Staats Review (2026): Pricing, Features & Honest Verdict

Staats is a cookieless website analytics service with no dashboard: your coding agent queries it over MCP and reports back in chat. It is the cleanest analytics pricing we have seen this year. Best for: indie makers running a portfolio of small products. Price: from $9/mo (free plan: yes). Rating: 7.5/10.

What is Staats?

Staats is website analytics built to be read by a machine instead of a person. You drop a 1.5KB script into your site, connect your coding agent to a Model Context Protocol server with one line, and then you ask questions in the same chat window where you write code. “How did the launch go?” replaces opening a tab and reading charts.

It launched on Product Hunt on 29 August 2026 and finished the day at rank 11 with 72 points. The terms were last updated on 23 August and the privacy policy on 21 August, so this is a very new product. The web dashboard exists, but it only manages sites, keys, usage and billing. Every actual number arrives through the agent. That is the whole bet, and whether you like Staats comes down to whether you believe an agent asking “traffic is up 24%, want me to look at why?” beats a chart you have to remember to open.

What Are Staats’ Key Features?

A single MCP connection that covers every site you own

The setup really is one command, because the account key lives in the URL path rather than in a header. You run claude mcp add --scope user --transport http analytics "https://api.staats.ai/functions/v1/mcp/ak_..." and the connection is live. One key covers every current project and every future one, so adding site number eleven does not mean touching your MCP config again. Clients that prefer standard auth can send a Bearer token to the bare endpoint instead. It works with Claude Code, Cursor, Windsurf, Codex and VS Code through Roo Code, Cline or Copilot.

Drop-off funnels and referrer tracing

Staats maps visitor journeys from landing page through to checkout and reports where people leave. The agent is meant to point at the leaking step rather than hand you a funnel chart to interpret. There is also anomaly tracing: a traffic surge gets traced back to its referrer and surfaced with context, so a spike arrives as “Hacker News sent 1,376 visitors” instead of as a bump on a line you have to go looking for. Whether that stays useful or turns into noise depends entirely on how disciplined the alerting logic is, and eight days in there is no way to know.

Deploy annotations and before-and-after comparison

The agent tags every deploy, then compares the metrics on either side of it. This is the feature that justifies the whole approach. Your agent already knows what changed in the code, so it is the only thing in your stack that can connect “p95 dropped” to “we shipped the index on Tuesday” without you typing the annotation yourself. Traditional analytics makes you remember to add the marker. Here the thing doing the deploying adds it.

Cookieless tracking with no visitor identifier to leak

No cookies, no cookie banner, no IP address written to disk. The IP is used in memory to compute a salted one-way visitor hash, and the salt rotates daily, so yesterday’s identifier cannot be joined to today’s. Geography comes from timezone rather than IP lookup. That is a weaker geography signal than most competitors offer, and Staats says so plainly rather than pretending otherwise.

Autonomous instrumentation with one HTML attribute

Adding a tracked event means putting data-track="event_name" on a button or link. No JavaScript, no SDK call, no event schema to maintain. The pitch is that your agent adds the attribute while it is writing the feature, so instrumentation stops being a separate chore you do afterwards and never do. For custom triggers there is still a window.track() escape hatch.

How Much Does Staats Cost?

Staats starts at $9 per month, and there is a free tier at $0 that gives you 10,000 events a month, one site, 30 days of history and every MCP tool. No trial clock, no card.

The $9 Pro tier is where this gets interesting. It buys 100,000 events a month, 10 sites and a year of history. Plausible charges the same $9 for 10,000 pageviews on a single site. Same price, ten times the volume, ten times the site count. Fathom wants $45 a month for 500,000 pageviews and sells extra sites in packs of 50 for another $10 a month. Staats Growth is $29 for a million events and 25 sites.

Two caveats keep this from being a rout. First, quotas count events, not pageviews, and every data-track attribute you add spends from the same budget as your traffic, so a heavily instrumented site burns through 100,000 faster than a raw pageview count suggests. Second, Plausible’s cheapest paid plan keeps three years of data; Staats gives you one on Pro and 30 days on free.

Run the arithmetic across the ladder and the volume price falls the way it should: $0.09 per thousand events on Pro, $0.029 on Growth, $0.0099 on Scale. The site allowance moves the other way. Pro works out at $0.90 per site, Growth at $1.16, Scale at $1.98. So the higher tiers sell you traffic headroom, not portfolio headroom, and each extra site costs more the further up you climb. If you are the maker with 30 tiny projects and 50,000 total events a month, you are paying $99 for Scale’s 50-site ceiling while using half a percent of its traffic allowance. That is the one place the pricing does not match the product’s own “portfolio overview” pitch.

Free and Pro budget exactly the same amount of traffic per site, which is either a nice piece of design or a coincidence: 10,000 events across one site, and 100,000 across ten. Growth loosens that to 40,000 per site and Scale to 200,000. So the free plan is not a crippled version of Pro, it is Pro at one tenth the scale, and that consistency makes the upgrade decision unusually easy to reason about.

One nice touch: going over your quota does not stop collection dead. You get warnings from 80% usage and collection only pauses at double the limit, which means a Hacker News front page does not silently cost you the data from your best day of the year.

Who is Staats Best For?

Use Staats if you run several small products, live inside Claude Code or Cursor all day, and have already noticed that you never open your analytics dashboard. The free tier is real enough to test the workflow on one project without a card, and the 10-site allowance at $9 is priced for exactly the person with a portfolio of half-finished launches.

Use it if you want analytics that will not trigger a cookie banner. The privacy design is thorough and documented rather than asserted.

Skip Staats if you need long data retention. Thirty days on free and one year on Pro is short, and if you care about year-over-year comparisons this is the wrong tool. Skip it if your work is not agent-shaped: without an MCP client the dashboard does not do the job, because the dashboard deliberately is not a dashboard. And skip it if you need to know who you are buying from, because the site names no company, no founders and no legal entity anywhere, including the terms. The only third party identified is Polar, the merchant of record for payments, and Supabase, where the data lives.

Best Staats Alternatives

Plausible Analytics. Starter is $9 a month for 10,000 pageviews on one site with three years of retention, Growth is $14 for three sites, Business is $19 for ten. Open source, mature, and it has a proper dashboard you will actually look at. Better for retention and for teams. Worse if you want the numbers to come to you.

Fathom Analytics. $45 a month at the 500,000 pageview tier, with forever data retention and extra sites in packs of 50 for $10 a month. Considerably more expensive at low volume, and there is no free plan, but the retention story is the opposite of Staats’ and the company is a known quantity.

PostHog. The free tier covers a million analytics events a month, which beats every Staats paid tier on raw volume, and you also get session replay, feature flags and experiments. The cost is complexity: PostHog is a platform, not a 1.5KB script, and the billing has a lot more surface area. If you want one tool for everything, this is the one.

Final Verdict: Is Staats Worth It?

Staats is the first analytics product we have seen that treats the coding agent as the user rather than as an integration. The pricing backs the pitch up: $9 for 100,000 events and ten sites is aggressive against Plausible’s 10,000 pageviews and one site at the same number, and the free plan is usable rather than a trap. The privacy engineering is careful, the docs are short and correct, and the setup really is one line.

The reservations are about the company, not the product. Nobody’s name is on this. There is no about page, no founding team, no registered entity in the terms of service, and the whole thing is eight days old going by the policy dates. You are handing your traffic data to an anonymous operator on infrastructure they do not own. For a side project that is a fine trade at $0, and the CSV export means you can walk away with your data. For the analytics of a business you depend on, wait a few months and see who turns up. Start on the free tier, connect one project, and let it earn the $9.