A dashboard answers every question you ask it, and since observations it also records on its own what changed. Someone still has to look, though. Anyone who wants to know on Monday morning whether last week was normal clicks through load times, errors and traffic, application by application, and hopes nothing slips past.
That is why, since this Monday, the weekly overview is live: one mail per organization and week that tells you what changed, where to look first and which numbers stand behind it. Mondays at 9 a.m. by default, for every member of your organization.
This is what it looks like. The mail below is an example from our test data, with applications and domains switched over to fastmon.eu.
The mail as it arrives, scroll through it: one sentence on top, then one row per application and what is shifting over several weeks.
One sentence first
The mail is a signal, not a report. It opens with one sentence that says what deserves attention this week: that a domain stopped sending data, that a new error showed up, that a blog suddenly had twice the pageviews. Read only that sentence and you already know whether there is something to do today.
A quiet week is a short mail. And an organization whose applications did not collect enough data during the week gets none at all.
One row per application
Below it sits a table with one row per application. An application is what you set up in fastmon, with all of its domains: a shop on www and m. is one row, not two with half the traffic each.
Every row has four numbers:
- Score, the Experience Score from 0 to 10 with a grade
- Visitors
- Load time, the LCP at the p75
- Errors, the first-party error rate: the share of pageviews with a JavaScript error from your own code. Errors from embedded third-party scripts do not count
Under each number, a second line gives the change against the week before, as short as it can be: “-42 %” for visitors, “24 % slower” for load time, “was 8.1” for the score. If the change is within the noise, it says “stable”. More visitors are neither good nor bad, which is why traffic never gets a colour.
Three rules keep the table honest. Bots count nowhere, not even in the score. Development and staging applications do not appear at all, because a load test on staging is worth nobody’s mail. And an application with fewer than 50 pageviews in the week gets “insufficient data” instead of a rating.
At most three findings per application
After the table come detail blocks for up to three applications, and only for those where something stood out. Urgency decides which ones: first outages such as a domain that stopped reporting, then new or sharply rising errors, then the largest drop in score, then traffic. Each block holds at most three findings, the most severe first, and an improvement always takes the last slot so good news does not get lost.
Every finding gives the numbers of both weeks and leads straight to the matching dashboard page.
The most important finding is the least spectacular one: a domain stopped reporting. If a domain had at least 50 pageviews last week and not a single one this week, the snippet usually went missing in a deploy. In the total of an application with several domains this barely shows, so the weekly overview checks every domain on its own and ranks this case above all others.
To keep the mail from turning into noise, a few rules apply:
- A new error only becomes an alarm once it has reached at least 0.5 % of the week’s pageviews. Below that it reads as needs improvement.
- An error that someone on your team marked as seen in the dashboard is not reported as new again.
- A regression that prints the same rounded number in both weeks, such as 1,180 and 1,245 ms, which both read 1.2 s, does not show up.
Releases: since, not because
If you send release markers to fastmon, the weekly overview says how many releases an application had in the week and which one was the last. That line is there even when nothing moved, so you can see that the markers arrive.
It also measures the newest release that has the same amount of traffic before and after it: 48 hours on each side where that fits, 24 otherwise. If LCP, INP, TTFB or CLS get clearly worse afterwards, from 20 % for the timings, a finding with the version number appears. It is always worded in time, never as a cause: “since release 2026.10.4”, not “because of”. A Friday deploy shares its hour with campaigns, CDN changes and the weekend, and a mail cannot tell them apart.
Over several weeks
Comparing with the week before has one weakness: a problem that started three weeks ago and stayed looks unchanged from week to week. That is exactly when the mail would go quiet.
Two things prevent that. The trends from observations have their own section, “Over several weeks”, for every application, not just the three with a detail block. A regression stays there for as long as it is open, an improvement appears once, in the week it opens. And a metric that sat in the poor range this week and the week before is named in the opening sentence as “still”, even if nothing about it moved.
Every detail block ends with a small table of load time, response to clicks, layout shifts and error rate, with up to four previous weeks as columns. An arrow means: this move matters. A circle means: too small to mean anything. Ratings follow Google’s thresholds for Core Web Vitals.
The same week in the dashboard
Everything in the mail is also in the dashboard, on a page of its own: Weekly overview, in the sidebar under Analytics. The page uses the same sentences as the mail, so the two never say different things, but it shows eight metrics instead of four and a window of its own with every detail for each application.
The mail itself loads nothing: the logo is an embedded SVG, there is no external image, no script and so no tracking pixel either. It arrives in the language set in your account.
Unsubscribe in one click
The weekly overview is a service mail about your own data, not advertising. That is why it is on for every member of an organization, whatever their role, and every member can switch it off for themselves at any time:
- in the dashboard under Account, Notifications, at the weekly overview switch. That is also where you set weekday and hour, in your own time zone.
- through the Unsubscribe link at the end of every mail.
Every mail ends with both routes.
The link opens a confirmation page, and only the button switches the mail off. That is not red tape: security filters in company mailboxes open links in mails on their own, and a link that unsubscribed on opening would sign out people who never clicked. Mail clients with built-in one-click unsubscribe use the same route and need no confirmation. Unsubscribing only affects the weekly overview, every other mail stays as it is.
Tell us what is missing
The weekly overview is built to rather leave a finding out than send a wrong one. If you miss one, if one was too loud or a sentence left you puzzled, write to us through “Send feedback” in the dashboard. We learn the most from the weeks in which the mail and your gut feeling do not agree.