Channel 13 / 13 · Test efficiently
Paid search, ads & sponsorships
Use paid reach to test a known message—not to invent positioning.
Use it when
You know the ICP and can track a real conversion event downstream.
Don't confuse it with
Buying broad awareness because competitors do it or treating last-click ads as the full story.
01 / What companies actually did
PostHog published actual paid-channel lessons
Its 2023 article describes experiments and pitfalls of advertising to developers. The 2025 follow-up separates conversion from awareness and says its own budget is split roughly 50/50 between those objectives. That is PostHog’s practice, not a recommended universal allocation.
[P4] PostHog — What we learned running paid ads ↗[P1] PostHog — 40 things learned about developer marketing ↗PostHog recommends small initial tests
Its 2025 lessons suggest roughly $500 and two weeks per experiment, including brand search, relevant search queries, and cheaper awareness on developer-heavy social platforms.
[P1] PostHog — 40 things learned about developer marketing ↗PostHog cautions about last-click attribution
Its writing gives the example of a user discovering the brand elsewhere and later clicking a Google ad; it also asks signups where they heard about the company.
[P1] PostHog — 40 things learned about developer marketing ↗02 / How to do it
- 01Choose a qualified action: signup with activation, meeting with ICP or evaluation—not just click.
- 02Split search-intent tests from audience-awareness tests and label them separately.
- 03Run small bounded tests with correct conversion events and a clear stop condition.
- 04Ask buyers where they first heard about you; inspect assisted journeys and branded-search trends.
- 05Scale only campaigns whose quality and economics stay sensible, not those with cheapest CPC.
Things you can publish or run
03 / What good looks like
Watch this
Qualified cost per evaluation/opportunity, incremental lift where measurable, branded-search growth, opportunity quality.
Common trap
There is no universal “developer ads don’t work” rule; neither is there a universal budget split.
04 / Copy the worksheet
COPYABLE WORKSHEET / 13
Paid search, ads & sponsorships template
Replace the brackets. Use this as a starting point, not a rigid formula.
SMALL PAID EXPERIMENT Hypothesis: [specific audience will respond to specific offer] Audience/keyword: [high-intent, narrow] Ad type: [search conversion | awareness | sponsorship] Landing page: [technical page matching ad claim] Primary action: [activated trial / qualified meeting] Secondary signals: [brand recall / direct visits / assisted actions] Budget + duration + stop condition: [...] / [...] / [...] Evidence: [ad platform + product data + self-reported source] Decision: scale / refine / stop Note: do not equate last-click attribution with causal impact
05 / Original sources