PostHog Pricing 2026: Rates and Real Bills

PostHog has no plan tiers. You add a payment method and pay per unit for each product you use, and every product carries its own free monthly allowance: 1 million analytics events, 5,000 web session recordings, 1 million feature flag requests, 1,500 survey responses and 100,000 error tracking exceptions. Analytics starts at $0.0000500 per event past the free million and slides down to $0.0000090 at very high volume. That makes PostHog one of the few tools in this category you can budget properly before speaking to anyone. It also makes the bill harder to predict than it looks, because there are roughly ten meters running at once and the headline event rate is not what most B2B products actually pay. The figures below were read directly off posthog.com/pricing and PostHog's own documentation in July 2026.

How much does PostHog cost?

There is no license fee, no seat fee and no platform fee. You pay for what you use, per product. Here is the whole published rate card:

Product	Free every month	Starting rate after free
Product analytics	1M events	$0.0000500 per event
Session replay (web)	5K recordings	$0.0050 per recording
Session replay (mobile)	2.5K recordings	$0.0100 per recording
Feature flags	1M requests	$0.000100 per request
Surveys	1,500 responses	$0.100 per response
Error tracking	100K exceptions	$0.000370 per exception
Data warehouse	1M rows plus free historical	$0.000015 per row
Data pipelines	10K events and 1M rows	Usage based
AI observability	100K events	$0.00006 per event
Logs	10 GB ingested	$0.25 per GB
PostHog AI	500 credits, worth $5	$0.01 per credit

Every rate above is the entry band. All of them step down as volume grows, which is the part that makes PostHog cheap at scale rather than just cheap at the start.

How does PostHog's pricing model work?

PostHog bills on sliding volume bands, and each band only applies to the events inside it. This trips people up, so it is worth being precise: if you send 3 million analytics events, you are not charged 3 million times the top rate. The first million is free, the second million is charged at the 1 to 2 million rate, and the third million is charged at the lower 2 to 15 million rate.

Here are the published analytics bands in full:

Monthly events	Price per event
First 1 million	Free
1 to 2 million	$0.0000500
2 to 15 million	$0.0000343
15 to 50 million	$0.0000295
50 to 100 million	$0.0000218
100 to 250 million	$0.0000150
250 million and above	$0.0000090

At the top band you are paying roughly a fifth of the entry rate for the same event, which is unusually aggressive for this category. It is also why PostHog reads as expensive to nobody and cheap to almost everybody: the pricing is built to get less painful exactly as your volume gets scary.

What is the difference between anonymous and identified events?

This is the single most important thing to understand before you estimate a PostHog bill, and most pricing write-ups skip it entirely.

PostHog captures two kinds of event. An anonymous event carries only non-identifying context like referrer and device type. An identified event is tied to a person profile with your own user data attached, which means PostHog has to create or update a record in its backend database for it. That extra work costs more to process, so PostHog charges more for it. Its own cost documentation states that anonymous events can be up to four times cheaper than identified ones, and its pricing blog breaks the anonymous discount down by volume band: roughly 80 percent cheaper below 2 million events, 67 percent to 15 million, 55 percent to 50 million, 40 percent to 100 million, 20 percent to 250 million, and 10 percent above that.

Read that alongside the rate table and the consequence is stark. PostHog's public per-event table shows a single set of prices, so you have to combine the two sources to see the real picture. A marketing site or a consumer app where most traffic is logged out gets close to the headline rate. A B2B SaaS where essentially every meaningful action happens behind a login sends almost entirely identified events, and pays a multiple of the number it budgeted from the pricing page. If you are comparing PostHog to anything else, work out your identified share first. It moves the answer more than any feature comparison will.

The lever you have is the person_profiles setting in the SDK, which controls when a person profile gets created. Capturing events anonymously by default and identifying only when you genuinely need person-level analysis is the documented way to keep this line down. It is worth an hour of an engineer's time before you sign anything.

Is PostHog free?

Yes, and the free tier is the most generous in product analytics, because it applies separately to every product rather than to the account as a whole. A team using analytics, replay, flags, surveys and error tracking together gets a free allowance on all five at once, which no competitor matches.

The limits that actually end the free plan are not the volume caps. A free account is restricted to one project and one year of data retention, with community-only support. Adding a payment method raises that to six projects and seven years of retention and gets you email support. Most teams upgrade for the retention rather than because they blew through an allowance, and if you care about year-over-year comparisons that twelve-month ceiling arrives faster than any event cap does. Worth checking too: PostHog runs a startup credits program, and an eligible early-stage company can cover a long stretch of usage with it.

How much does PostHog session replay cost?

Web session replay gives you 5,000 recordings a month free, then $0.0050 per recording at the entry band. Mobile replay is metered separately at 2,500 free and $0.0100 per recording, so mobile costs twice as much per session and starts charging sooner.

Practically, replay is the line most likely to surprise you, because recording volume tracks sessions rather than instrumentation. You control your event count by choosing what to track. You do not really control how many times people visit. A product doing 30,000 sessions a month is looking at roughly 25,000 billable recordings, or about $125 at the entry rate before volume discounts. That is not expensive, but it is not the zero that the free tier implies either, and it scales with success rather than with effort. PostHog lets you sample recordings and set conditions on which sessions get captured, which is the standard way teams keep this proportionate.

What does PostHog actually cost per month?

Nobody publishes worked totals, so here are three, calculated from the published band rates above. These are our arithmetic on stated public rates, not quotes from PostHog, and they assume anonymous-weighted event traffic. Replay and feature flag rates also step down at volume, so those lines are upper bounds.

Usage profile	Analytics	Session replay	Feature flags	Estimated monthly total
Early-stage SaaS: 2M events, 6K recordings, 500K flag requests	$50.00	$5.00	Free	about $55
Growth SaaS: 10M events, 25K recordings, 3M flag requests	$324.40	$100.00	$200.00	about $624
Scale-up: 40M events, 60K recordings, 10M flag requests	$1,233.40	$275.00	$900.00	about $2,408

Two things stand out. First, the early-stage number is genuinely small, and a great many products never leave that row. Second, look at what happens to the growth and scale-up rows if those events are identified rather than anonymous. Applying PostHog's own published discount for the 2 to 15 million band, where anonymous is about 67 percent cheaper, an identified-heavy 10 million event month is closer to $970 of analytics than $324. The total roughly doubles. That is the same product, the same traffic and the same tracking plan, with the bill decided by one SDK configuration choice.

Is there a PostHog pricing calculator?

Yes. PostHog puts an interactive calculator on its own pricing page where you set volume per product and see the monthly total update, and it is the right first stop because it always reflects current rates. Use it, then do two things it will not do for you.

Adjust for your identified event share, since the calculator works from the headline rates. And add the products you will end up switching on later. Teams almost always arrive for analytics and then turn on replay within a month and error tracking within a quarter, because they are already there and each one has a free tier that makes trying it free. That is a reasonable way to consolidate a stack, but it means the number you calculate on day one is rarely the number you pay in month six. Budget for the products you expect to adopt, not just the one you came for.

How does PostHog pricing compare to Mixpanel and Amplitude?

PostHog and Mixpanel are the two tools in this category that publish real per-unit rates, which makes them directly comparable in a way most competitors are not.

	PostHog	Mixpanel	Amplitude
Free analytics events	1M per month	1M per month	2M per month
Published rate above free	$0.0000500 per event, sliding to $0.0000090	$0.28 per 1,000 events, which is $0.00028 each	None published above Plus
Feature flags	Included, 1M requests free	Paid add-on	Not a separate charge
Experimentation	Included	Paid add-on	Included on every tier
Surveys	Included, 1,500 responses free	Not offered	Included, limited on lower tiers
Error tracking	Included, 100K exceptions free	Not offered	Not offered
Plan tiers	None. Pure usage	Free, Growth, Enterprise	Free, Plus, Growth, Enterprise

On the entry band PostHog is roughly five and a half times cheaper per event than Mixpanel, and it bundles three things Mixpanel either charges extra for or does not sell. The identified event multiplier narrows that gap considerably for B2B products, which is the nuance to hold on to. Our full Mixpanel vs PostHog comparison works through where each one wins feature by feature. Amplitude is the odd one out here: it publishes no rate above its Plus tier, but it gives the largest free allowance at 2 million events and bundles session replay, experimentation and in-app guides into every plan including the free one, so on a features-per-dollar basis its free tier is the strongest of the three, and the full breakdown of Amplitude pricing by plan covers what the two quoted tiers actually cost. Because both publish so much, that pairing is the easiest in the category to settle with arithmetic, and we work it through row by row in PostHog vs Amplitude. The wider product analytics tools field is worth scanning before you settle.

Is PostHog worth it?

For an engineering-led team, it is the best value in the category and it is not especially close. The rates are public, the free tiers are real, nothing is gated behind a plan you have to negotiate, and consolidating analytics, replay, flags and error tracking onto one bill can genuinely retire three vendors. If you are currently paying for a product analytics tool plus a flagging tool plus an error tracker, run the arithmetic above against those three invoices. It usually favors consolidation by a wide margin.

The costs that do not appear on the rate card are the ones to weigh. PostHog assumes engineering ownership: you will define events, tune sampling, manage person_profiles and maintain the taxonomy, and a team without someone who wants that job tends to get less out of it than a polished hosted tool would give them for more money. Data retention on the free plan is one year, so plan for the upgrade if you care about annual comparisons. And if you are considering PostHog specifically because it is open source and you want to self-host, read the honest account of what self-hosting PostHog involves before you commit: PostHog's own documentation says self-hosted deployments are officially unsupported, all paid features are Cloud-only, and new Kubernetes deployments are no longer supported.

Costs like these also have a habit of arriving as a surprise at the end of a quarter rather than the start, which is why teams running several usage-billed vendors at once increasingly track where each table and metric in the warehouse came from alongside what it costs to produce. PostHog's own data warehouse product, free to 1 million rows, makes that easier than it used to be.

The honest limit is the same one every tool on this page shares. PostHog measures behavior with real precision and tells you nothing about motive. It will show you that 41 percent of new workspaces never invite a second user, reliably and cheaply. It cannot tell you they did not invite anyone because the invite email lands in spam, because that fact lives in support tickets and reviews rather than in the event stream. If this is your team's first tool in the category, what product analytics actually is is the right place to start.

UserInsight sits on the other side of that gap. It runs alongside PostHog rather than replacing it, joining the behavioral signal you already trust to the tickets, reviews and survey replies from the same users, so a drop-off arrives with candidate causes attached and every finding links back to the evidence behind it.