Signals

Eleven ways a sale fails, found without you looking

Nobody watches a thousand recordings. Whence reads them for you, flags the visits where something went wrong, ranks them by what went wrong, and attaches the evidence.

Repeated clicks on a link that does nothing, flagged as a signal, in the Whence demo. Fictional store, demonstration data.

What gets flagged

Every signal carries its proof

A signal is never a bare number. It arrives with the session that produced it, so the first thing you do is watch it rather than argue about it.

The Frustration screen in Whence: what friction costs at checkout and in carts, then dead clicks, rage clicks, excessive scrolling and quick backs, and where each one happens.
In the Whence demo. Fictional store, demonstration data.

Behavior

High Rage click

The same spot clicked again and again in quick succession. Something looks pressable and is not.

Medium Dead click

A click that changed nothing on the page. Usually a control that broke quietly.

Medium Excessive scrolling

Up and down a full screen at a time, over and over, within half a minute. They were looking for something they could not find.

Low Quick back

A page opened and left for the previous one within ten seconds. What they clicked was not what they expected.

Low Shallow scroll

The visitor stayed but saw less than a tenth of the page, never the part you built the argument on.

Medium Long visit, no purchase

Real attention that produced nothing. Often the most instructive recording of the week.

Abandonment and faults

High Checkout abandon

They reached payment and left. The most expensive moment to lose someone.

Medium Cart abandon

A product was added and never bought.

High Form abandon

A form was started and given up on. The recording shows which field lost them.

Critical Script error

Your page broke during the visit. You would otherwise learn this from a complaint, or never.

High Resource error

An image or file failed to load. Frequently a product photo on the page that was supposed to sell.

Sorted by cost, not by count

Fifty shallow scrolls matter less than one broken checkout. Each signal has a weight, so the visits to watch first are the ones where the most went wrong.

  • Each signal has a severity, from critical to low, and a weight that sets the order of the visits to watch.
  • A completed purchase cancels the false alarm. If the visitor bought, the abandonment flag on that visit is withdrawn rather than left to inflate the list.
  • Repeats are grouped, so one broken button appears once with its count instead of two hundred times.

A flagged session

  • Signal checkout abandon
  • Severity high
  • Source Meta Ads
  • Device mobile
  • Also seen 2 dead clicks
  • Evidence session attached

It knows what a shop is

Generic replay tools see clicks. Whence recognizes the moments that decide whether a visit was worth anything.

  • Added to cart and reached checkout are read from the store’s own events, and orders come from the store itself, not guessed from page addresses.
  • The funnel is built from them, so a drop between two steps is a place you can go and watch.
  • Abandonment means something precise here, rather than a heuristic about time on page.
Funnel · Last 7 days
Visited
Added to cart
Reached checkout
Ordered

A list you can actually finish

Signals are meant to be closed, not admired. Each one carries a status so a week's work leaves a trace.

  • Open, investigating, resolved or ignored. What you mark resolved leaves the open list.
  • On the Decide plan, record what you changed and measure whether the signal moved.
  • New signals surface on their own as sessions accumulate. There is no report to run.

A week's list

OpenInvestigatingResolvedCritical open

What signals here do not do

  • Suggestions, not decisions. Each signal comes with a short note on where to look, and the AI quick actions, such as Conversion blockers, sum up what the evidence shows on plans with AI requests. Whether to change anything, and what, stays your call.
  • Detection needs volume. On a store with a handful of visits a day, patterns take a while to become patterns.

Find out what broke last week

One month free, no card, no call with anyone.

Start free trial