Everything you need to find your site's limit
Describe a visit, choose how the traffic arrives, and see which page or step gives out first. Every session runs in a real browser on our machines.
Describe one visit, then send hundreds.
A scenario is the path a visitor takes: open a page, type, select, click. Write it once and every session follows it in a full browser.
-
01
Write the steps
Add the actions in the editor, in the order a visitor takes them. Labels and selectors come from your own pages.
-
02
Or paste a Playwright script
Supported calls become saved steps you can edit. A ready prompt gets an AI assistant to write one for your pages.
-
03
Validate it, free
One browser session runs the steps before any load test. If one fails, you see which and why, and fix it first.
The Playwright compatibility reference lists every call that converts and how selectors are matched.
Shape the crowd the way it really arrives.
Pick how many sessions to send, the window they arrive in (ten seconds to an hour) and the shape of the arrivals.
Constant load
Starts spread evenly across the arrival window.
Rising load
More sessions start later in the run, like a sale warming up.
Early spike
Most arrive at the start and then taper off, like a newsletter landing.
Random
The arrival rate changes from moment to moment.
Custom
Draw your own day with 24 points: a lunch rush, an evening peak.
Locations
Pick from 17 locations; each session starts from one of them at random.
Mix several scenarios in one test and split the traffic between them, for example 70% browsing and 30% search. Configure traffic explains each setting.
Tests that stop when your site gives up.
A load test should find the limit, not take the site down for the afternoon. These are on for every run.
The run winds down on its own
Pick the share of sessions allowed to fail. Cross it and the run stops by itself.
- Checked once 10 sessions have finished
- A separate threshold for each test
- Sessions already in progress can finish
Only sites you have proved are yours
Add a DNS TXT record or a meta tag. Nothing to install on your servers.
Stop any time
One button ends the run. Sessions already running finish or time out.
A cap on sessions at once
Accounts start at 200 concurrent sessions. Support can raise it.
Validated before load
Scenarios that need checking are validated before the first load session starts.
A fresh browser each time
No cookies, cache or login carried over from one session to the next.
Testing a live site? The production safety guide covers permission, monitoring and cleanup.
Not “it was slow”, but which step, and why
Every run ends in a report you can read in a minute and hand to whoever has to fix it.
- 1
Every step, timed
The step that slows down is its own line, not lost in an average.
- 2
Every failure, with its status code
Failed requests and HTTP errors, and the grouped errors behind failed sessions.
- 3
Web Vitals per page
Load time, TTFB and LCP, rated against web.dev's limits, as a visitor's browser sees them.
The report also has the XHR and fetch calls your pages make, results by location, and a live chart of concurrent users and failed steps while the run is going.
Catch the release that got slower.
One test is a snapshot. Run the same one on a schedule and a slow deploy shows up the morning after.
Every run kept, with its settings
Daily, weekly or monthly runs
One link for your developers or client
Turn the link off and it stops working.
FAQs
Does UserLimit run my Playwright tests?
It converts them. Supported calls such as page.goto, clicks and typing become saved steps you can edit. Assertions, helpers and other test logic do not run. The Playwright compatibility reference lists what converts.
Is there a recorder?
No. Add the steps in the editor, paste a Playwright script, or copy a prompt that gets an AI assistant to write the script for your pages.
Does UserLimit test APIs?
No. Sessions run in real browsers, so UserLimit measures what visitors experience in the page. The XHR and fetch requests your pages make show up in the report, but there is no API-only load testing.
Can it test a site on a private network?
No. The browsers run in the cloud without an agent on your network, so the site has to be reachable from the internet. Each session also starts in a fresh browser, without your login.
How many sessions can run at once?
Accounts start with a limit of 200 sessions at once. Support can raise it, and can confirm the higher limit before you buy usage.
What does a test cost?
You pay for the time sessions use, in virtual-user hours: 300 sessions of 30 seconds each is 2.5 VUh. Validating a scenario is free.
Find yours on a quiet Tuesday, not on launch day
Start free. Your first test can be running in a few minutes.