Behavioral analytics See what people do, not just how many came
Events, funnels, retention, paths and heatmaps built from the same autocaptured data. The sample checkout goes 2,400 → 1,248 → 674 → 508 — and the 46% lost between cart and checkout is the first thing to fix.
- Autocapture, no tagging plan
- Funnels with a time window
- Cohort retention
- Click and scroll heatmaps
Checkout funnel
7-day windowViewed product
2,400
Added to cart
1,248
Started checkout
674
Paid
508
Checkout conversion · daily
Median time between steps
- Product → cart1 m 40 s
- Cart → checkout6 m 12 s
- Checkout → paid2 m 05 s
Page views tell you where. Behavior tells you why.
Three questions a visits-and-pages report cannot answer, and what TapCub does instead.
"Conversion dropped — which step?"
A site-wide conversion rate hides where people leave. You end up guessing between the cart page, the address form and the payment screen.
A funnel with a 7-day window shows each step. The sample loses 46% between cart and checkout; the other steps hold.
"We never finished the tracking plan"
Event-based tools want a spec, a sprint and a release before the first chart. Six months later half the events are still missing.
Autocapture records clicks, submits and page views from day one. You name the events that matter afterwards — the history is already there.
"Who exactly dropped off?"
Aggregate charts cannot be opened. You know 574 people left; you do not know they were mostly mobile visitors from one campaign.
Every step is a list of people. Open it, filter by device or channel, save it as a segment, send it to chat or an export.
Every click is already an event
TapCub autocaptures clicks, form submits, page views and SPA route changes. Name the ones that matter — "Add to cart", "Start checkout", "Order paid" — and the chart, the funnel and the retention table all use the same definition.
- Autocapture: clicks, submits, page views, route changes
- Name an event once; every report updates, history included
- Custom events from the SDK or server for revenue and sign-ups
Example scenario: A product manager renames the autocaptured click on the black "Checkout" button to Start checkout. The funnel below rebuilds for the last 30 days without a deploy.
Events · daily
Add to cartOrder paidStart checkoutAdd to cart
1,248
▲ 8.2%
Start checkout
674
▲ 3.1%
Order paid
508
▲ 11.4%
Where the 46% goes
A funnel is an ordered set of events with a time window. The sample checkout keeps 52% to cart, 28% to checkout and 21% to paid; the biggest loss is cart → checkout, and the median time there is six minutes — people are hesitating, not failing.
- Any sequence of events, 2 to 10 steps, strict or loose order
- Time window per funnel — 7 days for checkout, 30 for a trial
- Compare segments side by side: mobile vs desktop, paid vs organic
Example scenario: Filtered to mobile, the cart → checkout step drops to 42%. The address form has eleven fields on a phone. Shortening it is the first experiment.
Checkout funnel · 30 days
Last 30 daysAll visitorsViewed product
2,400
Added to cart
1,248
Started checkout
674
Paid
508
| Step | Conversion | Median time | Drop-off |
|---|---|---|---|
| Viewed product → cart | 52% | 1 m 40 s | −48% |
| Cart → checkout | 54% | 6 m 12 s | −46% |
| Checkout → paid | 75% | 2 m 05 s | −25% |
| Overall | 21% | 9 m 57 s | −79% |
Did they come back?
Pick a start event and a return event and TapCub builds the cohort table: each row is the people who started in that week, each column is how many of them returned. Later cohorts in the sample retain better — 44% → 54% in week 1 — which is what a working onboarding change looks like.
- Classic N-day or range retention — one switch
- Return event can be any action, not just a visit
- Incomplete cohorts stay blank, never zero
Example scenario: A growth lead sets the return event to Order paid instead of "any visit". Week-4 retention falls from 29% to 11% — the honest number for repeat purchase.
Weekly cohorts · returned
N-dayRange| Cohort | Users | Week 0 | Week 1 | Week 2 | Week 3 | Week 4 |
|---|---|---|---|---|---|---|
| Aug 25 | 1,204 | 100% | 44% | 36% | 31% | 29% |
| Sep 1 | 1,318 | 100% | 47% | 39% | 34% | |
| Sep 8 | 1,276 | 100% | 51% | 42% | ||
| Sep 15 | 1,402 | 100% | 54% | |||
| Sep 22 | 1,366 | 100% |
Retention curve
Sep cohortsAug cohortsWhat happened before and after
Paths show the real sequence of pages and events around a point you choose — what people did right before adding to cart, or where they went after abandoning checkout. In the sample, 26% of checkout abandoners went straight to the shipping policy page.
- Start from any page or event, forward or backward
- Depth and minimum share controls keep the chart readable
- Click a branch to turn it into a segment
Example scenario: Backward from Started checkout: 38% came from the cart, 14% from a product page directly, 48% from elsewhere — a "Buy now" button is skipping the cart more than expected.
Paths shown
8 of 142
Median steps
3.2
Reach "Paid"
38%
How many, how often
Not every question is a funnel. Distribution charts answer "how many items per order?" and "how many sessions before the first purchase?"; frequency and interval answer "how often do they return and with what gap?"
- Distribution of any numeric property or event count
- Interval between two events — first visit to first order
- Percentiles, not just averages
Example scenario: 48% of orders have a single item; 10% have five or more. The 31% who buy again do so within 8–30 days — the window for a follow-up email.
Items per order
1,168 ordersDays to second order
Repeat buyersMedian items
1
Median days to repeat
19
Repeat buyers
31%
Where they click, how far they scroll
Click and scroll heatmaps are built from the same autocaptured events — no separate script, no sampling. The product page in the sample shows 22% of visitors never reach the reviews, and 9% click an image that does not open.
- Click, move and scroll maps per page and device
- Element table: which buttons and links actually get used
- Compare two date ranges to check a redesign
Example scenario: On mobile, the "Add to cart" button is below the fold for 61% of visitors. Pinning it to the bottom lifts cart adds by 14% in the next funnel comparison.
Scroll depth
100% → 22%
Clicked elements · /product/linen-shirt
| Element | Clicks | Share |
|---|---|---|
| button.add-to-cart | 1,248 | 46% |
| a.size-guide | 612 | 23% |
| img.gallery-2 | 244 | 9% |
| a.reviews | 198 | 7% |
From raw clicks to a named event in four steps
Autocapture records everything; the AI inbox turns the noise into a short list of events worth naming. Confirm or dismiss — nothing goes into reports without you.
Capture
Clicks, submits, page views and route changes are recorded with the element, text and page. ≈2 KB script, no configuration.
Suggest
The inbox groups repeated interactions and proposes names: "Add to cart", "Start checkout", "Open size guide".
Confirm
Accept a suggestion and it becomes a named event, backfilled across the last 30 days. Dismiss and it stays raw.
Extend
Revenue, sign-ups and server-side events come from the SDK with properties — same event list, same reports.
Raw interactions · last 24 h
- click · button.add-to-cart1,248
- click · a.size-guide612
- submit · form.checkout674
- view · /order/confirmed508
Suggested events
- add_to_cart98%
Clicks on the cart button across 38 product pages
- start_checkout96%
Checkout form submitted
- open_size_guide91%
Size guide link on product pages
Confirm
- add_to_cartConfirmed
- start_checkoutConfirmed
- open_size_guideConfirm
Custom events from code
// Order value and currency come from your code
tapcub.track('order_paid', {
order_id: order.id,
amount: order.total,
currency: 'USD',
});Tracking plan first
Weeks before the first chart
Spec every event, implement, release, wait for data, discover the spec was incomplete.
TapCub
Charts on day one, names when you need them
Autocapture from the first page load; the AI inbox proposes the events, you confirm, history is backfilled.
Ten models, two ways to use them
Every model runs on the same events and segments. Use the guided mode with sensible defaults, or the pro mode with the full set of controls — and ask the AI in plain language when you do not know which model you need.
Pick a question, get a chart
Each model opens with the most common setup already filled in: last 30 days, all visitors, the events you confirmed. Change one thing at a time.
- Defaults from your own data
- Plain-language AI prompt on every model
- Save to a dashboard in one click
Every control, every property
Step order rules, exclusion steps, property breakdowns, custom formulas and SQL for the questions no model anticipates.
- Breakdown by any property or segment
- Formulas that combine metrics
- SQL and export for your own tools
Three teams, three questions
Example scenarios built on the sample data above.
Fix the checkout drop
Conversion fell for two weeks and nobody knew which step was responsible.
- Funnel shows cart → checkout at 54%, down from 61%
- Filter to mobile: 42%
- Paths show abandoners reading the shipping policy
Result: Free shipping threshold shown in the cart; the step recovers to 60% in the next comparison.
Find the activation moment
Trial users who pay look the same as those who do not — until you look at what they did in week one.
- Retention by return event: "Created report" vs any visit
- Interval: sign-up → first report, median 2.4 days
- Segment: trials with no report after 3 days
Result: Onboarding email on day 3 to the segment; week-1 retention rises from 44% to 54% across cohorts.
Make the reviews visible
The product page converts worse than its traffic suggests.
- Scroll heatmap: 22% never reach reviews
- Element table: image clicks that go nowhere
- Compare date ranges after moving reviews up
Result: Reviews moved above the gallery; product → cart improves from 52% to 57% on mobile.
Behavioral analytics questions
Still have a question?
Chat with the team behind TapCub — we usually reply within a few hours on business days.
Do I need a tracking plan before I start?
No. Autocapture records clicks, submits, page views and route changes from the first page load. You name the events that matter later, and history is backfilled for the last 30 days. Revenue and sign-ups come from the SDK when you are ready.
Why does the funnel start at 2,400 and not 48,213?
48,213 is every visitor in the sample. The funnel starts with the 2,400 who viewed a product. Funnels are about a specific journey, not the whole site.
What is the difference between N-day and range retention?
N-day counts people who returned exactly on day N; range counts anyone who returned between day 1 and day N. Range is always higher. TapCub shows both with one switch and labels which one you are reading.
Are heatmaps a separate script?
No. Click and scroll maps are built from the same autocaptured events as everything else, on every page, with no sampling and no extra load.
Is this on the free plan?
Yes. Funnels, retention, paths and heatmaps are included on Free with 6 months of data retention; Pro keeps 12 months and VIP 24 months or more. See pricing.
Can I query the raw events?
Yes. The pro mode includes formulas and SQL, and events can be exported. See data collection.