TractionScout
← Back to the blog
Honest growth metrics

DAU, WAU, MAU and stickiness: which source should count your users?

Definitions you can defend, what should count as active, and why two tools' users never add up

7 Nov 2026 · 5 min read

Daily, weekly and monthly active users (DAU, WAU, MAU) count the distinct people who used your product in the last 1, 7 and 28 finished days. Stickiness is the average daily active users over those 28 days divided by MAU. The formulas are simple; what makes the numbers trustworthy is deciding which actions count as use and letting one source, not several, count the people.

The definitions

MeasureCountsWindow
DAUDistinct people with product useThe last finished day
WAUDistinct people with product useThe last 7 finished days
MAUDistinct people with product useThe last 28 finished days
StickinessAverage DAU over the 28 days ÷ MAUThe same 28 days

Two choices in that table matter more than they look.

Finished days. Today is never complete. A window that includes today undercounts every morning and recovers every evening, so it moves while nothing changes. Count only days that have ended.

28 days, not a calendar month. Twenty-eight days is exactly four weeks, with each weekday counted four times. A calendar month has 28 to 31 days and a different mix of weekends, which makes February look worse than March for no reason.

Reading stickiness

Stickiness says how often a typical monthly user comes back. At 20%, the average monthly user is active on about 6 of 28 days (0.2 × 28 = 5.6). At 50%, about 14.

What counts as good depends on how often your product is meant to be used. A team chat app should be used daily; an invoicing tool used weekly is doing its job at a much lower stickiness. Compare your stickiness with itself over time, and with how often a customer genuinely needs you.

What should count as "active"

This is where most active-user numbers go wrong. Someone is active when they used the product, not when something happened to their account.

Counts as activeDoes not count
Creating, editing or viewing the thing your product is forA renewal or invoice paid automatically
Completing a step toward a result, like a sign-up or an upgradeAn automated email being sent or opened
Navigating between product screensA background sync, API job or scheduled export
An account being created, updated or merged

Logins sit in between: a sign-in shows someone came, not that they did anything. If sign-ins are all you have, label the measure "signed-in users" rather than active users.

Which source should count the people

Most products have several tools that see users: web analytics, product analytics, the authentication system, billing. Each identifies people differently, and you cannot match them reliably: an analytics ID, an auth account and a billing customer are different keys.

So pick one population and keep it. A sensible rule: use the source that identifies the most people with product activity, such as your product analytics with a user ID, and fall back to your authentication system's sign-ins only when nothing else identifies people. Never add active users from two sources together. The result double-counts everyone both tools saw and is not the number of anyone.

Comparing periods

Compare each window with the same window one period earlier: this week's WAU with last week's, not with this month's MAU. And be careful at the start: a window that begins before you started identifying users is partial, so its comparison means nothing.

How TractionScout counts active users

TractionScout Performance for a sample product: the Engaged step with bounce rate charted, and session time and average session duration beside it

The Engaged step in the TractionScout demo workspace. The sample data identifies no users, so it shows engagement from sessions; with a source that identifies people, active users and stickiness sit here.

TractionScout puts daily, weekly and monthly active users and stickiness on each project's board, on the Engaged step, over the last 1, 7 and 28 finished days, each compared with the same window one period earlier.

The population comes from one source: of the sources whose identified people used the product in the period, the one identifying the most, such as the TractionScout tracker or SDK, or PostHog with a user ID. If no source identifies people, an account source such as Supabase sign-ins can supply it, and each figure says which population it counts. Populations are never added together, and a project with no measurable population shows no active users instead of zero.

Because every product names its events differently, Scout reads each event name with its source and aggregate counts only, never property values or page addresses, and classifies it as product use, conversion, navigation, automated, bookkeeping or cannot tell. Product use, conversion and navigation count as activity; automated and bookkeeping events do not. See Smart KPIs.

Questions

What is a good DAU/MAU ratio?

It depends on how often your product is meant to be used. Daily tools aim high; weekly or monthly tools are healthy at much lower ratios. Track your own ratio over time rather than comparing with other products.

Should logins count as active use?

Only if nothing better is available, and then call the measure signed-in users. A login shows someone arrived, not that they got value.

Can I add active users from my web analytics and my product analytics?

No. The same person appears in both under different IDs, so the sum double-counts. Choose one source and report its number.

Empfohlen

See active users and stickiness from one honest source

Daily, weekly and monthly active users on each product's board, with the population named.