Quick answer
A walkthrough is a guided pass through a process one step at a time. In product work it means either a guided walkthrough that teaches a user a task, or an evaluation walkthrough where a team steps through a flow to find where users would fail.
What a walkthrough is
A walkthrough is a guided pass through a process, one step at a time, so that someone can follow it or check it. The word is used in several fields, and the meaning changes with the field. Search for it and you will see game guides, software onboarding screens and review meetings mixed together.
In product and software work it usually means one of two things:
- a guided walkthrough: a sequence of prompts that shows a new user how to complete a task in the product (also called a product tour or onboarding tour);
- an evaluation walkthrough: a team steps through a flow to find places where users would get stuck, before or without real users.
The rest of this page covers both, then the other meanings so you can tell which one a person is talking about.
The meanings of walkthrough compared
| Meaning | Who does it | Purpose | Output |
|---|---|---|---|
| Guided product walkthrough | The product shows it to a user | Teach a task in context | A completed first task |
| Cognitive walkthrough | A review team steps through a task | Find steps users would fail | A list of failing steps |
| Code walkthrough | An author explains code to colleagues | Share understanding, catch defects | Comments and decisions |
| Video game walkthrough | A guide writer or player | Help a player finish a game | A step list or video |
| Walk-through test (audit) | An auditor follows one transaction | Confirm a control works as described | Audit evidence |
| Tutoring or rehearsal walkthrough | A teacher or director | Practise before the real thing | Confidence |
The disambiguation list is public on Wikipedia. The rest of this page is about the first two, because they are the ones that affect a website or app.
The cognitive walkthrough, step by step
A cognitive walkthrough checks whether a first-time user can work out the right action at every step of a task. Nielsen Norman Group describes the workshop like this (NN/g):
- Set up. Define the user task and decide what counts as one step.
- Walk through each step. A facilitator performs the task and stops at each new screen or discrete action.
- Ask four questions at each step: Will users try to achieve the right result? Will they notice that the correct action is available? Will they connect that action with the result they want? After acting, will they see that progress is made?
- Record. If any answer is "no", the step is marked as a fail.
- Summarise the failing steps and decide what to change.
The people in the room are a facilitator, evaluators (designers, product owners, engineers, domain experts) and a recorder. No real users are needed, which makes it fast and cheap, and also limits it: it finds problems you can imagine, not the ones you cannot.
A worked example: a five-step signup
Sample data: a signup flow with five steps. A team runs a cognitive walkthrough and then checks the funnel from last week.
| Step | Walkthrough result | Sessions reaching the step | Loss from previous step |
|---|---|---|---|
| 1. Landing, click "Start" | Pass | 1,000 | n/a |
| 2. Enter email | Pass | 940 | 6.0% |
| 3. Choose a workspace type | Fail: choices use internal jargon, and the continue button is below the fold on phones | 850 | 9.6% |
| 4. Invite teammates | Pass | 510 | 40.0% |
| 5. Done | Pass | 470 | 7.8% |
Check: step 2 lost 60 of 1,000 = 6.0%; step 3 lost 90 of 940 = 9.6%; step 4 lost 340 of 850 = 40.0%; step 5 lost 40 of 510 = 7.8%. Overall 470 ÷ 1,000 = 47%.
Look at what each source says:
- The walkthrough predicted trouble at step 3 (jargon, hidden button) but not at step 4.
- The funnel shows the biggest loss at step 4, the invite screen, where 40% leave.
So the review and the data disagree, and both matter. The walkthrough found a real flaw that the funnel does not isolate. The funnel found the largest loss, which the team had judged a pass. The decision rule: fix step 3 because it is cheap and certain, and run a second walkthrough focused on step 4 asking what a first-time user thinks "invite teammates" will do. If the invite step is optional, add a skip link and watch the loss there.
Guided walkthroughs: what makes them work
A guided walkthrough succeeds when it helps a user reach a useful first result sooner. Four practical points:
- Tie it to one task. "Create your first report" beats a ten-screen tour of every menu.
- Show it when needed. Triggering at the moment of the task is better than a forced tour before the user has context.
- Allow skipping and resuming. Forced tours show up as abandon rates.
- Measure the result, not the clicks. The success measure is whether the user completed the task and came back, not how many people clicked "Next". See product tour and first-time user experience.
A quick check on "did the tour help?" with sample data: 1,000 new users are shown a tour and 620 finish it (62%). Of those 620, 372 create their first report (60%); of the 380 who skipped, 190 do (50%). The gap looks like a win for the tour, but the people who finish a tour may simply be more motivated. The fair test is to compare users randomly shown the tour with users who are not, or at least the same signup cohort before and after the tour launched.
Common mistakes
- Using a walkthrough instead of fixing the screen. If a step needs a pop-up to explain it, first ask whether the label can be clearer.
- Walking through the happy path only. Include an error, an empty state and a slow connection.
- Treating "users clicked through" as success. Completion of the tour is not task completion.
- Evaluating with the people who built it. Add someone who has never seen the flow.
- Never re-measuring. A walkthrough is a hypothesis; the funnel after the fix is the test. For the bigger picture of what comes after signup, see the blog on onboarding flow.
Checking a walkthrough in TapCub
Use the signup example above and measure it in TapCub.
- Send an event for each step. Call
tapcub.trackwith a snake_case name per step, for examplesignup_workspace_chosen. Page views alone cannot tell steps apart when they live on one URL. - Build the funnel. In All-platform analytics › Behavior › Behavior analysis, add the steps in order and read the loss between each pair, as in the table above.
- Look at the screen. For the step with the largest loss, open All-platform analytics › Behavior › Heatmaps to see where clicks land, and session replay (off by default) to watch a few attempts.
- Compare after the change. Annotate the release date with Configuration › Annotations, and re-run the funnel for a comparable period.
TapCub does not build or show in-app guides or tours. It measures the flow before and after you change it, in whatever tool you use to build the guidance.
Sources
- Nielsen Norman Group, Cognitive Walkthroughs.
- Wikipedia, Walkthrough (disambiguation).
Was this explanation clear?
We use your answers to fix and expand this page.
Couldn’t send. Please try again in a minute.
Thanks for the feedback. We’ll use it to improve this page.You’ve already sent feedback on this page. Thanks.

Data Analyst
Explores traffic sources, conversion funnels, and retention, using clear metric definitions and concrete examples to explain what the numbers mean.