Targets and returning users
Set the target an adoption check aims for, read who came back, and know what each verdict means.
An adoption impact check needs a bar to clear. You set two numbers: the share of active people who should use the feature, and the days they get to do it.
Every adoption check has a target: the share of active people who should use a new feature, and how many days they have. When the days are up, Needlr tells you if the feature cleared it. It also shows who came back, which tells you if the feature became a habit.
Think of a reading challenge
A teacher says: “By the end of the month, 1 in 5 students should borrow a book from the new class library.” That's the target. It's set on day one, before anyone borrows anything.
At the end of the month, the teacher counts. But the teacher also notices something else: some students came back for a second book. Those are the readers. One borrowed book is curiosity. Coming back is a habit.
Set the target
- 1On your project page, press + New impact check. Under What did you ship?, give it a name and pick Adoption.
- 2Set the Release date.
- 3Next to Target:, type the share in the first box: the percent of active people who should use the feature. Any number from 1 to 95. It starts at 20.
- 4In the second box, type the number of days they have, counted from the release date. Any number from 7 to 60. It starts at 30.
- 5Finish the rest as usual (how), then press Start measuring.
Under the boxes, Needlr says the day it will give its verdict. If the window is already over, it says the answer comes as soon as you start.
The first 100 people
Needlr counts every hour. But it shows no share until 100 active people are in. Before that, the page says how many it has so far, and that the first read appears after 100.
Why? With 8 people, one more person using the feature moves the share by more than 10 points. That's noise, not a result.
If the window ends with fewer than 100 active people, there is no verdict. The impact check stops, and the page says there was too little to go on. A longer window or a wider group of active people helps next time.
While it runs
After 100 people, the page shows the share so far, like “12 of 30 days in. 23% of active people used it so far: on pace for the 20% target.” In your project list it reads like 23% so far.
The verdict comes only once, when the window ends. Early shares usually look too good: as new people arrive who haven't tried the feature yet, the share tends to fall.
Who came back
Two numbers tell you if people kept using it:
- Came back: of the people who used the feature, the share who used it on 2 or more days.
- Came back next week, in the weekly table: of the people who used it in one week, how many used it again in the following week. For example 41% (38) means 38 people, which is 41% of that week's users.
- Weeks are 7 days long, counted from the release date. A day is a UTC day.
- A week's Came back next week says not yet known until the next week is over.
What the verdict means
Needlr works out a likely range for the true share, at the confidence you chose in step 3 under How sure do you want to be? Then it compares the range with your target.
The whole likely range is at or above the target. People use it. Keep investing, and watch the weekly table for people coming back.
The whole likely range is below the target. Often the feature is hard to find, not unwanted. The page has a Set up a split test button to test how people discover it, with the same KPI.
The range covers the target, so the numbers can't tell which side it's on. If the share was still climbing in the last weeks, it's on its way.
If the KPI stops firing while it runs, the impact check goes On hold instead (more).
Common mistakes
- Setting the target after launch. Pick it first. Otherwise you'll pick a number you already know you hit.
- A window that's too short for a weekly habit. If people use the feature once a week, 7 days can't show anyone coming back. Give it at least a few weeks.
- Calling it early. The share in week one is almost always higher than at the end. Wait for the verdict.
- Only looking at the share. A feature many people try once and never touch again is not a success. Check Came back.
Words to know
- Target
- The share of active people who should use the feature, within a number of days.
- Window
- The days from the release date to the verdict.
- Came back
- Used the feature on 2 or more days.
- Likely range
- Where the true share most likely is, at the confidence you chose.
What the video says
An adoption impact check needs a bar to clear. You set two numbers: the share of active people who should use the feature, and the days they get to do it.
Set the target before launch, from a similar feature or the business case. Anything from 1 to 95%, over 7 to 60 days. A target picked after the data proves nothing.
Needlr counts every hour, but shows no share until 100 active people are in. A percentage of a handful is noise. Under 100 when the window closes means no verdict.
Then the habit question. Came back counts users who used it on two or more days. The weekly table shows who used it again the next week, once that week is over.
When the window closes, you get one call. Target reached: the whole likely range is above your target. Below target: all of it is below. Near target: the range straddles it.
Reached? Keep investing. Below target? It's usually discovery, not demand, so split test how people find it. Hitting the target earns attention. Returning users prove the value.