Monthly Active Users

For a long time the number companies reached for to describe their size was cumulative signups — every person who ever pressed the register button, added up. It only ever goes up, which makes it comfortable to put in a pitch deck.

The trouble is that a shut-down product reports the same figure as a thriving one. A product claiming 500,000 registered users would turn out to have had 4,000 people show up this month, and investors started attaching a follow-up question: how many of them use it now? Cumulative signups have no answer to that.

So the practice became to fix a window and count only the people who actually showed up inside it. Monthly active users (MAU) is that count with the window set to one month. A month is the window that survived because anything shorter drops the people who stop by occasionally, and anything longer keeps people who have already left on the books for too long.

Fix a one-month window and count the distinct people who used the product inside it. Someone who came back twenty times counts as one, and so does someone who came once.

Two teams can pull different numbers off the same product because they settled the following four questions differently. Write all four down before publishing a number.

1. What counts as using it — including people who only opened the app versus only those who did something moves the result by around 20 percent. Pagedeck counts an account only if it opened a document. Someone who tapped a notification and left three seconds later is excluded.
2. Where the window starts and ends — a calendar month from the 1st to the last day and a rolling 30 days ending today give different answers. With calendar months the count restarts from zero every 1st, so the beginning of each month always looks steep on a chart.
3. What counts as the same person — accounts and devices are not the same unit. One person on a laptop and a phone is two under device counting, and someone with a work login and a personal login is two even under account counting.
4. Who gets excluded — leave employee accounts and automated accounts in and the number carries that much padding. At a 45-person company, those 45 are riding along in every reading.

Once the four are settled the number itself falls straight out of the logs. When a definition has to change, recompute the historical values under the new definition as well. If only the month of the change moves, everyone reads a measurement change as a product change.

There's no threshold that makes the raw figure good or bad. Eighty thousand is large for a collaboration tool sold to teams in one country and small for a messenger serving the world. Importing an industry average and grading yourself against it is the most common error here.

Build the benchmark from three places instead.

1. Your own past values — compare against last month and last quarter. For a seasonal product, put the same month of last year alongside it too.
2. Before and after a change — compare the month you shortened signup against the month before it. The comparison has to be that specific before you can say a change moved the number.
3. A figure derived backwards from the plan — work out what this number has to be for next quarter's revenue plan to hold, then look at the gap against today.

Left on its own the number invites a particular misreading, because a person who showed up once in the month and a person who showed up every day both count as exactly one. That's why teams read it next to daily active users (DAU) and weekly active users (WAU). A weekly count above half the monthly one means hardly anyone is an occasional visitor; a weekly count under a quarter of it means most of the base stops by once or twice a month and no more.

The direction of travel matters as much as the level. If the monthly count climbs from 8,400 to 12,000 while the daily count only moves from 900 to 1,100, a lot of people arrived and most of them looked once and left. Stickiness is that same relationship expressed as a ratio.

"We're at about 80,000 monthly actives, growing 12 percent a quarter."

Communicating the product's current size and speed to people outside the team.

"Does this number have internal accounts stripped out?"

Checking whether the definition is written down anywhere and whether the figure is padded.

"Don't just show the monthly number — put the weekly one up next to it."

A month-long window can't tell you how often people come back.

Pagedeck uses this number only in investor updates and company decks. The figure permanently pinned to the team's meeting dashboard is the weekly one. A monthly window takes a month to reflect anything, so it can't answer whether something changed this week worked.

Last quarter Mia changed the definition. Until then every account that logged in was counted, and 14 percent of those were accounts that arrived through a notification and did nothing at all. Switching to accounts that opened at least one document dropped the figure from 98,000 to 84,000. Reporting it to Grace, Mia led with the fact that the measurement had changed rather than the product, and included the previous four quarters recomputed under the new definition.

Treating it as a synonym for total registered users is the most frequent mistake. Both count people and both show up in the same slide, so they blur together easily. But registrations only ever accumulate while this one is recounted from scratch every month — if nobody shows up, it is zero. Mixing the two is how a team ends up saying "500,000 users" without noticing that 4,000 people actually use the thing.

Reading it as better-when-higher is the other one. It describes size, so that's a natural assumption. The catch is that advertising can buy it. Anyone who shows up inside the window counts, so pushing 30,000 extra people through the door lifts this month's figure by 30,000. If those people don't return next month the figure settles right back down and only the ad spend remains. Set this number as the goal and the team starts buying traffic instead of fixing retention.