Use it for
- Interviews with users who signed up in the last 30 days, both those who stayed and those who left.
- Session recordings or usability test notes of the first session.
- Your onboarding funnel with the drop-off at each step, and support tickets from accounts younger than 30 days.
Steps
- Map the onboarding steps
List the steps from sign-up to the first result, in order. Use names that match your funnel data, such as Sign up, Verify email, Connect data, Invite team and First report.
- Find the friction passages
Mark each passage where a new user hesitates, makes an error, asks for help, leaves or says they were unsure. Include friction that the user solved alone. That friction still costs users who did not.
- Code the type and the step
Give each passage one friction code and the onboarding step where it happened. The same code at different steps needs different fixes.
- Record the outcome
For each passage, record what the user did next: continued alone, got help, came back later or left. Friction that ends in a lost user is the top priority.
- Compare stayers and leavers
Count codes for users who reached the first result and for users who did not. Codes that only leavers mention are the likely causes of drop-off.
- Match codes to the funnel
Put the codes next to the drop-off rate at each step. A step with a high drop-off and a clear code is the first place to fix.
Codebook
Copy the codebook as a table. It pastes into a spreadsheet with one row for each code.
Sign-up form burden
The sign-up form asks for too much, or asks for something the user cannot or will not give.
- Include
- The user complains about the number of fields, a phone number field or a card for the trial.
- Exclude
- The user filled in the form but could not verify the account. Use Verification block for that.
- Example
- “It wanted my phone number and company revenue before I had even seen the product.”
Verification block
The user cannot confirm their email, phone or identity, so they cannot continue.
- Include
- The user mentions a late email, a code that fails, an expired link or a spam folder.
- Exclude
- The user can sign in but needs a teammate or admin to approve them. Use Dependency on others for that.
- Example
- “The confirmation email took twenty minutes, and by then I was in a meeting.”
Dependency on others
The user needs another person, such as an admin, IT or a colleague, to complete a step.
- Include
- The user waits for approval, an invite, admin rights, an API key or data that another person owns.
- Exclude
- The user needs their own data from another tool and can get it alone. Use Data setup effort for that.
- Example
- “I needed IT to approve the Google connection, and they take a week for anything.”
Data setup effort
The user must import, connect or enter data before the product is useful, and the work is long or hard.
- Include
- The user describes a long import, field mapping, file format errors or manual data entry.
- Exclude
- The import works but the product still shows nothing useful. Use Empty state for that.
- Example
- “I had to rename every column in my spreadsheet before it would import.”
Empty state
The user reaches a screen with no content and no clear way to fill it.
- Include
- The user sees a blank dashboard, an empty list or a zero chart and does not know what to do.
- Exclude
- The screen offers a next step but the user does not understand it. Use Unclear next step for that.
- Example
- “I logged in and there was just a big empty box that said No projects. Now what?”
Unclear next step
The user does not know which action to take next to reach a result.
- Include
- The user hesitates between options, asks what to do next, or clicks around with no plan.
- Exclude
- The user knows the next step but does not understand a word in it. Use Unfamiliar terms for that.
- Example
- “There were five buttons on the home page and I did not know which one starts things.”
Unfamiliar terms
The user does not understand a word, label or concept that the product uses.
- Include
- The user asks what a term means, guesses wrong, or uses a different word for the same thing.
- Exclude
- The user understands the words but not the order of steps. Use Unclear next step for that.
- Example
- “What is a workspace versus a project? I made three of the wrong one.”
Guided tour interference
A tour, tooltip, checklist or pop-up gets in the way of what the user wants to do.
- Include
- The user closes tips without reading, feels interrupted, or loses their place because of a tour.
- Exclude
- The user wanted help and could not find it. Use Unclear next step for that.
- Example
- “The tour kept popping up over the button I was trying to click, so I just closed all of it.”
Slow time to value
The user completes setup but the first useful result takes too long to appear.
- Include
- The user waits for processing, data to sync or results to build, and says it took too long.
- Exclude
- The user is stuck on a setup task and cannot finish it. Use the code for that task.
- Example
- “It said my first report would be ready in 24 hours. I forgot about it by then.”
Expectation mismatch
The product the user finds is different from what the website, ad or sales call promised.
- Include
- The user says they expected a feature, a price, a template or a use case that is not there.
- Exclude
- The feature exists but the user could not find it. Use Unclear next step for that.
- Example
- “The ad showed a ready-made template for my industry, but I had to build everything from scratch.”
Setup error
A technical error stops or breaks a step during onboarding.
- Include
- The user sees an error message, a page that does not load or a step that fails.
- Exclude
- The step works as designed but is long or hard. Use Data setup effort for that.
- Example
- “I clicked Connect and got Something went wrong, three times in a row.”
Report outline
- The onboarding steps and the drop-off at each one, from your funnel data.
- Friction by step: the top codes at each step, with the outcome for the users who had them.
- Codes that separate users who left from users who stayed, with quotes from both groups.
- The first result: how long it takes and what slows it.
- Recommended changes per step, each with the funnel metric that will show the effect.
- Method: who was interviewed or observed, when they signed up and the limits of the sample.
Mistakes to avoid
- Do not interview only users who stayed. They solved the friction. The users who left show which friction is fatal.
- Do not code only problems that users complain about. Watch the recordings. Hesitation and wrong clicks show friction that users do not mention.
- Do not fix friction without the step. "Unclear next step" on the home page and in the import screen need different fixes.
- Do not add more tour steps to fix every code. A longer tour often creates Guided tour interference.
Guides that use it