Hear about it when a number breaks its pattern.

Pick something you already measure. ClickClacks learns what normal looks like for each day of the week, tells you when today falls outside it, and shows where the change came from.

Start tracking free All features
kinds of metric: pageviews, sessions, funnels, retention, sources
5
sensitivity levels, and no thresholds to set
3
previous same weekdays behind every expected range
6
between checks, for every monitor
5 min

What it does

You don’t set thresholds. Each monitor builds an expected range from the same weekday in previous weeks, checks it every five minutes, and fires when the value lands outside. A quiet Sunday isn’t an outage, and a flat Monday is worth knowing about.

Metrics

Watch what you already measure.

Daily pageviews and sessions, the conversion of any saved funnel, week-1 retention, and the events from each source. Before you start, the dialog says how much history the metric has to learn from.

  • Start from the Alerts page, or from a board tile with “Watch this…”.
  • Fired alerts land in the in-app notifications of everyone in the organization.

Baseline

Sensitivity, not thresholds.

The expected value is the median of the same weekday over the previous six weeks, counted up to the same minute of the day. The range around it comes from how much those days varied. Pick how wide it is: Big changes only, Moderate or Everything unusual.

  • It fires on a rise as well as a drop.
  • With less than 42 days of history the range is held wider, and marked baseline still learning.

What changed

Where the change came from, with the evidence.

Each alert splits the gap by source, and by step for funnels: expected, actual, the gap and each one’s share of the drop. Every row links to the events or the funnel behind it. It’s a filtered count, no model and no weighting.

  • The headline names the source that holds most of the change.
  • Evidence opens Events filtered to that source and window, or the funnel.

Lifecycle

It closes itself when it’s over.

When the metric comes back inside its range, the alert resolves on its own. Acknowledge one to stop the reminders, mute a monitor to pause its checks, or change its sensitivity right from the alert.

  • One open alert per monitor, so a long dip doesn’t fire twice.
  • A timeline records when it fired, was acknowledged, re-checked and resolved.

Questions it answers

Ask it the way you’d ask a colleague.

How ClickClacks answers it

Pageviews are compared with the previous six Mondays, each counted up to the same minute of the day, so a slow morning isn’t measured against a full day.

−23.4%

Pageviews, daily · vs typical Mon

Who it’s for

Three ways teams use it.

  • Agencies

    Watch pageviews and the main funnel in each client organization at Big changes only, so you hear about a broken snippet or a dead landing page before the client does.

    Big changes only, per client

  • Indie developers

    You can’t watch a dashboard all day. Watch sessions and your trial funnel; when something moves, the alert already says which source it came from.

    Two monitors, Moderate

  • Product teams

    Turn a funnel’s monitor up to Everything unusual for launch week, then set it back to Moderate from the alert page when things settle.

    Adjust sensitivity

How it works

Under the hood

  1. 01Pick a metricStart from the Alerts page, or from a board tile with “Watch this…”. The dialog shows how much history the metric has.
  2. 02Choose how sensitiveThree levels, from big changes only to everything unusual. Nothing to tune by hand.
  3. 03We check every five minutesEach check compares the metric with the median of the same weekday over the previous six weeks, and the range those days allow.
  4. 04Read it in the appA fired alert lands in your in-app notifications and on the Alerts page, with what changed and a timeline. It closes itself when the metric recovers.

Spec sheet

from the product

Checks
Every 5 minutes
Baseline
Same weekday, previous 6 weeks
Expected value
Median
Spread
1.4826 × MAD, at least 5% of the median
Range
± 3, 2 or 1.2 × spread
Learning
Under 42 days, 1.75× wider
Thresholds to set
None
Delivery
In-app

Same events, different question

Works well with

All features →

FAQ

Alerts, in detail.

  • A fixed number is wrong most days: too tight on a Monday, too loose on a Sunday. Each monitor works out its own range from the same weekday in previous weeks, and you only choose how wide it is.

  • Pageviews, sessions and source volume are counted up to the same UTC minute on each of the previous six matching weekdays. Funnels use finished conversion windows ending yesterday, and retention uses the last finished weekly cohorts.

  • The metric has less than 42 days of history. It can still be watched, as long as there is at least one earlier matching weekday, but the range is held 1.75 times wider until the history fills in.

  • Yes. A monitor fires when the value leaves its range in either direction. A rise shows in green, a drop in red.

  • Acknowledging stops the reminders but keeps the alert open until the metric recovers. Muting pauses the monitor’s scheduled checks.

  • Not yet. Alerts are delivered in-app, to the notifications of everyone in the organization. The new alert dialog lists email as coming soon.

Start hearing your clicks.

One script tag. Your first report in minutes.

Start tracking free Sign in

4.3 KB · pageviews, clicks and scroll depth, automatically