How much does Rybbit Cloud cost compared with invisit?
invisit costs less than Rybbit Standard, or the same, at every volume up to 3M events, and less than Rybbit Pro at every volume. As of November 2026, Rybbit’s pricing is $19 a month for 100K events on Standard and $39 on Pro, $69 and $139 for 1M, and $249 and $499 for 10M. invisit is free to 20K events, then $14 for 100K, $49 for 1M and $199 for 10M.
The steps in between, on monthly billing:
| Events a month | invisit | Rybbit Standard | Rybbit Pro |
|---|---|---|---|
| 250K | $24 | $29 | $59 |
| 500K | $49 | $49 | $99 |
| 2M | $99 | $99 | $199 |
| 3M | $99 | $149 | $299 |
| 5M | $199 | $149 | $299 |
Rybbit Standard is cheaper at 5M, where invisit’s next step is its 10M plan. Paid yearly, Rybbit costs eight months’ price and invisit nine (invisit billing), so at 1M a year is $552 on Rybbit Standard and $441 on invisit.
What counts differs. Rybbit’s slider is labelled in pageviews, but its billing docs count page views, custom events, web vitals, outbound link clicks, and autocaptured button clicks, copies, form submits and input changes. A site with 300K page views and 30K custom events uses 330K events on both and pays $49 on either. Turn on Rybbit’s web vitals or button-click autocapture and those count too, so the same site can cross into Rybbit’s 1M step at $69. invisit bills page views, custom events and its own outbound clicks and downloads, and never bills web vitals, heatmap clicks or crawler reads.
At the limit, Rybbit stops collecting for the organization’s sites until the next month or an upgrade. invisit warns at 80% of your plan, records all of a first month over, and pauses new events at the limit only in a second month over in a row.
Which Rybbit features need Standard or Pro?
Every current Rybbit Cloud plan is Standard or Pro, and Pro adds session replay, unlimited sites and members, and two more years of data. invisit has one feature set on every paid plan, and its free plan has nearly every feature.
| Feature | Rybbit Standard | Rybbit Pro | invisit paid plans |
|---|---|---|---|
| Price at 1M events | $69 | $139 | $49 |
| Sites | 5 | Unlimited | Unlimited |
| Team members | 3 | 100 | Unlimited |
| Data kept | 3 years | 5 years | Forever |
| Session replays | No | Up to 10% of the events | No |
| Heatmaps | No | No | Yes |
| API requests a day | 5,000 | 25,000 | REST API |
| Revenue | No | No | From your provider |
| Import history from | Last 24 months | Last 60 months | No import |
Funnels, goals, journeys, retention, sessions, user profiles, web vitals and error tracking are on both Rybbit plans (Rybbit’s billing). New Rybbit sites need a trial or a plan to collect anything: sites created since February 13, 2026 in an organization without one record no data. The trial is 7 days, once per organization; invisit’s is 14 days on a monthly plan, with a free plan after it.
Can you self-host Rybbit instead of paying for Cloud?
Yes. Rybbit is open source under AGPL-3.0, and invisit is hosted only: its tracker and SDKs are MIT, but its dashboard and backend are not available to run yourself.
Rybbit’s quick start asks for a VPS with at least 2 GB of RAM and a domain with HTTPS, and installs Rybbit with Docker on ClickHouse. A self-hosted instance has session replay without a plan gate, no replay quota, no API rate limits and unlimited imports.
Some of Rybbit stays on Cloud. Its self-hosted vs Cloud table leaves web vitals, email reports, Search Console, the Bots page, the Pages view and custom SQL dashboards out of self-hosted Rybbit, and a self-hosted instance sends no email, so there are no invitations or password resets. You take on the server: updates, backups and ClickHouse.
Does Rybbit need a cookie banner, and how does it tell visitors apart?
Neither Rybbit nor invisit sets cookies. The difference is how long a visitor stays recognisable. Rybbit’s site settings say its user ID is a hash of IP address and user agent, which links a visitor across days and “may be considered storing personally identifiable data under GDPR”. Turning on User ID Salting makes the hash change daily, at the cost of user history past 24 hours and a retention report Rybbit calls effectively useless.
Salting is off for a new site: the setting defaults to false in Rybbit’s schema, although Rybbit’s features page says IDs are re-salted daily. invisit’s hash rotates every day for every site, so nobody is followed from one day to the next.
Signed-in users differ too. rybbit.identify() keeps the ID you pass in localStorage, and Rybbit’s identify docs say custom user IDs are not hashed. invisit hashes the id you pass to identify() and keeps only the hash, and redacts email addresses in property values. invisit can also honour Global Privacy Control and Do Not Track, and has an optional consent mode (privacy guide).
Rybbit Cloud stores analytics on Hetzner in Germany and Finland, and lists Cloudflare for CDN, DNS and object storage, including session replay recordings (Rybbit’s subprocessors). invisit stores everything in Falkenstein, Germany, with Cloudflare in front for DNS and CDN, and a DPA is available.
Rybbit session replay vs invisit heatmaps
Rybbit records sessions and has no heatmaps; invisit has heatmaps and records no sessions. Which one you want depends on whether you watch individual visitors or read a page’s clicks in aggregate.
Rybbit’s session replay is a Pro feature on Cloud. It records mouse movement, clicks, scrolls and navigation, masks inputs by default, and lists clicks and rage clicks on a timeline beside the player. Each month it records up to 10% of the plan’s event limit, 100,000 replays on a 1M plan, and Pro trials of 500K events or more leave replay out. Recording loads replay.js, which measured 81,673 bytes gzipped on November 16, 2026, on top of the 11,767-byte script.js.
invisit’s heatmaps record clicks, rage clicks, dead clicks and scroll depth over your real pages, on every plan and never billed. Each visit’s pages and events stay in order at any age, in the dashboard, the API and MCP (visits), with no recording of the screen.
Both track JavaScript errors. Rybbit groups them by message with stack traces and counts them per session (errors). invisit groups them, ties them to visits and gives them their own allowance, 5,000 reports a month on Free, that never counts as events (errors).
Does Rybbit show AI assistant traffic and crawler reads?
Both do, in similar ways. Rybbit’s Channels report has an AI channel and a Paid AI channel, with referrers from ChatGPT, Claude, Gemini, Perplexity, Copilot, DeepSeek, Grok, Mistral and others (Rybbit’s list). invisit recognises 19 assistants by referrer and utm_source in its own AI tab (how it works).
Rybbit Cloud’s Bots page shows which AI agents and crawlers read which pages, split into agents, answer engines and training, and how many visits each operator sent back. It counts only bots that run JavaScript, unless your server reports the rest through Rybbit’s tracking API, and it needs Block Bot Traffic on. invisit records crawler reads server-side with @invisitdev/server, including user-triggered fetches like ChatGPT-User. Neither bills them.
When Rybbit is the better choice
Rybbit is the better pick in these situations, and some are things invisit does not do at all.
- You need to self-host. invisit has no self-hosted edition.
- You want session replays. Rybbit Pro records them on Cloud, and self-hosted Rybbit has no quota; invisit has no recordings.
- You want product analytics for signed-in users: retention cohorts, user profiles with traits, and user journeys. invisit has no retention cohorts.
- You want feature flags, experiments or custom SQL dashboards in the same tool.
- You are leaving Plausible, Umami or Simple Analytics and want your history: Rybbit imports all three on paid plans.
- You pay yearly, or run around 5M events: Rybbit Standard is $149 at 5M against $199 on invisit.
Switching from Rybbit to invisit
Run both side by side before you remove anything. Both trackers are cookieless and can sit on the same pages.
- Install invisit next to the Rybbit script and leave Rybbit running for two to four weeks, so you can compare the numbers.
- Replace each
window.rybbit.event()call withtrack(), as below. invisit event names use letters, digits and_ . : -, up to 64 characters, with up to 32 properties; Rybbit names can run to 256 characters, soSignup Button Clickedbecomessignup_button_clicked. - Replace
rybbit.identify()withidentify(). Traits have no equivalent; put the ones you need on events as properties. - Replace
rybbit.error()withcaptureError(), and turn on error tracking for the site. - If you proxied Rybbit’s script, the
@invisitdev/serverproxy does the same for invisit. - Keep your Rybbit history before you cancel, because invisit cannot import it. On Cloud, Rybbit’s API returns events and sessions; self-hosted, your history is already in your ClickHouse database.
| Rybbit | invisit |
|---|---|
script.js?siteId=123 in the script URL |
init({ site: '<site id>' }) |
window.rybbit.event('signup', { plan }) |
track('signup', { plan }) |
data-rybbit-event="signup" |
data-invisit-event="signup" |
data-rybbit-prop-plan="pro" |
data-invisit-prop-plan="pro" |
window.rybbit.identify(user.id, traits) |
identify(user.id) |
window.rybbit.error(error) |
captureError(error) |
// Rybbit: the script tag defines window.rybbit
window.rybbit.event('Signup Button Clicked', { plan: 'pro' })
window.rybbit.identify(user.id, { plan: 'pro' })
// invisit
import { identify, init, track } from '@invisitdev/web'
init({ site: '<site id>' })
track('signup_button_clicked', { plan: 'pro' })
identify(user.id)See identifying signed-in users for when to call identify(), and connect your payment provider to see revenue, which Rybbit does not track.