The Funnel, Explained for a Store With Real Numbers

A funnel is the ordered list of steps a buyer passes through - arrive, view a product, add to cart, start checkout, pay - with a count beside each step. The drops between counts are where money leaks. Building yours takes one sitting with the dashboards you already have, and reading it takes one skill: comparing each step's drop with the steps around it, and with the same step last period. This guide builds a funnel from illustration numbers, shows the standard readings, and names the limits - a funnel shows where, never why. The why comes from recordings, surveys and reading the page like a stranger.

Illustration, round numbers, one month: sessions 5,000; product views 2,600; adds to cart 340; checkout starts 210; orders 138. As step rates: 52 percent of sessions reach a product; 13 percent of product viewers add; 62 percent of adders start checkout; 66 percent of starters pay. Overall conversion: 2.76 percent. Already the shape speaks: the add-to-cart step is the narrow neck of this funnel - not because 13 percent is a failing grade in the abstract, but because it is the smallest multiplier in this store's own chain, so improvements there compound through every later step.

A drop is not automatically a defect. Many product viewers were browsing, comparing, or landed on the wrong product - their leaving is the store working as a shop, not failing as a machine. Judge a drop by three things: its size relative to the funnel's other drops, its movement over time, and the intent of the audience at that step. The checkout-start-to-order drop deserves the harshest reading because intent there is proven - those visitors chose a product and began paying. A fall at that step is nearly always friction, cost shock or failure, and it outranks a larger but softer drop earlier.

The blended funnel hides the actionable versions of itself. Build the same five lines for mobile and desktop separately, and for your largest traffic source separately. Small stores routinely discover that the desktop funnel is healthy and the mobile funnel collapses at one specific step - often the add-to-cart or the payment form - which converts a vague 'mobile converts badly' into a single page to fix. Segmentation is what turns the funnel from a poster into a tool.

Service and lead businesses run the same shape with different steps: visit, service page view, enquiry form start, form submit, qualified conversation. Email capture has its own small funnel: saw the offer, clicked, submitted, confirmed. Whatever the business, the discipline is identical - name the steps in the order a person meets them, count each, and refuse to optimise a step you have not counted. The leak bench walks each stage of the standard funnel in a full guide of its own.

Counts do not explain themselves. A falling add rate might be price, photos, a stock-out in the popular size, a slower page, or a traffic shift to colder visitors - the funnel narrows the search; it does not end it. Pair every funnel review with one qualitative instrument: a handful of session recordings at the weakest step, or the month's support messages read with the funnel open. Numbers choose the page; evidence from people chooses the fix.

A funnel decays quietly. A new campaign sends traffic to a collection page instead of products, and the sessions-to-product step drops for reasons that are really about routing. A theme update renames a step in analytics and the add rate appears to halve overnight. Once a quarter, walk your own funnel as a customer - phone in hand, real product, all the way to the payment screen without paying - and confirm the steps your report names still exist in that order. Then check the counts for step definitions that tools may have changed in an update. Ten minutes of walking the funnel protects every conclusion drawn from it, and it regularly surfaces a defect - a vanished size chart, a coupon field error - that no dashboard would ever name.