Resurrected User

When users were sorted into just two buckets, new and existing, there was nowhere to put someone who returned. An account dormant for three months opens again one morning and you have to pick. File it as new and you're counting a person you already acquired as an acquisition twice. File it as existing and the three missing months disappear from the record.

That ambiguity turned into a real problem wherever the numbers got used. A team reports 1,200 new users for the month, and only later does it surface that 300 of them were old accounts. The acquisition cost computed by dividing ad spend by 1,200 came out lower than reality, and next month's budget, set on that number, misses by the same margin. Nobody in the room disagrees that the user count went up; the disagreement starts the moment someone has to explain why.

Resurrected user is the third bucket. Separate the people arriving for the first time, the people who never left, and the people who left and came back, and the same increase starts telling you what produced it.

A resurrected user gets counted when an account previously treated as gone shows up again. Three things have to be written down before any number comes out.

1. What counts as gone — how many days of silence before an account is treated as departed. Match it to how often the product is meant to be used. Thirty days is reasonable for a collaborative document tool people open on working days, but apply 30 days to a tax filing service used once or twice a year and nearly every user registers as departed every month. Pick the window once and leave it alone, because moving it from 30 days to 60 changes the count without anything changing in the product.
2. What counts as coming back — whether a login is enough or whether the core action is required. Pagedeck uses opening or editing a document. Count an account that clicked through an email and only looked at a screen, and the campaign that sent the email looks better than it was.
3. Split the buckets three ways — new, existing, resurrected. With that split you can decompose how monthly active users (MAU), the count of people who used the product at least once in a month, actually moved. Going from 8,400 to 8,900 resolves into 1,200 new plus 300 resurrected minus 1,000 departed.

It also gets read as a rate. If 4,000 accounts were sitting in the departed bucket and 300 of them returned this month, the resurrection rate is 7.5 percent. That number is managed separately from new acquisition, because the pool consists of people who have already used the product once, which makes calling them back far cheaper than acquiring someone new.

It's counted apart from the retention numbers too. Methods that look at a single point in time, like N-day retention, the share of a cohort that returned on exactly day N, leave out anyone who came back afterward. Without a separate count, these people appear in no table at all.

The count earns its keep in the meeting where somebody has to explain why the user number moved. Four uses come up most often.

1. Decomposing the change — knowing the base grew by 500 doesn't determine what to do next. Split into 1,200 new, 300 resurrected, and 1,000 departed, and it becomes clear whether the thing to work on is acquisition or retention. The same 500 built from 200 new and 400 resurrected is a different situation calling for a different quarter.
2. Correcting acquisition cost — ad spend has to be divided by genuinely new users. Mix resurrections in and the cost per acquisition reads low, and a budget increased on that reading misses next month.
3. Measuring win-back work — the output of a reactivation campaign, the set of messages sent to wake dormant users, shows up as this number. Twenty thousand emails producing 300 returns is 1.5 percent, and that figure decides whether another round runs.
4. Finding out why people left — returners are the only group that knows both why they left and what brought them back. When a team works on reducing churn, the phenomenon of users or customers leaving, the people who already left are unreachable and the people who stayed never left. Asking the 300 who returned what changed gets an answer faster.

"Of this month's 1,200 new users, 300 are resurrections. We need to recalculate cost per acquisition."

Returning accounts were being counted as people acquired for the first time.

"Resurrection rate came in at 7.5 percent, up two points from the last round."

The win-back message performed better than the previous one.

"Let's start the interviews with the people who came back. They're the only ones who can tell us why they left."

Using this group as the sample for understanding departures.

At Pagedeck, Ravi sends a feature announcement each quarter to accounts that have been silent for more than 30 days. Last quarter 300 out of 4,000 opened a document again. Mia looks separately at what those 300 did once they were back. More than half of them opened a feature they had never used before, and that feature became the first line of the next quarter's announcement email.

The most common mistake is reading a large resurrection count as a good sign. Coming back looks like evidence that the product improved. But for that number to be large, a lot of people have to be available to return, which means a lot of people left. If 300 returned in a month when 1,000 departed, the revolving door is spinning and the total shrank. Resurrections belong on the same line as departures, always. Report one without the other and the movement can be narrated as good news in either direction. The pairing is also what tells you whether the win-back work is worth continuing: a rising resurrection count against a flat departure count is progress, and against a rising one it is treading water.

Treating one return as a save is the other one. The act of coming back looks like a change of mind. But a returning account is likelier to leave again than a first-time one. It already has a history of not finding a reason to stay, and whatever working rhythm it had was broken during the absence. Of 300 returners, a month later having only 60 left is an ordinary outcome. So a resurrection is handled as a starting line, not an arrival. Many companies run these accounts through onboarding, the process of helping a new user get settled, a second time.