What the PostHog Startup Program Gives You
PostHog's startup program offers up to $50,000 in credits toward the PostHog platform, which bundles product analytics, session replay, feature flags, experiments, surveys and error tracking onto one usage-based bill.
That breadth is the point. Most analytics credits buy you one tool. This one covers a category stack that a growth-stage company would otherwise assemble from four or five separate vendors and four or five separate contracts. AI Perks tracks it alongside $7.7M in credits across 194 companies.
The honest way to read $50,000 is not "a year of free analytics." For a pre-product-market-fit team, $50,000 of list usage is closer to three or four years of natural spend. The credit is not a budget you need to spend. It is a window in which price stops being the reason you do not instrument something, and the whole outcome depends on what habits you form inside that window. Eligibility depends on your stage and funding, and current terms are listed on getaiperks.com.
What PostHog Actually Replaces
PostHog's value is not any single product in the bundle. It is that your analytics events, your session recordings, your feature flags and your experiments share one identity graph, so you can go from a metric dropping to the exact replay of a user hitting the bug without exporting anything.
Every component has a stronger standalone competitor. What you are buying is the join, plus one vendor relationship instead of five.
| Product in the bundle | What teams usually buy instead | Why the join matters |
|---|---|---|
| Product analytics | Amplitude, Mixpanel | Funnels and retention on the same events |
| Session replay | FullStory, Hotjar, LogRocket | Jump from a funnel drop-off straight to the recording |
| Feature flags | LaunchDarkly, Statsig | Flags and the metrics they move share one dataset |
| Experiments | Optimizely, Statsig | A/B results computed on events you already send |
| Surveys | Typeform, Sprig | Target a survey at a behavioural cohort |
| Error tracking | Sentry, Rollbar | Errors attached to the session that produced them |
PostHog is also open-source and stores events in ClickHouse, which means you can query raw event data with SQL rather than waiting on a vendor to add a chart type. For a technical founding team that is usually the feature that decides it.
The test of whether you need this yet is simple: do you have a metric you check weekly that a decision actually depends on? If not, a pageview counter is enough and the credit is better activated later.
How PostHog Pricing Behaves at Scale
PostHog does not have one price. Each product in the bundle meters separately, each carries its own monthly free allowance, and per-unit rates fall as volume climbs.
That structure is founder-friendly at the bottom and surprising in the middle. The free allowances are large enough that most pre-launch teams pay nothing at all. The trouble starts at the point where usage grows faster than revenue.
| Meter | Roughly what it prices | What makes it spike |
|---|---|---|
| Analytics events | Fractions of a cent per event past a monthly free allowance | Autocapture on an interactive app: every click becomes billable |
| Session replay | Around half a cent per recording, tiered down with volume | Recording 100% of sessions on a high-traffic consumer product |
| Feature flags | A fraction of a cent per flag evaluation request | Evaluating flags client-side on every page load instead of caching |
| Experiments | Billed through the flag and event meters | Long-running tests left on after a decision was made |
| Surveys | Cents per response | An always-on survey on a high-traffic page |
| Error tracking | Per exception captured | A noisy error loop shipping to production on a Friday |
Those rates are approximate and PostHog publishes current pricing openly. Verify before you model anything.
A worked example. A B2B product with 50,000 monthly active users, with autocapture on, plausibly generates around 10 million events a month. Recording every session at roughly three sessions per user is another 150,000 recordings. At the rough rates above, that lands somewhere near $1,000 to $1,500 a month, or $12,000 to $18,000 a year. A $50,000 credit covers that footprint for roughly three years.
Now change one variable. Move from B2B to a consumer app where users click constantly, and event volume per user can rise tenfold without your revenue moving at all. That is the shape of the risk: PostHog's bill tracks user activity, not user value. AI Perks lists the credit terms. The discipline is yours.
What PostHog Credits Stack With, and What They Cancel Out
Analytics credits are a separate invoice from compute and a separate invoice from model APIs, so they stack cleanly with both. They do not stack with other product analytics credits, because you can only instrument your product once.
This distinction is worth more than it sounds. Founders collect credits as though they were all additive, then discover that three of them cover the same job.
- Cloud credits cover where your code runs, and do not touch a third-party SaaS invoice
- Model and API credits cover the inference your product calls
- Observability credits cover whether your infrastructure is healthy
- PostHog credits cover whether anyone is actually using the thing
Note the fourth line against the third. Observability tells you the server is up. Product analytics tells you the feature is dead. They answer different questions and both bills are real, which is why they stack rather than overlap.
Where PostHog genuinely substitutes is against Amplitude, Mixpanel, LaunchDarkly, Hotjar and Sentry credits. Holding a large PostHog grant and a large competing analytics grant at the same time is not $100,000 of value, it is $50,000 of value and a migration you will eventually have to do. Seeing which grants are compatible instead of redundant is the reason AI Perks exists as a tracked list rather than a folder of bookmarks.
What Founders Get Wrong About Product Analytics Credits
The most expensive mistake is letting a credit remove the price signal at the exact moment your team is deciding what to instrument, because whatever you turn on while it is free is what you inherit at list price later.
Five patterns, in rough order of cost:
Leaving autocapture fully on. Autocapture is the right default for week one and the wrong default at scale. It is what makes analytics bills grow with clicks rather than with customers. Narrow it once you know which events matter.
Sending identified events for anonymous traffic. PostHog prices events that carry a full person profile differently from anonymous ones, so sending person profiles for logged-out marketing-site visitors is a common and avoidable multiplier. Configure identification deliberately rather than accepting the default.
Recording every session. Session replay is the fastest-growing line on most PostHog bills. Sampling, or conditioning recording on a cohort that matters, keeps the useful recordings and drops the cost.
Instrumenting everything instead of the decision. Hundreds of event types nobody queries cost money every month and produce no decisions. A dozen well-named events beat four hundred autocaptured ones.
Planning the offboarding too late. Credits are denominated at list price and the month they end is a cliff, not a ramp. Decide at 70% consumed what you will sample down or turn off, not at 100%. Other analytics-category credits that cushion that transition are tracked at getaiperks.com.
Where PostHog Sits Among Analytics Credits
Analytics is the one category where holding more credits leaves you worse off, so the question is which single vendor to standardize on, not how many programs to collect.
AI Perks lists PostHog in the Analytics category next to the other product analytics, observability and data warehouse programs, with current terms for each.
Two things decide the outcome. Route changes the terms: a meaningful share of large software credits move through accelerator and investor partner channels, and the amounts differ by route, so your investor perks page is worth reading alongside the public list. Activation timing is a lever: credit clocks generally start when the account is switched on rather than when it is granted, so switching on before you have traffic worth measuring spends the balance on a product nobody is using yet.
Frequently Asked Questions
How much is the PostHog startup program worth?
Up to $50,000 in credits toward PostHog, covering product analytics, session replay, feature flags, experiments, surveys and error tracking on one bill. For an early-stage team that typically represents three or more years of natural list spend. Current terms and eligibility are tracked at getaiperks.com.
Is PostHog free for small startups?
Effectively yes. Each product in the bundle carries a monthly free allowance, and most pre-launch and early-traction teams stay inside it without paying anything. The startup credit matters at the point where usage outgrows those allowances, which is usually well after launch rather than at it.
Does PostHog replace Amplitude and Mixpanel?
For most startups, yes, and it also absorbs feature flags, session replay and surveys that would otherwise be separate vendors. Amplitude and Mixpanel remain stronger at large-enterprise analyst workflows. The tradeoff is depth in one product against one identity graph across six.
Do AWS or Google Cloud credits cover PostHog?
No. PostHog Cloud is a third-party SaaS invoice and sits outside your cloud credit balance, which is exactly why the two stack rather than overlap. Holding compute, model and analytics credits together covers three separate recurring bills. See which combinations are compatible at getaiperks.com.
What makes a PostHog bill grow unexpectedly?
Event volume from autocapture and session replay volume, in that order. Both scale with how much users click rather than with revenue, so a consumer product can ten-times its bill without changing its business. Narrowing autocapture and sampling replays are the two changes that fix most of it.
Can I self-host PostHog instead of paying?
PostHog is open-source and a self-hosted deployment exists, but running ClickHouse and the ingestion pipeline yourself is real infrastructure work, and the managed product is where the company focuses. For a small team the engineering time usually costs more than the bill a credit would have covered anyway. AI Perks tracks $7.7M in credits across 194 companies at getaiperks.com.
Measure what your users actually do. Let someone else pay for the first three years of it.