Product Analytics Pricing: Event vs MTU vs Seat Models
Product analytics tools bill on one of three meters: events (you pay per action tracked), MTUs (you pay per monthly tracked user, no matter how busy they are), or seats (you pay per teammate who logs in). Which one is cheapest depends entirely on your event-to-user ratio. A chatty app with heavy users gets punished by event pricing and rewarded by MTU pricing; a big, mostly-idle user base is the reverse. Below is a normalized cost table so you can stop comparing apples to invoices.
The eval that made me write this
Last quarter I was pricing analytics for a B2B onboarding tool. Roughly 120k monthly users, but a nasty distribution: about 8% of accounts fired 200+ events a session because they lived in a bulk-import wizard, and the rest barely touched the product. I ran the same usage profile through three vendors' calculators on the same afternoon, with the same event export.
The spread was absurd. One quote came back around $600 a year. Another cleared $40k. Same data. The only variable was the meter.
That's the whole game with product analytics pricing. The sticker on the pricing page tells you almost nothing until you map it onto your own event-to-user ratio. So let's build the map.
The three meters, plainly
Event-based. You pay per tracked event. Mixpanel and PostHog both run this way. It's beautifully honest for low-engagement apps and brutal for anything with high per-user activity. Every page_view, every button_clicked, every autotracked scroll is a line on the bill.
MTU-based (monthly tracked users). You pay per unique user who fires at least one event in the billing month. Amplitude is the reference implementation here. Heavy users are free to be heavy, which is great, but you get charged for a user whether they did one thing or ten thousand. Amplitude's own docs describe a dual model where you have both an MTU allowance and an events-per-MTU allowance, and blowing either one triggers overage.
Seat-based. You pay per human who logs into the dashboard, and volume is unmetered (or lightly metered). This is the classic BI and dashboard-tool model, and it survives in blended pricing where a vendor sells you a base platform fee plus seats plus a usage line. Pure seat pricing for product analytics is rarer now, but the seat component hides inside a lot of "custom" enterprise quotes, which is exactly where it gets expensive quietly.
Here's the thing about the labels: most vendors mix meters. Amplitude counts MTUs and events. Mixpanel's enterprise tier adds seat-flavored packaging on top of events. Treat "event vs MTU vs seat" as the primary axis, then check the fine print for the second meter that catches you.
Normalized cost table
To compare anything, I fixed an assumption: about 20 events per user per month, which is a middling-engagement B2B web app. That converts user counts into event counts so the two meters land on the same chart.
| Scenario | ~Events/mo | Event model (Mixpanel) | Event model (PostHog) | MTU model (Amplitude) |
|---|---|---|---|---|
| 100k users | ~2M | ~$30k-70k/yr | ||
| 1M users | ~20M | few hundred to ~$1k/mo, tapering | ~$150k-250k+/yr | |
| 10M users | ~200M | Enterprise, custom | Enterprise, custom | Enterprise, well above list |
A few notes before anyone quotes these at a vendor.
The Mixpanel numbers come straight from its published Growth math: the first 1M events are free, then it's $0.00028 per event, capped at 20M events a month before you're forced into Enterprise. OpenPanel's 2026 breakdown and UserCall's cost guide both peg ~5M events at roughly $1,120/mo and ~10M at roughly $2,520/mo, which lines up with my own calculator run.
PostHog is the outlier on the cheap end. Its pricing page gives you the first 1M events free, then starts at $0.00005 per event, so 2M events really is about $50 a month. At scale the per-event rate tapers hard, with volume discounts PostHog says reach up to 82%, and there's no minimum spend or annual lock. That's why the 1M-user cell is a range, not a point: the taper makes it genuinely hard to predict without running the exact volume through their calculator.
Amplitude's cells are contract ranges, not list prices, because Growth and Enterprise are quote-only. Userpilot's 2026 pricing analysis puts 50k-150k MTU deals around $30k-70k a year and 1M+ MTU deals at $150k-250k+, based on real buyer data. Notice what happens across the row: at 100k users the MTU model is roughly ten times the price of event pricing, and at 1M users it's still multiples higher. MTU pricing only wins when your events-per-user ratio is high enough to blow up the event bill. Flip my assumption from 20 events per user to 500, and the columns swap places.
That's the single most useful takeaway here, so I'll say it flatly: cheap pricing model is a function of your events-per-user ratio, not the vendor's brand. Compute your ratio first, then read the pricing page.
Watch the columns swap
Because this is the part people don't believe until they see it, here's the same 1M-user scenario run at two different engagement levels.
At 20 events per user, you're generating 20M events a month. On Mixpanel's Growth math that's roughly $5.3k a month right at the plan cap, and it's the moment event pricing starts to hurt. On Amplitude's MTU model, 1M users sits in that $150k-250k+ annual band. Event pricing wins by a wide margin.
Now crank the same million users up to 400 events each, because you built something sticky that people live inside all day. That's 400M events a month. Event pricing is now firmly in custom-enterprise territory and climbing with every event, while the MTU bill barely moves, because Amplitude still only sees 1M users. The user count didn't change. The meter did, and it flipped which vendor is cheaper.
I've watched teams sign a three-year MTU deal on the strength of "we're high-engagement," only to discover their real EPU was 15 because most signups never activated. They overpaid for a meter that assumed usage they didn't have. Measure before you commit.
Do the math on your own ratio first
Pull one number before you talk to anyone: total monthly events divided by monthly active users. Call it your EPU.
- Low EPU (under ~30): event pricing usually wins. You have a lot of lightweight users, and paying per user (MTU) means paying full freight for people who barely did anything.
- High EPU (over ~150): MTU pricing usually wins. Your users are busy, and event pricing bills you for every twitch.
- Somewhere in between is where the calculators earn their keep, and where "generous free tier" marketing does the most damage, because the free allowance covers your test data and expires the week you go to production.
Worth knowing: autotracking quietly inflates EPU. If a tool captures every click and pageview by default, your event count can be 3-5x what your hand-instrumented count would be, which is fantastic for coverage and terrible for an event-metered bill. On MTU pricing that same autotracking is free. This is not an accident in how the meters were designed.
The pricing gotchas, before the feature talk
I always name the gotcha before the feature list, because the gotcha is what shows up on the renewal, not the demo. Here's what actually moved the numbers in my evals.
Overage rates are punitive on purpose. Volument's 2026 analytics pricing guide notes overages frequently run 2-3x the standard rate, and Amplitude specifically charges 1.2x the contracted rate on both its MTU and event allowances. Translation: a viral week or a botted signup wave doesn't just cost more, it costs a premium. On usage-based event models you can usually set a hard billing cap; on annual MTU contracts you often can't, and the overage lands as a surprise line item.
Data retention beyond 12 months is a tier, not a setting. That same Volument guide flags extended retention adding 10-30% to contract value. If your analysis leans on year-over-year cohorts, budget for it up front or your historical data silently ages out.
Warehouse export is frequently paywalled. Getting your own events into Snowflake or BigQuery can require the Enterprise tier or a separate pipeline fee. This is the one that stings, because it's your data and you're paying to get it back out at volume.
Sampling is a hidden discount you didn't ask for. Some free and mid tiers sample high-volume events to keep costs down, which quietly degrades the accuracy of exactly the funnels you're trying to trust. Check whether the tier you're pricing reports on 100% of events or a sample.
Seats multiply in enterprise quotes. When a "platform fee" arrives with a per-seat line, adding your data team, your PMs, and read-only stakeholders can double the number faster than event volume ever will. If a vendor won't give you unlimited or generous seats, price the org you'll have in 18 months, not today.
Reading a quote without getting played
When the custom quote lands, three lines decide your real cost, and none of them is the headline price.
First, find the meter and its allowance. If it's MTU, ask what counts as a tracked user, because some vendors count anonymous visitors and some don't, and that definition can double your number overnight. Amplitude's MTU guide is worth reading closely on this; the counting rules are not intuitive.
Second, find the overage clause and whether you can cap it. A contract with uncapped 1.2x-to-3x overage and no billing ceiling is a contract that hands your budget to your growth curve. On a viral month that's a feature for the vendor, not for you.
Third, find every line that says "add-on": retention, warehouse export, extra seats, sampling removal, SSO. Add those to the base and compare that total across vendors, because the cheap base with four mandatory add-ons routinely beats out more expensive on the invoice. The sticker price is marketing. The loaded total is the number you'll actually defend to finance.
So which model should you pick
For most early and mid-stage teams with low-to-moderate EPU, event-based usage pricing is the safer default. It scales down to near-zero when you're small, you can cap it, and you're not signing an annual commitment before you know your real volume. PostHog and Mixpanel both let you start metered and grow into it, and the generous free tiers actually cover a real early-stage product rather than just a demo.
MTU pricing earns its premium when you have a heavily engaged user base, a stable and predictable user count, and enough budget to negotiate a multi-year deal with locked overage rates. Amplitude's depth is real; you're paying for it, and you should only pay for it once your EPU is high enough that event pricing would cost more anyway.
Seat or blended-enterprise pricing makes sense when analytics is a company-wide utility with a fixed, known audience of dashboard users, and you value flat, predictable billing over pay-for-what-you-use. Just interrogate the second meter hiding under the seat line.
If you want the deeper feature-by-feature teardown behind these price tags, we compared the tools head to head in Mixpanel vs Amplitude vs PostHog vs Heap. And if you're trying to sanity-check whether your EPU and engagement are even normal for your category, the 2026 product analytics benchmarks are a decent reality check before you argue with a sales rep.
One more model worth a line: some newer AI-native platforms, Kixo among them, bundle product analytics, replay, and messaging under per-project MAU-bracketed plans (FREE / GROWTH / ENTERPRISE) rather than metering raw events, which shifts the math again if you were going to pay for several point tools separately. Run your own volume through it the same way; the meter, not the marketing, decides.
Who should not optimize for this
A warning, since I end every verdict with who should walk away. If you're pre-product-market-fit with under 50k users, stop pricing analytics tools. Every option on this page has a free tier that covers you, and the hours you'd spend modeling event volume are better spent getting users worth measuring. The pricing model only matters once your bill is real. Come back when it is, EPU in hand.