We never wrote a tracking plan. Autocapture collected the clicks, AI named them, I confirmed six events in the inbox. The activation funnel worked the same afternoon.
JL
J. L.
Product manager · B2B SaaS
For product managers Where do new users drop off?
Autocapture records the clicks from day one. You confirm the events that matter, then read the funnel, the paths and the cohort that came back.
Questions you answer every day
- Which step of onboarding loses the most people?
- Did last week’s release change week-1 retention?
- Which feature do paying accounts actually use?
- What do people do right after they see the pricing page?
Signups
2,400
▲ 6.8%
Activated
1,248
▲ 3.1%
Reached core feature
674
▼ 4.4%
Week-1 return
46%
▲ 2 pts
Activation funnel
7 daysSignup
2,400
Onboarding done
1,248
Core feature
674
Return
508
Weekly retention by signup cohort
| Cohort | Users | W0 | W1 | W2 | W3 | W4 |
|---|---|---|---|---|---|---|
| Sep 1 | 1,840 | 100% | 46% | 38% | 33% | 31% |
| Sep 8 | 2,012 | 100% | 49% | 41% | 36% | |
| Sep 15 | 1,966 | 100% | 52% | 44% | ||
| Sep 22 | 2,130 | 100% | 58% |
The daily friction
Three things that slow a product decision
The data exists. It is just not in a shape a product manager can read before the planning meeting.
The tracking plan is never finished
Every funnel needs an event that was not instrumented. The ticket waits a sprint, the question is forgotten by then.
Numbers without the “who”
A funnel says 46% left at step three. It does not say whether they were trial users, paying accounts or people from one campaign.
Three platforms, three users
The same person signs up on the web, opens the app, pays in the mini-program. Three tools count three users and no journey.
Numbers you report
The activation story in four numbers
Shared sample funnel used across the site: 2,400 signups, 508 reach the end.
- Signed up
- 2,400
- Signed up
- Finished onboarding
- 1,248
- Finished onboarding
- Largest drop · core feature
- −46%
- Largest drop · core feature
- Week-1 return
- 46%
- Week-1 return
Sample data. Funnel steps are counted per person inside a 7-day window.
Built for product
Six reports a product manager lives in
Funnels, paths and cohorts are the daily ones. Autocapture and the AI analyst are how you get there without a sprint of instrumentation.
Funnels that show the step, and who left
Build a funnel from confirmed events in a minute. Click the drop to see the people in it, their source, device and the page they went to instead.
- Per-person counting inside a window you choose (1 to 90 days)
- Break the same funnel down by plan, platform, campaign or any property
- Save it, share it, compare it against last month
Example scenario: Step three lost 46%. Breaking down by platform showed the mini-program version lacked the “skip” button. One fix, one week, the drop halved.
Activation · by platform
7-day windowSignup
2,400
Onboarding done
1,248
Core feature
674
Return
508
Autocapture first, naming later
The snippet records clicks, page views and form submits from day one. AI groups them and proposes event names in the inbox. You confirm the ones that matter.
- No tracking plan before the first funnel
- Suggested names come with a confidence score and the raw selector behind them
- Server events (orders, plans, subscriptions) sit next to autocaptured ones
Event inbox · 6 suggestions
This weekstart_trialbutton.cta-primary · /pricingConfirmedinvite_membera[href*=invite] · /settings/teamConfirmedexport_csvbutton[data-export] · /reportsConfirmopen_pricingnav a[href=/pricing]Confirmcomplete_onboardingform#onboard submitConfirmedview_changeloga[href=/changelog]Dismiss
Retention by cohort, read weekly
Who comes back after the first week, grouped by the week they signed up. A release that helped shows up as a brighter row.
- Signup, first purchase or any event as the cohort start
- Daily, weekly or monthly grid, up to 12 periods
- Switch the return event: any visit, or only the core feature
Week-1 to week-4 return · signup cohorts
| Cohort | Users | W0 | W1 | W2 | W3 | W4 |
|---|---|---|---|---|---|---|
| Sep 1 | 1,840 | 100% | 46% | 38% | 33% | 31% |
| Sep 8 | 2,012 | 100% | 49% | 41% | 36% | |
| Sep 15 | 1,966 | 100% | 52% | 44% | ||
| Sep 22 | 2,130 | 100% | 58% |
Paths after the pricing page
Where people go in the three steps after an event. The branch you expected and the branch you did not, with the size of each.
- Start from any page or event, forward or backward
- Up to 10 branches per step, the rest grouped as “other”
- Click a branch to turn it into a segment
Three steps after /pricing
Heatmaps on the page that lost them
Click and scroll maps for the step that dropped. See the button nobody finds and the fold nobody passes.
- Click, move and scroll depth per page and device
- Rage clicks on elements that did not respond
- Split by segment: new users only, or one campaign
Scroll depth
100% → 22%
Only 22% reach the second call to action. The primary button sits under the fold on mobile.
Sample dataAsk the question in one sentence
“Which feature do paying accounts use most?” The AI analyst picks the events, runs the query and shows the chart with the definition under it, so you can check the logic.
- Natural language to a saved analysis you can edit
- Weekly digest of what changed and why it probably did
- Anomaly alerts on activation and retention
Feature use · paying accounts · 28 days
Share of accountsA day in the role
From the morning funnel to the sprint review
A product manager’s week in five moves. Each one is a saved report that is already waiting.
-
09:001
Open the activation funnel
Yesterday’s signups against the 7-day window. Step three dropped 46%, four points worse than last week.
Funnel · activation
-
09:402
Find who dropped
Break down by platform. The mini-program cohort is the one that left. Click into the people and see the page they went to.
Breakdown · platform
-
10:303
Look at the page
Heatmap of the onboarding step on mobile. The skip button is below the fold; rage clicks on a disabled “next”.
Heatmap · /onboarding/3
-
14:004
Write the ticket with evidence
Share the funnel link and the heatmap in the ticket. Engineering sees the same numbers you do.
Shared analysis link
-
Next Tue5
Check the cohort
The release went out Friday. The Sep 22 cohort row is brighter than Sep 15. The drop halved; the retention table says it stuck.
Retention · signup cohorts
No ticket to instrument a new event was needed. The clicks were already captured; two were confirmed in the inbox.
Working with other roles
Hand the same funnel to the people who act on it
Product finds the step. Engineering fixes it, growth measures it, support hears about it first. One shared analysis beats four screenshots.
Developers
Get the exact selector behind an autocaptured event and the server events that need a name.
OpenGrowth & operations
Reads the same cohort table to decide whether the activation change deserves a campaign.
OpenSupport teams
Sees the path a user took before opening chat, so a confusing step is reported with context.
OpenShare “Activation funnel · by platform”
Copy link- PMProduct managerAnalyst
- ENEngineeringAnalyst
- GRGrowthAnalyst
- CSSupport leadViewer
Shared analyses are read-only, show up to the last 90 days, and respect the project’s filter allowlist.
Analysts build and save analyses, segments and dashboards. Viewers read them. The event inbox and tracking rules need Analyst or above; billing and members need Admin.
Templates & solutions
Start with a pack that already has your funnel
The SaaS pack installs signup, onboarding, feature use, trial and subscription. Other packs do the same for their industry.
Templates that fit this role
Solutions worth reading next
In their words
What a product manager notices first
Two scenes from the job, written as the person doing it might tell them.
The drop was only on the mini-program. Three tools would have shown three users. One profile across web, app and mini-program showed one person giving up at the same step.
HY
H. Y.
Senior PM · consumer app
Illustrative scenarios written from the perspective of people who do this job. Not real reviews.
FAQ
Questions product managers ask
Still have a question?
Chat with the team behind TapCub — we usually reply within a few hours on business days.
Do I have to write a tracking plan first?
No. The snippet autocaptures clicks, page views and form submits from day one. You name and confirm events in the inbox; funnels use confirmed events.
Can I still send my own events?
Yes. Orders, plans and other business fields come from your server through the SDKs or API. They sit next to autocaptured events on the same person.
How is the same user matched across web, app and mini-program?
Each platform sends its own anonymous ID; a login ID merges them into one person. Funnels and retention count that person once.
Are heatmaps included?
Yes. Click, move and scroll heatmaps are part of behavioral analytics, per page and device, and can be filtered by segment.
Can I write SQL?
Yes. Pro mode exposes the event table to SQL for questions the visual models do not cover. Results can be saved as an analysis like any other.
Install the snippet today, read the first funnel tomorrow
Free plan, no credit card. Autocapture starts the moment the snippet loads.