of five: the General Counsel sets "what we crawl and how, how we treat robots.txt and publisher opt-outs, what data provenance we stand behind" and defends it "in customer diligence"; the Legal Engineer builds "playbooks the sales team can use without waiting on a lawyer" and proves systems match the terms "in a customer security review"; Commercial Counsel negotiates data-licensing deals.Source
• Search resultsSeries B announcement: Alexandria combines paid official data providers, a Research Index, a Developer Index, a Government Index and site connectors; Firecrawl already pays providers such as Wikimedia Enterprise and plans a self-service system for contributors to earn when agents use their knowledge.Source
• HiringPartnerships Manager, Alexandria: triage the provider waitlist, go outbound to "sources ranked by agent demand", negotiate rights and redistribution, and report catalog gaps; measured on "which sources come online, how fast, and on what terms". Providers get "paid, attributed access" to agent demand.Source

Offer framework and agent platform maintainers a revenue share on paid usage that starts in their product, plus co-marketing, in return for Firecrawl as the default or first-listed web tool.

2.5M weekly SDK downloads show developers already pick Firecrawl when asked. A share aligns the platforms with promoting it, and attributed usage gives the CFO a channel with a measurable cost.

• Open-source projects resist paid defaults

  Offer the share as sponsorship of the project, disclosed openly.

• Your websiteHomepage: "Give your AI agents web data and beyond"; claims 150,000+ companies, 1.25M+ developers, 5 billion+ requests, 400,000+ MCP servers installed and 2.5M+ weekly SDK downloads; logos include Shopify, Lovable, Canva, Zapier, Apple, Replit, DoorDash, Gamma and Sierra.Source
• Search resultsGitHub: about 192.5k stars and 10.2k forks; AGPL-3.0 with MIT SDKs; the README says the cloud "includes additional features", but the open-source vs cloud comparison is only an image. An agent-onboarding skill lets an AI agent sign a user up and get an API key.Source
• Your websiteBlog: more than 20 posts between October 1 and 13, mostly "best X" roundups and tutorials (plugins, APIs, connectors), with launches such as Universal Scrape (October 8) and the official ChatGPT plugin (October 6); benchmarks and customer stories are rare.Source
• HiringChief Financial Officer: set policy for credits, prepaid and pay-as-you-go plans, enterprise contracts and discount authority; deploy the $75M Series B and prepare the next raise; treat tokens as a P&L line. Hiring "at a pace that will nearly double the team".Source

Publish a monthly post and repo from the AI Enablement Lead: the agent workflows each team runs, the skills and MCP configs, token spend against output, and what was switched off because it did not pay. Every workflow that uses web data uses Firecrawl.

It is honest content no other vendor can write about itself, it continues a story the press already covered, and it shows Firecrawl being used inside real business workflows rather than demos.

• Sharing internal workflows exposes sensitive detail

  Review every post for customer data and secrets.

• HiringAI Enablement Lead: build the internal agent setup, start with the revenue team, review token spend for usage without output; "measured by who's using agents well now who wasn't before".Source
• Search resultsFirecrawl drew coverage in 2025 for job ads seeking to hire AI agents as employees, first at $15k a year and later with $1M for three agents.Source
• HiringChief Financial Officer: set policy for credits, prepaid and pay-as-you-go plans, enterprise contracts and discount authority; deploy the $75M Series B and prepare the next raise; treat tokens as a P&L line. Hiring "at a pace that will nearly double the team".Source

Pitch Nieman Lab, Press Gazette, Digiday and Axios Media Trends with a data-led story: which kinds of publisher content agents request most (from the demand board), what Alexandria pays, and a CEO interview.

The Partnerships Manager needs publishers and research indexes to come online. Coverage in the press publishers read reaches the people who decide on licensing, and reframes Firecrawl for the next raise.

• Reporters focus on Firecrawl's scraping instead

  Be ready with robots, opt-out and payment policies on the page.

• #3 A public Alexandria "most requested sources" board so providers come to you

• Search resultsSeries B announcement: Alexandria combines paid official data providers, a Research Index, a Developer Index, a Government Index and site connectors; Firecrawl already pays providers such as Wikimedia Enterprise and plans a self-service system for contributors to earn when agents use their knowledge.Source
• HiringPartnerships Manager, Alexandria: triage the provider waitlist, go outbound to "sources ranked by agent demand", negotiate rights and redistribution, and report catalog gaps; measured on "which sources come online, how fast, and on what terms". Providers get "paid, attributed access" to agent demand.Source
• HiringLegal team of five: the General Counsel sets "what we crawl and how, how we treat robots.txt and publisher opt-outs, what data provenance we stand behind" and defends it "in customer diligence"; the Legal Engineer builds "playbooks the sales team can use without waiting on a lawyer" and proves systems match the terms "in a customer security review"; Commercial Counsel negotiates data-licensing deals.Source

Add an optional setting to the open-source version: when a self-hosted scrape is blocked, retry it through the cloud with an API key and pay per page. Show self-hosters their blocked-page rate so they can see what they are missing.

It turns the self-hosting community into a paying channel without changing the license, and the open-source vs cloud difference becomes something users experience rather than an image in the README.

• Community sees it as upselling

  Opt-in, off by default, with no change to self-hosted features.

• Search resultsGitHub: about 192.5k stars and 10.2k forks; AGPL-3.0 with MIT SDKs; the README says the cloud "includes additional features", but the open-source vs cloud comparison is only an image. An agent-onboarding skill lets an AI agent sign a user up and get an API key.Source
• HiringSearch Engineer: reverse engineer how sites detect automated traffic, build diagnostics and monitoring, and "Own scrape success rate: Measure it, explain regressions, and ship the fixes."Source
• HiringGrowth team roles also include a Social Video Producer (fast, short-form social video, "a founder with a strong take") and an Open Source Engineer.Source
• HiringJob posts: "We hit 8 figures in ARR in year one and more than doubled it in year two"; $75M Series B led by Smash Capital (September 2026); 192k+ GitHub stars, "top 40 repositories of all time"; 11 to 50 employees in San Francisco with a Toronto hub.Source

Run a fixed set of 200 prompts ("scrape this site into markdown", "find pricing for these companies", "monitor this page") weekly across the major agents and chat assistants, record which tool or library each chooses, and treat that share as a growth metric with experiments on docs, skills and MCP descriptions.

More and more developers let an agent choose the tool. A weekly share by agent and task shows where Firecrawl loses the choice, and the Agent Experience Engineer's A/B tests can then target those surfaces.

• Agent behaviour changes with each model release

  Track by model version and treat drops after releases as signals, not noise.

• HiringAgent Experience Engineer: A/B tests across agent-facing surfaces (MCPs, CLIs, skills, APIs, documentation); form hypotheses "about how agents discover, interpret, and select tools", for "an audience that isn't human".Source
• HiringHead of Growth Marketing: own weekly signup pacing across SEO/GEO, partnerships, brand and paid; "Take over our paid ads program and the agency running it: audit it in your first month and decide to scale, fix, or end it"; inherit product marketing, SEO, paid and YouTube. Eric Ciarla, the Chief Growth Officer, owns narrative, launches and open source.Source
• Your websiteHomepage: "Give your AI agents web data and beyond"; claims 150,000+ companies, 1.25M+ developers, 5 billion+ requests, 400,000+ MCP servers installed and 2.5M+ weekly SDK downloads; logos include Shopify, Lovable, Canva, Zapier, Apple, Replit, DoorDash, Gamma and Sierra.Source
• Search resultsGitHub: about 192.5k stars and 10.2k forks; AGPL-3.0 with MIT SDKs; the README says the cloud "includes additional features", but the open-source vs cloud comparison is only an image. An agent-onboarding skill lets an AI agent sign a user up and get an API key.Source