When a marketing push or a bot crawl hits, several meters rise at once, and none of that traffic necessarily earns you anything. Dull but predictable: with a seat-based tool you know the invoice before the month starts. Believers in that view are right that a free tier is only a teaser if you can't say what happens after it. ### Why billing limits and volume tiers answer most of it PostHog lets you set a spending limit on each product, and per-unit prices fall as volume grows, so the worst case becomes a number you chose. The steps below go from the control that bounds the bill first to the habit that keeps it low: 1. Set a billing limit on each product in PostHog's billing settings. 2. Filter internal traffic and known bot events before they are counted. 3. Tune autocapture so you collect the interactions you analyze, not every click. 4. Sample session recordings instead of capturing every visit. 5. Shorten data retention to the window you actually query. 6. Review usage weekly, so a spike is an afternoon's problem instead of a month-end surprise. ### Where PostHog billing limits fall short Hit the limit and you stop collecting; it is a cutoff, not a forecast. That means gaps in the data you wanted most during a traffic spike. Setting it too low trades a surprise invoice for blind spots, and setting it too high gives up the protection. On rare behavior, sampling also costs you accuracy, which matters for small segments. And someone has to own the weekly review. ## Hidden costs and limits teams hit on PostHog Five places produce the costs that reach the invoice. Cap or measure each before launch. - **Autocapture records clicks and pageviews nobody asked for**, so event counts rise with traffic instead of with your questions. - **Session recordings meter separately from events**, so a busy site can burn through that allowance while the event meter looks calm. - Retention periods vary by product and plan, so history you assumed was there may expire before the year-over-year comparison you need. - Some controls, SSO enforcement among them, sit behind paid plans, so a security review can force an upgrade the usage numbers never predicted. - Support tiers and instrumentation effort cost money and engineer hours respectively, and the second never shows up as a line item. ### PostHog autocapture and event volume costs Under autocapture, every interaction becomes an event, including the ones you'll never put in a funnel, which makes it the fastest way to get data and the fastest way to get a surprise. Turn it off for noisy surfaces, or keep it and filter at the source. ### PostHog session replay costs and retention Because replay is priced on its own meter, it needs its own limit. Set a billing cap and a sampling rule, or one popular page can spend the budget meant for analytics. Check retention before you promise anyone a long look back. ### PostHog features locked behind paid plans Even though the free tier is generous on volume, volume isn't the only gate. Admin controls, enforced SSO and faster support are tied to paid plans, so a team that passes a compliance check may owe more than its traffic suggests. If you want events or alerts in a CRM, Slack or email, [Activepieces](https://www.activepieces.com) runs that as a workflow, and a ten-step flow costs the same 1 credit per run as a two-step one. It's still not a pricing fix: it moves data where people will see it, but it doesn't reduce the events PostHog counts. ## PostHog versus Amplitude, Mixpanel and Heap on price Treat the competitor figures below as prompts to check rather than quotes, since they remain unverified, though PostHog's per-event metering is the easiest of the four to model. Where the cost cell says to use the vendor's calculator, I have no confirmed number, and you shouldn't budget from my table. | Vendor | Free tier size | Pricing unit | Bundled products | Cost at 1M events | Status | |---|---|---|---|---|---| | PostHog | Monthly free allowance per product, listed on its pricing page | Events, with separate metering for recordings and flags | Product analytics, session replay, feature flags, experiments, surveys | Read from PostHog's pricing calculator | Unverified, checked 2026-11-02 | | Amplitude | Free starter plan, limits on its pricing page | Events, with user-based metering on some plans | Analytics, session replay, experimentation | Use Amplitude's calculator or ask sales | Unverified, checked 2026-11-02 | | Mixpanel | Free plan, limits on its pricing page | Events | Analytics, session replay | Use Mixpanel's calculator | Unverified, checked 2026-11-02 | | Heap | Free plan, limits on its pricing page | Sessions | Analytics, session replay | Use Heap's calculator or ask sales | Unverified, checked 2026-11-02 | Although all four offer a free entry point, the allowances are metered in different units, so the headline sizes can't be compared directly. It depends on how chatty your instrumentation is. A product that fires many events per visit burns an event allowance quickly and barely touches a session allowance. If your event volume is predictable and you can set billing limits, PostHog's separate meters favor you. If your traffic is session-heavy with few events per session, Heap's unit may suit you better. ## What Activepieces does about this Activepieces can't lower PostHog's event count, and I won't pretend it can. What it covers is the weak spot in the billing-limit advice above: a limit is a cutoff, not a forecast, and someone has to own the weekly usage review. That job suits a scheduled workflow. A flow can run once a day, pull your usage from PostHog's API with the HTTP step, compare it with the limit you set, and post to Slack or send an email when you pass, say, 70% of it. I haven't verified a ready-made PostHog integration, so assume you'd be calling the API directly. The alert arrives while there is still time to filter a noisy source, before you hit the cap and lose data during a spike. The pricing model suits this kind of monitoring. Activepieces charges 1 credit per run, and a ten-step flow costs the same as a two-step one. A check that fetches usage, branches on thresholds, and notifies two channels doesn't cost more because you split it into clear steps. On Zapier, which bills per task, or Make, which bills per module, those extra steps are what you pay for. Here the run is the meter, which is the same logic that makes PostHog easy to model. The core of Activepieces is open source under the MIT licence, so you can self-host the automation layer as well. The same trade-off from the self-hosting question applies: you take on servers and upgrades in exchange for dropping usage fees. For a daily usage check, the hosted version is usually the simpler choice. Build the alert first, set the PostHog limit second, and review the flow's output weekly in place of the manual review. ## Frequently asked questions Is PostHog really free? Yes, up to a monthly allowance on each product, and you pay only for usage beyond it. The allowance resets every month and unused amounts don't carry forward, so a quiet month earns you no credit against a busy one. Treat the free tier as a monthly threshold, not a balance.  Does PostHog charge per seat or per user? No, team size doesn't change the price; usage does. Adding a colleague costs nothing, but that colleague's dashboards, replays and experiments generate events, and events are what you're billed for. Compare this with seat-based tools, where a hire raises the bill the day they join. Here the bill moves only when your volume does. ### How do I stop PostHog bills from spiking? Set a billing limit on each product, then cut the events you don't need. A limit caps spend at a figure you choose, and once you reach it PostHog stops ingesting that product's data until the next cycle. So a cap protects your budget but can leave a gap. Set it above your normal usage and alert on it before you hit it. Then trim at the source: drop noisy events, sample session recordings, and avoid autocapturing everything on high-traffic pages. The cap is a safety net, and volume control is what keeps the bill low. ### Does PostHog offer startup or nonprofit discounts? If you're a startup, you can get credits that offset usage charges through a PostHog program, and eligibility rules are set in its docs. I can't confirm a nonprofit program from the material here, so check PostHog's current pricing and program pages before assuming one. Either way, the free allowance already covers a small team. A discount matters only once your volume outgrows it. ### Can I self-host PostHog to avoid usage fees? If you self-host the open-source version, PostHog's usage fees go away, but it's not free. You take on the servers, storage, upgrades and on-call time, and those costs grow with event volume just as the cloud bill would. For a small project, that trade can work. For a team without spare engineering hours, the cloud plan with a billing limit is usually the cheaper line once you count labor. Price both before deciding. ## Related reading - [Backblaze Pricing 2026 – Which Plan Fits You?](https://www.activepieces.com/blog/backblaze-pricing-2026-which-plan-fits-you) - [Claude Team Plan (2026 Guide)](https://www.activepieces.com/blog/claude-team-plan-2026-guide) - [Housecall Pro Pricing (2026): Plans by Team Size](https://www.activepieces.com/blog/housecall-pro-pricing-2026-plans-by-team-size) ## References - [PostHog](https://posthog.com/docs/product-analytics/pricing)Pay by traffic and you are not paying by revenue.