PostHog Pricing: How Usage-Based Billing Adds Up

PostHog sells many products under one roof: product analytics, session replay, feature flags, experiments, surveys, error tracking and a data warehouse. Each is billed on its own usage, and each has a free monthly allowance. That structure is generous at small scale and needs attention at large scale.

This is a guide to reading that model. We make OnyxMetric, a competing product with flat pricing, so we flag where PostHog is the better pick too. For exact rates, use PostHog's pricing page, since numbers and allowances change.

How the billing is structured

At time of writing, PostHog's allowances include around one million analytics events and a few thousand session recordings per month before charges begin. Beyond that, each product bills per unit, with rates that fall in tiers as usage grows. You can set billing limits per product to cap spend.

Three consequences follow from this design.

You pay per product you turn on. Analytics, replay and flags each have a meter. A team that uses all of them watches several meters.

Cost tracks behavior. Events, recordings, flag requests and exceptions all depend on traffic and on how your code is written. A noisy flag check or a chatty client can raise a meter without any change in business results.

The free tier is real. Many early-stage products stay inside the allowances for a long time. That is a strong argument for PostHog at the start.

Where bills tend to grow

Teams usually report the same few pressure points:

• Session recordings. Replay is priced per recording. A busy site can burn through the allowance quickly unless you sample or filter.
• Autocapture. Capturing every click and pageview by default creates volume you may never analyze. Turning it down helps, but takes a deliberate pass.
• Feature flag evaluations. Each request for a flag value counts, so apps that evaluate flags on every render add up.
• Multiple environments. Staging and preview traffic count unless excluded.

PostHog gives you controls for these: sampling, billing caps, event filters. Using them is work, and the work is an ongoing cost.

A worked example

Take a content site with 300,000 pageviews a month and replay enabled for everyone. Pageview events alone may sit inside the free allowance, but if recording is on for every session, recordings can pass the free allowance within days. The fix is sampling, for example recording ten percent of sessions or only sessions that hit an error. That works, and it is a decision someone has to make and revisit as traffic changes.

With a flat plan, the decision is simpler but not absent: replay is limited to the Pro tier, so you choose a tier once rather than tuning a sampling rate every quarter.

Strengths worth acknowledging

PostHog is broad. If you want surveys, a warehouse, group analytics and SQL access inside the same product, few competitors match its scope. Its engineering-first approach also suits teams that want to configure deeply and read the code behind their tools. For a product organization that will actually use that breadth, usage-based pricing is a reasonable way to pay for it.

What flat pricing changes

OnyxMetric charges per site per month: Base $99, Plus $199, Pro $299; yearly is ten times the monthly price. There are no meters for events, recordings or flag requests, so a spike does not change the invoice.

Included, as the product ships today:

• Real-time traffic, sessions, goals, custom events, funnels and retention.
• Session replay on the Pro tier.
• Web Vitals and error tracking.
• Feature flags and experiments.
• Google Search Console integration and an MCP server for AI assistants.
• Cookieless tracking.

The tradeoffs are real. OnyxMetric has no surveys and no data warehouse. And at low volume, PostHog's free allowances will cost you less than $99 per month. The flat model pays off when your usage is high and steady, or when you want one number to put in a budget.

What to do if you stay on PostHog

Switching is not always the answer. If you stay, set a billing limit per product so a bad deploy cannot create a surprise invoice, sample session recordings, exclude internal and staging traffic, and review autocapture settings once a quarter. These habits keep a usage-based bill close to the value you receive, and they take an hour or two to set up.

A simple test

Pull last month's numbers from PostHog's billing page: events, recordings, flag requests, exceptions. Then:

1. Price the same usage at twice the volume.
2. Add the cost of any engineering time spent tuning sampling and filters.
3. Compare the total with the OnyxMetric tier that covers your needs.

If your PostHog bill is near zero, stay there and revisit when it grows. If it is already in three figures and climbing, flat pricing may be worth a trial.

Questions to settle

• Do you need surveys or warehouse features? If yes, PostHog.
• Do you want a predictable bill and a focused interface? OnyxMetric is built for that.
• Is replay central? Check both: replay is Pro-tier in OnyxMetric, and metered in PostHog.
• Who will maintain event hygiene? Metered tools reward someone who owns it.

For a feature-by-feature view, see our PostHog comparison. If you are mapping out other options, the Amplitude alternatives and Mixpanel alternatives guides cover neighboring tools.