A daily check on every client site's key flow
Clients find out their booking form is broken from their customers. Set up one scheduled test per site, get told when it fails, and send the client the proof.

The call you do not want#
A client rings on Monday. Bookings stopped on Friday night. A plugin update, an expired mail key, a form that no longer submits. Their customers noticed first, and now it is your problem.
Uptime monitoring would not have caught it. The site was up the whole time. The home page loaded; it was the last step of the form that did nothing.
What catches it is someone going through the flow every day. This post sets that up once per client, so that you hear about it first, with a recording.
One test per site: the flow that makes money#
Do not start with a suite. For each client, pick the single journey that pays for the site:
- A clinic or a restaurant: submit the booking form and see the confirmation.
- A lead-generation site: send the enquiry form and receive the confirmation email.
- A shop: add a product to the cart, reach the checkout and see the right total.
Write it the way you would explain it to a new junior:
Open the booking page and fill in the form as a new patient,
using this run's email address and the name "Test Patient".
Choose the first available appointment and submit the form.
Check that the page says the request was received.
Check that the confirmation email arrives and names the appointment time shown on the page.The last line matters. With Give this test a mailbox turned on, the run gets its own email address, so "the confirmation email arrives" is checked for real, and it is checked against what the page showed. See Mailbox, phone number and endpoints.
Run it every morning#
On the test's page, under When it runs, add a schedule: the days and the time, in your time zone. The default is every day at 02:00, and a test can have up to 5 schedules. See Schedules.

Pick a time before the client's day starts. If it fails at 6 am, you have fixed it or warned them before the first enquiry would have been lost.
After the first passing run, the tester replays the path it learned, so a healthy site is checked quickly. When the site changes, it explores again.
Hear about failures without watching#
You do not open a dashboard every morning. Nobody does.
- Email. A scheduled run that fails, is blocked or is inconclusive emails the workspace's owners and admins, with what was seen and a link to the run.
- Slack. Connect one channel and choose which projects post to it. A failure arrives with what was seen and a link to the proof.

Set-up for both is in Email and Slack.
Send the client proof, not a paragraph#
When something breaks, the usual message to a client is three sentences of reassurance. A link is better.
On any result, Share creates a read-only link that works without an account, for 7, 30 or 90 days. The client sees the checks and the values they were decided on, the replay of each step and the video. They see nothing else in your workspace. See Share links.

It works in the other direction too. A passing result from this morning, sent at handoff or with the monthly invoice, shows the client that the flow they pay you to look after was checked today.
Two things to know before you promise anything:
- The link shows your workspace's name. Name the workspace after your agency. There are no white-label or branded reports.
- For a client who wants to look whenever they like, invite them as a Viewer: they see results and can change nothing. Inviting people is on paid plans. See Team and roles.
One project per client#
Keep each client in its own project, so their tests, results and usage stay apart, and so a Slack channel can follow only the clients that matter to it.
If a client should pay for their own testing, give them their own workspace instead. Each workspace has its own plan and billing.
What it costs to run#
A scheduled run counts like any other: one test, run once, is one run. A daily check uses about 31 runs a month, so divide a plan's monthly runs by 31 to see how many client flows it covers. Runs that a site blocks are not counted.
| Free trial | Starter | Team | Growth | |
|---|---|---|---|---|
| Price | Free | $49 a month | $149 a month | $299 a month |
| Runs included | 10, once | 100 a month | 500 a month | 1,200 a month |
| Extra runs (packs of 50) | — | $30 | $20 | $15 |
| Web runs at a time | 1 | 5 | 10 | 20 |
| App runs at a time | 1 | 1 | 3 | 10 |
| Android and iOS app tests | Yes | Yes | Yes | Yes |
| Members | 1 | 5 | 25 | 100 |
| Projects | 1 | 2 | 5 | 15 |
| Tests | 20 | 300 | 1,000 | 5,000 |
| Assistant drafts a day | 40 | 300 | 300 | 300 |
| Phone number and endpoint calls | — | Yes | Yes | Yes |
| Invite people | — | Yes | Yes | Yes |
Some client sites sit behind bot protection that refuses cloud browsers. Those runs come back as Blocked. The fix is to allow our addresses in the client's firewall or CDN: see Network and allow-listing.
Start with your most nervous client#
Pick the client who would ring first. Write their one test, schedule it, and connect the alerts. The first morning it catches something, you will know what the other sites are worth.