Hyrax

Hyrax Marketing Budget

FY27 request and supporting model

Date
September 3, 2026
Prepared by
Tristan Benozer
Go-to-Market
Audience
Finance and executive
review
Request
$78,599 per month
$831,594 in year one

Contents

Six sections
IObjective and request1–4
IIAssumptions5–7
IIIInvestment required8–13
IVExpected return14–15
VRelease schedule and controls16
VIAppendix17–21
SECTION I

Objective and request

Objective

1 of 21

Make $10 million in the first year.

This document sets out how much marketing spend is required to do that.

Revenue objective$10,000,000 annualised
TimeframeFirst year
Marketing spend required, per month$78,599
Marketing spend required, year one$831,594

Spend is released in four quarterly steps rather than in full at the outset. Conditions on slide 16.

Request

2 of 21

$78,599 per month for Hyrax marketing, released quarterly against performance. Year one commitment is $831,594.

Monthly at full release$78,599
Year one, released in four quarterly steps$831,594
Current approved monthly spend$45,000
Increase over current, at full release$33,599 per month
Revenue target$10,000,000 annualised
Modelled position at month 12$7.84M annualised, 1,379 customers
Modelled month the target is metMonth 16
Cost to acquire one paying customer$363.33
Payback periodUnder one month

Decision required: approval of the year one commitment and the quarterly release conditions on slide 16.

Current Spend and Result

3 of 21
$45K$34K$23K$11K$0 AD SPEND 200150100500 SIGNUPS $17,588 $9,768 $19,625 $45,000 3937183 JuneJulyAugustSeptember actualactualactualrun rate, $1,500 a day Google Reddit LinkedIn X Signups, right axis

Bars are ad spend, left axis. The line is signups, right axis. Spend from the Google Ads (8518001543), Reddit (a2_iyhg34o7cahk), LinkedIn (537800361) and X (18ce55vw38i) accounts. Signups from PostHog project 436309, account_created and wizard.started. Pulled 2026-09-02. The monthly cap has been raised three times: $15,000, $20,000, $25,000, $45,000. September is the current run rate, not a completed month.

Revenue Target

4 of 21

Basis for the target

  • $10,000,000 annualised is the amount judged achievable in roughly one year for a product at this stage in this category.
  • It is a planning target, not a commitment. The model reaches it in month 16 rather than month 12.
  • Comparable companies in the category are on slide 7.

What the target requires

Monthly revenue$833,333
Paying customers1,389
New paying customers per month150
Free signups per month3,750
Website visitors per month75,000
Ad impressions per month1,875,000

Assumes 20 seats per customer at $30 per seat and 95% monthly retention. Both assumptions are set out on slide 6.

SECTION II

Assumptions

Funnel Model

5 of 21
1,875,000
Ad impressions a month
4.0% CTR
75,000
Website visitors a month
5.0% web-to-signup
3,750
Free signups a month
4.0% upgrade
+150
New paying customers a month

Conversion rates, landing team size and retention are from the Hyrax capital allocation proposal and growth funnel model, 2026-09-03. Settled seat count is revised to 20 on the evidence linked on slide 6.

Model Assumptions

6 of 21
Revenue target$10,000,000 / year
That works out per month at$833,333.33 / month
Price per seat$30 / user / month
Seats when a team starts8
Seats once the team is fully on board20
Average revenue per customer$600 / month (20 × $30)
Paying customers we need1,389 teams
New paying customers a month+150 teams / month
Free signups a month3,750
Average cost per signup$20.96
Website visitors a month75,000 / month
Ad impressions a month1,875,000 / month

Seat counts from published deployments at comparable company sizes: LSports 150 of 150 engineers (Codacy), trivago 133 weekly users of 668 staff (CodeRabbit), Pomelo 200 of 200 (Snyk), SalesRabbit 20 of 20 (CodeRabbit). Adoption rates once deployed run 70 to 100%: Brex 100% of code (Greptile), Brex 70%+ and Rippling 60% (Cursor), Qodo 46% monthly actives. Compiled in Track 7 research, Notion, 2026-09-03. Twenty seats is the low end of the observed range.

Comparable Companies

7 of 21
CompanyRevenue pathTimeWhat drove it
Same category, AI code review
CodeRabbit$5M (Apr 2025) → $15M (Sep 2025) → $40M (Apr 2026) → approaching $100M (Aug 2026)16 monthsBottom-up GitHub Marketplace adoption, then a sales team. 17,000 customers, 2M reviews a week. $143M Series C at $1.5B.
Adjacent, AI code generation
Cursor (Anysphere)$100M (Jan 2025) → $500M (Jun 2025) → $1B (Nov 2025) → $4B (Jun 2026)17 monthsIndividual developers first, then enterprise. Enterprise is now 65% of revenue and 60%+ of the Fortune 500 are customers.
Replit$10M (Q4 2024) → $70M (Q1 2025) → $100M (Jun 2025) → $250M (Oct 2025)10 monthsAgent launch in Sep 2024 was the inflection after six years at under $3M. Moved from developers to non-developers.
Lovable$7M (Dec 2024) → $70M (Jun 2025)6 monthsPivot from a command-line tool to a visual product. 30,000 paying customers at $20 a month.
Windsurf (Codeium)$40M ARREnterprise and on-premises focus. Acquired by OpenAI for $3B in May 2025.

Read across

CodeRabbit is the closest comparable by product and category. It went from $5M to $40M in twelve months, and did so on bottom-up developer adoption before adding sales headcount.

Where the comparison breaks

All five had a product in market with demonstrated retention before these curves began. Hyrax is at the start of that. The comparison supports the target being achievable in the category, not the timing.

Sources, collected 2026-09-03: CodeRabbit Sacra and ARR Club. Cursor ValueAdd VC and Andrew Chan. Replit ARR Club and Scaled Enablement. Lovable and Windsurf SaaStr. Figures are third-party estimates and company statements, not audited.

SECTION III

Investment required

Budget by Category

8 of 21
Paid Media  ·  $54,500  ·  69.34% PR & Events Prof. Services  ·  $15,200
Paid Media — $54,500 — 69.34% PR & Events — $7,500 — 9.54% Professional Services — $15,200 — 19.34% Software — $1,399 — 1.78%
CategoryMonthly cost% of total
Paid Media$54,50069.34%
PR & Events$7,5009.54%
Professional Services$15,20019.34%
Software$1,3991.78%
Total per month$78,599100.00%

Every platform, contractor and software licence, listed separately.

Paid Media — $54,500

9 of 21
Line itemMonthlyDaily% of total
Google Ads (SEM & Demand Gen)$20,000$65825.45%
Paid Developer Influencer Fees (1 video spotlight)$10,000$32912.72%
Product Hunt Sponsored Ads (fixed rolling slot)$5,000$1646.36%
LinkedIn Ads (conversion targeting)$4,000$1325.09%
X Ads (developer timeline)$4,000$1325.09%
Meta Ads (founder video retargeting)$3,500$1154.45%
Reddit Ads (developer subreddits)$3,000$993.82%
ChatGPT / AI search network placements$2,500$823.18%
TikTok Ads (short-form code demos)$1,500$491.91%
Bing SEM (enterprise long-tail intent)$1,000$331.27%
Paid Media total$54,500$1,79369.34%

Google and X are set at current account spend. Google $658 a day, measured 31 August to 2 September in account 8518001543. X $4,000, its August total in account 18ce55vw38i. Google cost per signup over the 14 days to 2 September was $21.49, calculated against account_created in PostHog project 436309.

PR & Events — $7,500

10 of 21
Line itemMonthly% of total
PR Agency Retainer & Tech Press Outreach$4,5005.73%
Developer Conferences & Hackathon Tiers$2,0002.54%
Product Hunt Launch Kit & Creative Design$5000.64%
Developer Swag & Onboarding Welcome Kits$5000.64%
PR & Events total$7,5009.54%

The launch kit pairs with the $5,000 Product Hunt sponsored slot carried in Paid Media.

Professional Services — $15,200

11 of 21
Line itemMonthly% of total
Video Editing Services (4 founder videos @ $2k each)$8,00010.18%
Organic Social Content Management (X / LinkedIn)$5,0006.36%
Gauge Media (ad creative validation)$1,2001.53%
Video & Motion Graphics Production (UI overlays)$1,0001.27%
Professional Services total$15,20019.34%

No full-time hires. SEO documentation is written in-house, so it costs nothing here.

Software Stack — $1,399

12 of 21
ToolMonthly
PostHog (product data and funnels)$450
Synter Media (ad buying)$300
Perplexity Sonar / Agent API usage$150
Vercel (website hosting)$100
Parallel (web data)$100
Exa AI (developer link mapping)$100
LinkedIn automation & management$100
Jina AI (clean text for crawlers)$50
Loops Email (automated emails)$49
Software total$1,399

How the main three connect

Synter Media — $300Runs and scales
the ad budget
PostHog — $450Tracks clicks through
to a connected repo
Loops Email — $49Emails people who
never connect a repo

Function notes: Perplexity Sonar checks how the category is described online, Parallel pulls raw web data, Exa maps developer links, Jina turns pages into clean text for crawlers, Vercel hosts the site, the LinkedIn tool handles cold outreach. Published pricing: PostHog, Vercel, Loops, Perplexity, Exa, Jina, Parallel.

Software Overlap and Savings

13 of 21
ToolsTodayOverlapSavings
Perplexity Sonar
Parallel
Exa AI
Jina AI
$150
$100
$100
$50
Four tools that all read the web. Sonar searches and summarises, Parallel searches and fetches, Exa finds related pages, Jina cleans them up. Two of them cover it.$250
Gauge Media
Professional Services
$1,200Synter already tests ad creative and PostHog already has the funnel data. Our Gauge access ran out on 30 August.$1,200
LinkedIn automation
vs Synter Media
$100
$300
Synter buys LinkedIn ads. The other tool sends cold outreach. Different jobs.$0
PostHog
vs Synter Media
$450
$300
PostHog has the product data, Synter has the ad spend. The reports look similar but the data is different.$0
Total savings$17,400 per year$1,450

Vercel and Loops do not overlap with anything else. Pricing per the vendor pages linked on the previous slide. Gauge access status per the GTM log of 2026-08-30.

SECTION IV

Expected return

Modelled Year One

14 of 21
End ofCustomersAvg seats eachTotal seatsRevenue a monthRevenue a yearMonths to make it back
Quarter 14289.94,250$127,485$1.53M0.85
Quarter 279512.710,094$302,820$3.63M0.79
Quarter 31,10914.816,385$491,537$5.90M0.82
Quarter 41,37915.821,778$653,338$7.84M0.77
Month 161,68016.527,800$834,000$10.01M0.77

Seats per customer

A customer starts at 8 seats and reaches 20 over six months. The average across the base rises from 9.9 to 15.8 in year one, held down by continued new customer intake at 8 seats.

Implied company size

Twenty seats at the 70 to 100% adoption rate in published deployments implies a 20 to 30 engineer company. Target range is 15 to 250 engineers.

Target date

Month 16. Year one closes at $7.84M because a large share of the customer base is recent and has not reached full seat count.

150 new customers a month, 95% retained monthly, seats growing 8 to 20 across six months. Months to make it back is that quarter's ad spend per new customer divided by monthly revenue per customer. Retention and conversion assumptions are from the capital allocation proposal; seat progression is from the Track 7 research linked on slide 6.

Unit Economics

15 of 21
$363.33to win one customer
$12,000what a customer is worth
33.0:1back for every dollar spent
0.61months to make it back

Acquisition cost

$54,500 of monthly ad spend divided by 150 new customers.

Lifetime value

Customers stay 20 months on average and pay $600 a month.

Basis for the ratio

33% of signups are modelled as arriving at zero media cost through organic search, AI answer engines and referral.

We make back what we spent on a customer inside their first month of billing. The automatic shut-off is on the next slide.

SECTION V

Release schedule and controls

Release Schedule and Controls

16 of 21

The full year is approved at the outset. Funds are released one quarter at a time, conditional on the prior quarter meeting its stated measure.

QuarterPer monthQuarter totalUnlocks when
Q1$55,000$165,000Released on approval.
Q2$65,000$195,000Q1 cost per signup at or below $20.96.
Q3$78,599$235,797Free-to-paid conversion at 4%, measured on actual customers.
Q4$78,599$235,797Average customer at 20 seats by month nine.
Year one$831,594$111,594 below a flat release of $78,599 for twelve months.

If a condition is not met

  • Spend holds at the current quarter's level. No funds are recovered or cancelled.
  • The condition is reassessed at the following quarterly review.

Within-quarter controls

  • Any channel exceeding its cost-per-signup target is paused automatically.
  • The same applies to any channel producing accounts that do not connect a repository within 72 hours.
  • Paused budget is reallocated to channels meeting target. Monitored continuously through Synter Media.

Ad spend is 69.3% of each quarter. The rest covers PR and events, contractors and software.

SECTION VI

Appendix

Appendix A — Target User

17 of 21

User profile

Who they areSoftware engineer or senior software engineer
Daily workflowWrites, reviews and revises code through GitHub pull requests
Tools in useCursor, Copilot, Devin, Claude Code, Codex, Augment
MotivationAI speed without submitting code that feels generic, inconsistent or architecturally weak
Current painReview churn, missing context, repeated corrections, no certainty the generated code fits
Adoption behaviourWill try it alone or with a small team if setup is light and value shows up in real work
Activation eventHyrax catches a codebase-specific issue before a human reviewer does

Qualifying characteristics

  • Uses AI tools for a real share of implementation work
  • Reviews AI-assisted pull requests from teammates
  • Joining a mature or unfamiliar codebase
  • Works across several services, repos or product surfaces
  • Dislikes submitting code that merely passes tests
  • Gets asked to revise for conventions nobody wrote down
  • Open-source maintainer or heavy GitHub contributor

Source: Hyrax Ideal Customer Profile, Notion, last updated 2026-09-01.

Appendix B — Target Company

18 of 21

Company profile

TypeSoftware-led or codebase-centric business
Engineering size15 to 250+ engineers. Strongest expansion above 50
WorkflowGitHub is the source of truth, pull requests are central
AI adoptionEngineers are allowed and encouraged to use AI coding tools
CodebaseMature enough to have its own patterns, history and boundaries
InfrastructureAWS, GCP, Vercel, Kubernetes, modern CI/CD, microservices
CultureQuality, learning and craft matter, not only output volume
GeographyNorth America, the UK, major European tech hubs

Verticals

  • B2B SaaS with complex product surfaces
  • DevOps, infrastructure and developer tooling
  • Fintech and compliance-sensitive software
  • AI infrastructure
  • Engineering-led businesses where reliability and architecture are strategic

Skip these

  • Not GitHub-centric, or no real review workflow
  • AI-generated code banned
  • Codebase too simple to need product-specific context
  • Only wants style and syntax checks
  • An engineer cannot try it without executive approval

A 15 to 50 engineer company qualifies if the codebase is complex, AI usage is high, and engineers can adopt tools without procurement. Source: Hyrax Ideal Customer Profile, Notion, 2026-09-01.

Appendix C — Purchase Triggers and Buying Committee

19 of 21

Purchase triggers

  • An AI-generated pull request gets heavily reworked
  • They are working somewhere unfamiliar in the codebase
  • AI output is fast but keeps missing the house style
  • Review is the bottleneck and PRs sit waiting
  • They want to use AI more without dropping their standards
  • A teammate shares a Hyrax finding that actually mattered

Buying committee

Software engineer
first user
Wants a codebase-aware review before submitting
Senior engineer
amplifier
Repeatedly leaves the same review comments
Staff engineer
sponsor
Currently the manual enforcement layer for architecture
Engineering manager
buyer
Purchases to standardise a workflow already in use
VP Engineering or CTO
budget owner
Engages after engineer adoption is established
Security or AppSec
approval required
Reviews once usage extends beyond one team

Representative account: a senior engineer at a 150-person B2B SaaS company using Cursor daily across several services. Source: Hyrax Ideal Customer Profile, Notion, last updated 2026-09-01.

Appendix D — Channel Attribution

20 of 21
ChannelShareSignupsVisitorsImpressionsCost per signup
Paid — 55%
Google search and video25%93718,740468,500$21.34
Paid developer influencers12%4509,000225,000$22.22
Product Hunt ads8%3006,000150,000$16.67
Other paid social and search10%3757,500187,500$52.00
Organic — 33%
SEO and technical docs15%56311,260281,500$0.00
AI answer engines10%3757,500187,500$0.00
Organic social posting8%3006,000150,000$16.67
Word of mouth — 12%
Direct and in-product sharing8%3006,000$0.00
Private developer communities4%1503,000$0.00
Total100%3,75075,0001,650,000$20.96

Other paid social and search covers LinkedIn, X, Meta, Reddit, ChatGPT, TikTok and Bing. SEO docs are written in-house. AI answer engines are handled by the software stack. Direct and private community signups come from repository merge tags, engineers inviting teammates, and links shared in private Slack and Discord groups.

Appendix E — Product Phasing

21 of 21
Phase 1 · funded
Current product
  • Free developer workspace
  • Repository connected inside 72 hours
  • Team upgrades at $30 a seat
  • 8 seats at start, 20 at settle
The full $78,599. All channel and cost targets are set against the current product.
Phase 2 · when the product changes
New features
  • New message, new buyer, new channels
  • Its own numbers and cost targets
NEED FROM PRODUCT: what it does, when it ships, who buys it.
Phase 3 · when the product changes
New features
  • No channel is funded before the product supports it
NEED FROM PRODUCT: what it does, when it ships, who buys it.

Phase 1 is the product as it works now: a free workspace, a repository connected within 72 hours, and $30 a seat.