Your funnel says that people sign up and few reach the screen you call "activated". That is a fact about where people stopped. It says nothing about what they came for, what they expected to happen next, or what they decided when it did not. Onboarding interviews fill that gap, and they work best when you stop asking about your product and start asking about their first week.
This guide is about onboarding to a product or service, not onboarding a new hire. It gives you three things to copy: a first-week timeline worksheet, an interview guide that walks the timeline from the decision to sign up to the moment the person got value or stopped, and a short table for matching the stories you hear to the product events you already track. It works for customers, trial users, new staff on an internal tool, and partners joining a programme. The text says "customer" for short.
The short answer: fix the step where expectation and experience split
Do not ask "why didn't you activate?" That question asks for a verdict, and a verdict is not a story. Ask them to rebuild the week, step by step, and look for the first step where what they expected and what they got came apart. That step is your candidate. Then decide what kind of gap it is: they were confused, they had no reason or prompt to continue, or the product could not do what they needed. Each kind has a different fix. Mix them up and you redesign a screen that was never the problem.
Interviews give you the "why". Your event data gives you the "how many". Use each for its own job. NN/g puts the split plainly:1. Your funnel already counts. The interviews explain.
Define "value" in the customer's words, not your metric
Your activation metric is your guess about what value looks like. "Created three projects" or "invited a teammate" is a stand-in. Nobody signs up to create three projects. Before you interview anyone, you need to know what the customer would call success.
So make that the first thing you ask. Not "what did you hope to get from us?", which invites a polite answer, but "think back to the day you signed up. What was happening in your work that day?" The Jobs to Be Done switch interview, as its originators teach it, starts from the2. That event is where value gets defined. Picture someone who signed up the morning after a double-booked Saturday, and someone else who signed up to "see what's out there". They want different things, even if both land on the same welcome screen.
Write each person's value statement in their own words at the top of their worksheet, and compare it with your metric afterwards. Where they differ, you have found something the dashboard could never show.
Who to interview, and when: sign-ups who succeeded, stalled and quit
Interview three groups, not one.
| Group | Who they are | What they teach you |
|---|---|---|
| Got value | Signed up recently and reached something they would call a win | What the working path looks like, and what they did that others did not |
| Stalled | Signed up, did some things, then went quiet but have not left | Where the path broke, and what they are still half hoping for |
| Quit | Cancelled, deleted their account, or told you it was not for them | What the product could not do, or what they decided it was not worth |
Do not interview only people who quit. You will collect complaints and miss what the successes did differently. Do not interview only the happy either: they cannot tell you about the gap. Qualitative researchers have a name for choosing on purpose:3, and one of its strategies is to pick a spread of contrasting cases rather than whoever is easiest to reach. Start with a few people in each group. Add more only while new interviews keep teaching you something about the same step.
Recruit real, recent sign-ups. GOV.UK's service manual is blunt:4. Colleagues and friends do not count. Long-time power users do not count for this study either, because they are no longer living their first week. Build the invitation list from your own records. The same guidance tells you to4, to identify the groups you need. Write the invitation so it does not hint at the answer. "We would like to hear how your first week went" works. "Help us fix our confusing setup" does not.
Follow your organisation's consent and privacy rules when you pull names from your records and when you record the call.
When to talk: while the week is still fresh
Talk to people as soon after sign-up as you can. Do not wait for a quarter's worth of data. The sooner you ask, the less of the first week the person has to reconstruct from scratch.
The switch-interview originators give a looser outer limit:2. Treat that as the ceiling, not the goal. Ask for the exact screen, the exact moment, and what they said to a colleague about it.
Reconstruct the first week, step by step
The method is a timeline. You are not asking for opinions about onboarding. You are asking for the story, in order. GOV.UK's guidance on interviews says to5. Its guidance on researching a user's experience asks you to find out6.
That gives you three questions for every step: what did you expect, what did you do, what did you get. Ask them in that order. Ask about expectation first, before the person has told you how the step turned out. The guide below follows that order.
Use the plain follow-ups the same interview guidance recommends, such as5 and "can you tell me more about…?". Do not suggest answers. If you catch yourself saying "was it because the setup was too long?", stop. You have just told them the reason.
Here is the interview guide. Copy it, cut what does not apply to your product, and keep the order.
Before you start (read aloud, in your own words)
- Thank you for making time. I want to understand your
first week with [product or programme], in your words.
There are no right answers, and nothing you say will
be taken as a complaint about you.
- With your permission I will take notes [or record].
Say if you would rather I did not.
1. The day you decided (5 minutes)
- Take me back to the day you signed up. What was
happening in your work or life that day?
- When did you first start thinking about changing how
you handle [the job]? What happened just before?
- What else did you look at or consider? What did you
almost do instead?
- What did you hope would be different a week later?
- On the day you signed up, what would a good outcome
have looked like? (Write the answer down as their
definition of value.)
2. The first hour (10 minutes)
- You signed up. What did you see first? What did you
expect to happen next?
- What did you do? Tell me about the last thing you did
in that first session.
- Tell me about any point where you paused. Where were
you? What was going through your mind?
- What happened next? What did you do after that (in
the product or somewhere else)?
3. Each step after that (10-15 minutes, repeat per step)
- Before the next step, what did you expect it to do?
- What was the next thing you tried? Why that?
- What actually happened?
- What did you think? How did you feel?
- Who else was involved?
- What did you do when that happened?
4. The moment of value, or the moment you stopped (5 minutes)
- When did you first feel it was worth it? What had just
happened? [or] When did you decide it was not for you,
or not now? What was the last thing you did?
- If you could change one step in that week, which one?
What would you want to happen instead?
5. Close (2 minutes)
- Is there anything I did not ask that I should have?
- Is there anyone else who signed up around then that
we could talk to?
- Thank you. Here is what happens with what you told me:
[...]
The first-week timeline worksheet
Fill in one worksheet per interview, in the order the person tells it. One row per step. Leave a cell blank rather than guessing.
| Day or moment | Step (what they did) | What they expected | What they got | What they thought or felt | Their words (exact phrase) |
|---|---|---|---|---|---|
| Before sign-up | What was happening that day | ||||
| Sign-up | |||||
| First session | |||||
| Step 3 | |||||
| Step 4 | |||||
| Value reached, or stopped |
At the top, record three lines: their definition of value in their words, their group (got value, stalled, quit), and the first row where expectation and result differ. After five or six interviews, lay the worksheets side by side. The rows where several people's expectations break are your shortlist.
This is the shape of an experience map, which GOV.UK says7. Yours is built from interviews instead of assumption, and that matters. NN/g says the data behind a journey map8.
Sort each gap: confusion, no reason to return or missing capability
Once you have the broken step, name the kind of break. Three kinds are worth separating. The first two borrow from the Fogg behaviour model, which holds that9. If they could not work out how, that is ability. If the person lacked a reason, or nothing nudged them to act when the moment came, that is motivation or prompt. The third kind is about your product, not the person: it could not do the job.
| Gap | What you hear | What it means | Where the fix lives |
|---|---|---|---|
| Confusion | "I didn't know whether to…", "I wasn't sure what that button did", they did something other than what you intended | They wanted to continue but could not tell how | Wording, order of steps, what the screen offers next |
| No reason or prompt to return | "It seemed fine, I just never got round to it", "I forgot", no urgent event behind the sign-up | Nothing gave them a reason to come back, or nothing nudged them when it mattered | The reason to return: a smaller first win, a reminder tied to their situation |
| Missing capability | "I needed it to do…", they left to use something else for that job | The product could not do what they came for | Roadmap, positioning, or who you let in |
Be strict about the sorting. "It was confusing" is not a finding until you know what they were trying to do at that moment. If the thing is something your product does, it is confusion. If it does not do it, it is capability. If they had stopped trying before they got there, look for a missing reason or prompt.
Keep the three apart in your notes and fix them separately. The one you fix first is the one with the strongest evidence at the earliest broken step, not the one that is easiest to build.
Match interview stories to the events you already track
You do not need new tracking to start. Take the step names you hear in interviews, find the nearest event you already record, and write the pairing down. Do it per step, not per person.
Here is an invented example to show the method. It follows a small company that sells booking software to independent dog groomers, and every row below comes from it. The full case comes in the next section.
| Step in their story | Closest product event you track | Does the data agree? | Next question |
|---|---|---|---|
| "I signed up after a bad Saturday" | Sign-up completed | Yes: it is the first event | Which sign-ups have a trigger like this? |
| "I tried to connect my calendar but wasn't sure it was safe" | Calendar connection started | Check: started but not completed? | Where do the stalled group stop? |
| "I invited my receptionist straight away" | Teammate invited | Yes, in the got-value group only | Do others invite later, or never? |
| "I wanted to take deposits, so I left" | None | No event exists | What do people do right before they cancel? |
Two things happen when you fill this in. Where the story and the event agree, you now know why the number is what it is. Where there is no event, you have found a blind spot: something that matters to customers that your data cannot see. Add that event, or ask about it in the next round. The same NN/g article that explains journey maps has the warning that applies to funnels too:8.
The dog-groomer case in full
This is the invented case behind the table above. Its funnel shows plenty of sign-ups and few who take a first online booking. The team interviews nine recent sign-ups: three who took bookings, three who stalled, three who cancelled.
Value statements, in the customers' words: the three who succeeded each describe a specific day when phone bookings went wrong ("I missed three calls while I had a dog in the bath"). The stalled group describe a vaguer hope ("get a bit more organised"). Two of the three who cancelled wanted to take deposits.
The timelines show where expectation and result split:
- Sign-up to first session. Nobody reports a problem. The welcome screen is not the issue.
- Connecting the calendar. Two of the stalled group stopped here. One was afraid it would overwrite her existing appointments. The other did not know whether it was safe to try on a Tuesday afternoon. Neither asked for help. Both expected a preview of what would change and got a permission screen. That is confusion plus fear, and both are fixable with wording and a preview.
- Sending the booking link. The third stalled customer connected the calendar and then did nothing for four days. He had no urgent trigger. That is a missing reason to return.
- Deposits. Two of the three who cancelled needed them. That is missing capability, and no wording change will fix it.
The team matches the stories to events. "Calendar connection started" exists and shows starts without completions, which agrees with the interviews. All three who took bookings invited their receptionist in the first session ("I invited my receptionist straight away"), and "Teammate invited" appears in the got-value group only. Nothing records people looking for deposits. The decision: fix the calendar step first, because the evidence is strongest and the break is earliest. The deposit gap goes to the roadmap discussion with its own evidence attached, and the team adds a "looked at payment settings" event so the next round of interviews has data to check against.
What this will not tell you
Interviews give you reasons and sequences, not sizes. Nine conversations show what kinds of break exist at a step. They do not show how many sign-ups hit each one. For that you need your event data, and the pairing table above is where the two meet. Treat the first step you fix as a test, then check your numbers afterwards.
Every account you hear is a recollection, not a record. NN/g separates1, so weight what people can show you over what they say they felt. Ask for the screen, the message, the moment, and where you can, watch them do the step again.
Finally, you hear only from the people who agreed to talk. Say so in your notes and keep your conclusions narrow.
Next step
Pick six recent sign-ups this week, two who got value, two who stalled and two who left, and send the invitations today. Use the guide as written, fill in the worksheet, and bring the first broken step to your next planning meeting.
For the questions that move people past polite first answers, read our guide to customer interview follow-ups. If you are not sure onboarding is the right study at all, which customer research study should you run helps you choose. There is also a page on onboarding research.
