Hyrax GTM - Weekly Update
Sep 22 2026
TL;DR
Mostly organic marketing and minimal paid spend until we get a roadmap. Leads are down 54% after cutting paid spend 90%.
METRICS
Spend and leads, last 7 days
| Last 7 days | vs prior | |
|---|---|---|
| Leads | 153 | -54% |
| Paid spend | $5,202 | -26% |
| Cost per lead | $34 | — |
| Business-domain leads | 27 | — |
| Cost per business lead | $193 | — |
Onboarding funnel, last 7 days
| People | Lost here | |
|---|---|---|
| Account created | 98 | |
| Wizard started | 43 | 55 |
| Org connected | 27 | 16 |
| Repo selected | 22 | 5 |
| Confirm reached | 19 | 3 |
| Wizard completed | 17 | 2 |
Against benchmark
| Hyrax | Benchmark | |
|---|---|---|
| Cost per click | $4.66 | $5.58 – $9.76 |
| Cost per lead | $34 | $104 – $524 |
| Business-domain share | 18% | — |
| Connected an org | 28% | 25% – 35% |
WordStream 2025 and Metadata 2026. Activation across 500+ PLG products. Activation rate 35.7%, zero paid subscriptions.
SHIPPED
- Product Hunt launch — yesterday. Top 5 with 221 upvotes. We missed top 3 so we do not pay the agency.
- Paid — cut Google from $2,000 a day to $250 on the 16th.
- Budget — sent Friday with a Loom.
- Homepage CTA test — the main button on the site is either Start free or Book a demo. Testing which one people click. Even so far.
PRIORITIES — NEXT CYCLE
- Promoting the Product Hunt results — email, organic social and the website.
- Synter customer referral agreement — sign this week. 10% of anything Joel refers that pays for a year.
- Homepage design — putting in better designs based on Neil's recommendations.
- Demo video — adding a new shorter one to the website.
ASKS
- None this week
OTHER
Budget on hold until after the offsite.·Onboarding we changed the flow. New users land in a workspace with three example repos already audited and scored, 47 findings, before they connect anything. app.hyrax.dev/start/w/acme/repositories·Agencies met Product Hunt and the influencer agency to go through the results. Pausing both until we know what the product is going to be.