Cohort analysis for product tours: finding what works
Your product tour has a 61% completion rate. Congratulations. But does that number actually predict whether users stick around past week two?
Completion rate is a vanity metric without context. A user who clicked through five tooltip steps in eight seconds and a user who paused, explored each feature, and came back the next day both count as "completed." Cohort analysis separates the two and tells you which tour design actually drives retention.
We ran this analysis on our own onboarding flows and the results were not what we expected. This guide walks through building cohort analysis around product tour events: which cohort types matter, how to connect tour step data to your analytics tool, and what the benchmarks say about trigger types, step counts, and timing windows.
npm install @tourkit/core @tourkit/react @tourkit/analyticsWhat is cohort analysis for product tours?
Cohort analysis for product tours groups users by shared tour-related behaviors (when they signed up, which tour variant they saw, how they triggered it, which step they reached) then compares retention, activation, or conversion rates across those groups over time. Unlike aggregate tour completion metrics that flatten all users into a single percentage, cohort analysis reveals which specific tour experiences correlate with long-term product engagement. As of April 2026, Chameleon's benchmark dataset of 15 million tour interactions shows that trigger type alone creates a 2x gap in completion rates (Chameleon, 2025), yet most teams never segment their tour data by trigger method.
Standard tour analytics answers "how many users finished the tour?" Cohort analysis answers a harder question: did tour completers retain better than tour skippers at day 30, and does the answer change depending on how we triggered the tour?
Why cohort analysis matters for tour-driven onboarding
Product teams spend weeks crafting tour flows but measure success with a single metric: completion rate. That's like measuring a sales team by how many demos they booked without checking if any converted to paying customers. Cohort analysis connects the tour experience to the business outcome you actually care about, whether that's 30-day retention, feature adoption, or paid conversion. Without it, you're optimizing tours in the dark.
Why tour completion rate alone misleads you
The average product tour completion rate sits at 61%, based on Chameleon's analysis of 15 million interactions. Sounds healthy. But the median SaaS app loses 75% of its daily active users within the first week (Mixpanel, 2022 Product Benchmarks). If your tour completers churn at nearly the same rate as non-completers, the tour isn't doing its job. The completion rate just hides that fact.
This is the false positive problem in tour analytics. High completion doesn't guarantee comprehension. Users clicking "Next" to dismiss a tooltip register identically to users who read each step and tried the feature. Only a cohort split (completers-with-feature-usage versus completers-without) surfaces the difference.
Here's what happened at Slack. Their growth team identified that teams sending 2,000+ messages retained at dramatically higher rates. They rebuilt onboarding to push new teams toward that behavior faster. Result: 30-day retention for new teams jumped 17%. The insight came from behavioral cohort analysis, not from measuring whether people finished a welcome tour.
The four cohort types that matter
Not all cohort splits are equal. These four produce the most actionable signal when applied to product tour data.
Acquisition cohorts
Group users by signup week or month, then compare tour completion rates across cohorts. This catches seasonal effects, marketing campaign quality shifts, and the impact of tour changes over time. If you shipped a new 3-step tour in March and your March cohort completes at 74% versus February's 58%, you've got a signal. But only if you also check whether March completers retained better, not just finished faster.
Behavioral cohorts
The most powerful split for tour analysis. Group users by what they did during or after the tour:
- Completed tour and used the featured action within 24 hours
- Completed tour but never used the featured action
- Abandoned tour at step 3
- Skipped tour entirely
Then run retention curves for each group at day 7, 14, and 30. Users completing onboarding within the first 24 hours show 40% higher retention at 90 days in cohort analyses across SaaS products. But only behavioral cohorts tell you whether your tour is the mechanism driving that or just coincidental.
Trigger-type cohorts
This is the cohort dimension most teams miss entirely. Chameleon's benchmark data shows stark differences by how a tour starts:
| Trigger type | Completion rate | Source |
|---|---|---|
| Click-triggered (user initiates) | 67% | Chameleon 15M benchmark |
| Launcher-driven | 67% | Chameleon 15M benchmark |
| Checklist-triggered | +21% vs baseline | Chameleon 15M benchmark |
| Auto-popup (set delay) | 31% | Chameleon 15M benchmark |
A 2x completion gap between click-triggered and auto-popup tours is massive. But the unanswered question, and the experiment you should run, is whether click-triggered completers also retain at higher rates. Users who chose to start a tour have higher intent, meaning completion is a genuine engagement signal rather than a dismissal pattern. Cohort-split by trigger type, measure retention at day 30, and you'll know.
Tour-length cohorts
Step count matters more than most teams realize. The data from Chameleon's 2025 benchmark report across 550 million interactions:
- 3-step tours: 72% completion
- 4-step tours: 74% completion (the sweet spot)
- 5-step tours: 34% completion
- 7+ step tours: 16% completion
That cliff between 4 and 5 steps aligns with cognitive load research. People hold roughly 5 to 7 items in working memory (Smashing Magazine, 2023). But completion isn't the whole story. Run a cohort comparing 3-step completers versus 4-step completers on 30-day retention. If your 4-step tour covers the same ground, the extra step may reinforce the aha moment without crossing the cognitive overload threshold.
Piping tour events into cohort analysis
Most implementations have a technical gap where the tour library emits events and the analytics tool builds cohorts, but the two aren't connected at the step level. We hit this exact problem when measuring our own onboarding tours: Amplitude showed a 61% completion rate, but we couldn't answer whether completers retained any better than skippers. Here's how to bridge that gap with Tour Kit and any analytics platform.
Emit granular tour events
// src/components/TrackedTour.tsx
import { TourProvider, useTour } from '@tourkit/react';
import { useAnalytics } from '@tourkit/analytics';
const tourSteps = [
{ id: 'welcome', target: '#dashboard-header', title: 'Your dashboard' },
{ id: 'create-project', target: '#new-project-btn', title: 'Create a project' },
{ id: 'invite-team', target: '#invite-btn', title: 'Invite your team' },
];
function OnboardingTour() {
const analytics = useAnalytics();
return (
<TourProvider
tourId="onboarding-v3"
steps={tourSteps}
onStepChange={(step, prevStep) => {
analytics.track('tour_step_viewed', {
tour_id: 'onboarding-v3',
step_id: step.id,
step_index: tourSteps.indexOf(step),
trigger_type: 'checklist', // tag the trigger method
timestamp: Date.now(),
});
}}
onComplete={() => {
analytics.track('tour_completed', {
tour_id: 'onboarding-v3',
total_steps: tourSteps.length,
trigger_type: 'checklist',
});
}}
onDismiss={(step) => {
analytics.track('tour_dismissed', {
tour_id: 'onboarding-v3',
dismissed_at_step: step.id,
steps_completed: tourSteps.indexOf(step),
});
}}
/>
);
}The key properties: tour_id, step_id, trigger_type, and dismissed_at_step. These become your cohort dimensions. Without step-level events, you can only split on "completed vs didn't," which is barely more useful than the aggregate completion rate.
Build behavioral cohorts in your analytics tool
Once events flow, create these cohorts in Amplitude, Mixpanel, PostHog, or whichever tool your team uses:
- Completers: users who fired
tour_completedfor a giventour_id - Abandoners at step N: users who fired
tour_step_viewedfor step N but never fired the event for step N+1 - Skippers: users active in the same period who never fired any
tour_step_viewedevent - Feature activators post-tour: users who fired
tour_completedAND performed the tour's target action within 24 hours
Compare retention curves across all four. If cohort 1 and cohort 4 look identical, the tour is informational but not activating. If cohort 4 retains 2x better than cohort 1, the tour works, but only when users act on it.
Tour Kit's @tourkit/analytics package connects to PostHog, Amplitude, Mixpanel, and custom backends through a plugin interface. Step-level events fire automatically. See the analytics integration docs for setup.
Choosing the right cohort window
Picking the wrong retention window is the fastest way to misread your cohort data, so match the measurement period to how often people actually use your product. A daily-use collaboration tool and a quarterly-use reporting platform need completely different windows. The gotcha we hit: measuring enterprise trial users on a day-7 window made our tour look ineffective, when really those users just hadn't returned for their second evaluation session yet.
| Product type | Cohort windows | Why |
|---|---|---|
| Daily-use SaaS (Slack, Linear) | Day 1, 7, 14, 30 | Fast feedback loops; churn signals appear within days |
| Weekly-use tools (project management, analytics) | Week 1, 2, 4, 8 | Weekly cadence means Day 1 retention is meaningless |
| Enterprise / long sales cycle | Week 1, 4, 12, 26 | 90-day evaluation periods; early inactivity isn't churn, it's evaluation timing |
| Event-driven (invoicing, tax prep) | Event 1, 2, 3 (not time-based) | Calendar time doesn't reflect engagement; measure by usage events |
Most product tour articles default to "measure day 7 and day 30 retention." That works for daily-use SaaS. It's actively misleading for enterprise tools where a user who logs in once during a 90-day evaluation and then becomes a daily user looks like a churned user at day 14. Match the window to the product.
Common mistakes in tour cohort analysis
Three patterns consistently produce misleading results when teams start building cohort analysis around their product tour data. We measured these pitfalls against our own onboarding analytics and each one distorted the numbers in a different way.
Survivorship bias in completion cohorts. Users who complete a 5-step tour are already more engaged than average. They stuck around long enough to finish. Comparing their retention to non-completers and concluding "the tour caused better retention" confuses correlation with causation.
The fix: compare completers against a control cohort who never saw the tour at all (an A/B hold-back group). Tour Kit supports this through conditional tour rendering. Show the tour to 80% of new users, hold back 20%, and compare cohorts cleanly.
Ignoring dwell time per step. Two users complete the same tour. One spent 45 seconds per step. The other clicked through all five steps in 8 seconds. Both are "completers."
Add step dwell time to your tour events and split your completer cohort into meaningful-engagement and click-through subgroups. The retention difference tells you if your tour content actually resonates.
Using acquisition cohorts when you need behavioral ones. Acquisition cohorts (grouped by signup date) are useful for measuring changes to your tour over time. But they're terrible for measuring tour effectiveness because of confounding variables. A spike in signups from a Product Hunt launch changes the composition of your weekly cohort entirely. Behavioral cohorts (grouped by tour interaction pattern) isolate the tour's impact from acquisition channel noise.
Tools for tour cohort analysis
Your analytics tool determines how granular your cohorts can get, and for developer teams building with a headless tour library like Tour Kit, the priority is event flexibility. You need to send custom properties (step ID, trigger type, dwell time) and build cohorts from them. Not every tool makes this easy.
PostHog stands out for this use case. It's open-source, self-hostable, and its behavioral cohort builder accepts any event property as a cohort criterion. Connect Tour Kit's analytics plugin to PostHog and you get step-level cohort analysis without vendor lock-in or per-seat pricing.
Amplitude and Mixpanel both support behavioral cohorts with generous free tiers (10M events/month and 100K tracked users/month, respectively). But PostHog's open-source model and 1M events/month free tier make it the natural pairing for teams already using open-source tour infrastructure.
For a deeper walkthrough of the PostHog integration, see our guide on onboarding analytics dashboards with PostHog. And for Amplitude, Amplitude + Tour Kit: measuring onboarding impact on retention covers the full pipeline.
One honest limitation: Tour Kit doesn't include a built-in analytics dashboard or cohort visualization. It's a tour library, not an analytics platform. The analytics package handles event emission and plugin routing, and you bring your own analysis tool.
For teams that want cohort analysis built into the same tool that runs their tours, Pendo and Chameleon offer that (at significantly higher price points and with less UI flexibility).
Getting started with Tour Kit
Tour Kit's headless architecture means you control every event property your tour emits. That control is what makes rich cohort analysis possible. You aren't limited to the four or five built-in events a SaaS tour tool decides to track.
npm install @tourkit/core @tourkit/react @tourkit/analyticsStart with the analytics integration docs, connect your analytics provider, and run your first behavioral cohort split within a week. The investment is small. Knowing which tour design drives retention, not just completion, compounds over every product iteration.
Visit usertourkit.com for the full documentation, GitHub repo, and live examples.
FAQ
What is cohort analysis for product tours?
Cohort analysis for product tours segments users by tour-related behaviors (completion status, trigger type, step reached, or signup timing) and compares retention or activation rates across those groups. Unlike aggregate tour metrics, cohort analysis reveals whether a specific tour experience correlates with long-term engagement. Chameleon's dataset of 15 million interactions shows trigger type alone creates a 2x completion gap.
Which cohort type is most useful for measuring tour effectiveness?
Behavioral cohorts produce the most actionable signal for product tour analysis. Group users by what they did during and after the tour: completed and activated, completed but didn't activate, abandoned at step N, or skipped entirely. Acquisition cohorts (by signup date) carry too many confounding variables to isolate tour impact cleanly.
How many steps should a product tour have for best retention?
Benchmark data from 550 million interactions shows 4-step tours hit peak completion at 74%, while 5-step tours drop to 34%. This aligns with cognitive load research on working memory limits. Run a cohort analysis to check whether 4-step completers retain better at day 30 before optimizing purely for completion rate.
What analytics tools support tour-based cohort analysis?
PostHog, Amplitude, and Mixpanel all support behavioral cohorts built from custom event properties. PostHog is open-source with a 1M events/month free tier. Amplitude offers 10M events/month free. Mixpanel covers 100K tracked users/month. All three accept the step-level tour events that Tour Kit's analytics package emits.
How do I avoid survivorship bias in tour completion cohorts?
Use an A/B hold-back group. Show the tour to 80% of new users and withhold it from 20%, then compare retention between the two groups. Completers self-select for higher engagement, so comparing them to non-completers overstates the tour's impact. Tour Kit supports conditional rendering for hold-back experiments.
Related articles
Behavioral triggers for product tours: event-based onboarding
Build event-based product tours that trigger on user actions, not timers. Code examples for click, route, inactivity, and compound triggers in React.
Read article
How to calculate feature adoption rate (with code examples)
Calculate feature adoption rate with TypeScript examples. Four formula variants, React hooks, and benchmarks from 181 B2B SaaS companies.
Read article
Setting up custom events for tour analytics in React
Build type-safe custom event tracking for product tours in React. Wire step views, completions, and abandonment to GA4, PostHog, or any analytics provider.
Read article
DAU/MAU ratio and onboarding: how tours improve stickiness
Learn how DAU/MAU ratio connects to product onboarding quality. See benchmarks, formulas, and code examples for tracking tour-driven stickiness in React.
Read article