The ratio of daily to monthly active users. Multiplied by 30, it gives the number of days a month the average user comes back.
In practice
The one thing worth remembering about this ratio is that it translates: multiplied by 30, it gives the number of days a month the average user comes back. A DAU/MAU of 0.2 is six days out of thirty — put that way, a team immediately knows whether it's good, because it knows what its product is for. That's also what makes the often-quoted 20% threshold close to useless outside the category it came from, consumer social: a payroll tool at 0.05 isn't failing, it's being used when payroll runs. Two traps to know. The ratio is an average over a population that is often bimodal, so it rarely describes anyone real. And it climbs mechanically when casual users leave — hence the rule never to read it without the MAU count next to it.
The formula
DAU/MAU = (average daily active users over the month) ÷ (monthly active users) — and the ratio × 30 = days used per month
- DAU
The average of your daily active counts across the month, not one good Tuesday. A single day is a sample of one, and the day you happen to look is rarely representative.
- MAU
Unique users active at least once over the month, each counted once however often they came. Counting them per visit turns the ratio into nonsense.
- Active
The same definition on both sides of the division, and something more demanding than opening the app. A ratio built on "opened" measures notifications, not usage.
The multiplication is the part worth remembering: a ratio of 0.2 is not an abstract score, it is six days out of thirty. Stating it that way is usually enough to settle whether the number is good, because a team knows what its product is for.
Worked example
A ratio that describes nobody
- 50,000 MAU, DAU averaging 10,000 over the month. DAU/MAU = 0.20, so the average user shows up six days out of thirty.
- Split the base and there is no such user. 8,000 people open it about 22 days a month; the other 42,000 open it about 3.
- Check the arithmetic: (8,000 × 22) + (42,000 × 3) = 302,000 user-days, ÷ 30 = about 10,000 DAU. The same 0.20, from two populations that share nothing.
- The two groups need opposite decisions. The 8,000 are a daily habit to protect and to learn from; the 42,000 are either a weekly product misread as a daily one, or a group that never found the habit at all — and only looking at what they do differently answers which.
Same lesson as cohort analysis, on a different axis: cohorts split by when people arrived, this splits by how often they come. Both exist because an average over a mixed population describes the average and nobody in it.
Orders of magnitude
- The threshold most often quoted is 20% for consumer social products, with 50%+ treated as exceptional — figures that come from a category where daily use is the whole point. Quoting them at a product not meant to be opened daily is comparing ambitions, not performance.
- A payroll tool at 0.05 — a day and a half a month — is not failing; it is being used exactly when payroll runs. The first question is never "is the ratio high" but "should this product be opened daily at all", and answering it honestly retires the metric for a good share of B2B software.
- For a weekly-cycle product, WAU/MAU — weekly over monthly active users — is the same idea at the right grain and is far more readable. Reading a weekly product on a daily ratio produces a number that is always low and never actionable.
How to improve it
- Decide first whether daily use is the right ambition. Half the effort spent lifting this ratio is spent making people open something they had no reason to open, which shows up later as churn.
- Segment the ratio rather than averaging it, and look at what the frequent group does that the others do not. That behaviour, not the ratio, is the thing to design for.
- Read the ratio and MAU together, always. A rising ratio with a falling MAU is not an improvement: it is casual users leaving, which mechanically raises the average frequency of those who stay.
- Tie frequency to a real trigger in the user's week — a report that lands, a queue that fills, a teammate who acts — rather than to a notification. A trigger that exists outside your product survives someone turning notifications off.
In the Tour
Do you track a retention rate (D7/D30 or similar)?
- Yes, tracked and reviewed regularly20 pts
- We can pull it, but rarely look7 pts
- No0 pts
The Tour asks whether you track a retention rate at all, for 20 points. DAU/MAU is the companion figure: retention says whether people come back at all, this says how often. They fail in opposite ways, which is why neither replaces the other — retention can look healthy on users who return once a month, and the ratio can look healthy on a shrinking core of enthusiasts. A team answering 20 here and never having multiplied its ratio by 30 is holding a number it has not yet read.
Questions people ask
What is a good DAU/MAU ratio?
Multiply it by 30 and ask whether that many days a month is what your product is for. A messaging app at 12 days is struggling; an invoicing tool at 12 days is being used far more than its job requires. The benchmark lives in the product's purpose, not in a table.
Is DAU/MAU a retention metric?
Not quite — it measures frequency inside a month, retention measures survival across months. A cohort can retain beautifully at a low frequency, and a high ratio can sit on a base that halves every quarter. Read the cohort curve for survival and this for intensity.
Should I use WAU/MAU instead?
If the natural cycle of your product is a week, yes, and the choice matters more than the precision of either. Pick the grain that matches the rhythm you are actually asking people to adopt, then keep it — switching grains mid-year produces a step change that looks like a result.
Can the ratio rise while the business gets worse?
Routinely, and it is the trap to know. Lose your casual users and the ones left are by definition the frequent ones, so the ratio climbs on a shrinking base. That is why it is never quoted alone: the absolute MAU next to it is what says whether the climb was earned.