Purchase funnel analysis

Purchase funnel analysis is the practice of measuring how shoppers progress through the ordered stages of buying, from landing on a site to completing checkout, and quantifying where they fall away between each stage. Rather than reporting a single store-wide conversion number, it follows the path: how many people view a product, how many add to cart, how many begin checkout, and how many pay. If 20,000 visitors view a product, 8,000 add to cart, 2,400 begin checkout, and 1,440 buy, the funnel exposes that the steepest loss sits between add-to-cart and checkout, not where guesswork would suggest.

It matters because a single headline conversion rate tells you something is wrong without telling you where. Funnel analysis localises the leak. A store that loses people evenly across every step needs a different intervention from one that converts well until a single broken checkout field, and treating those two cases the same wastes effort. The same logic holds whether the funnel is one product page or an entire catalogue.

The most useful version follows the steps people actually take rather than the tidy diagram. Shoppers revisit the cart, abandon and return days later, and leave through unexpected exits. Measuring the real flow, including the routes you did not plan, is what turns funnel analysis from a chart into a diagnostic that connects to average order value and revenue.

There is no single equation for an entire funnel, because a funnel is a chain of steps. You measure it by computing the conversion rate at each transition and the cumulative drop-off across the whole path. The core unit is the step conversion rate: shoppers reaching the next step divided by shoppers reaching the current one.

1. 1

   Define the funnel stages

   List the ordered steps a shopper passes through, from first view to completed payment. Keep the definition stable, because moving stage boundaries mid-measurement makes period-over-period comparison meaningless.

2. 2

   Count distinct shoppers at each stage

   For every step, count the distinct shoppers who reached it. Use distinct counts rather than events, so one person reloading the cart does not inflate that stage.

3. 3

   Calculate step conversion rates

   For each transition, divide the count reaching the next stage by the count reaching the current stage. This is where the leak shows itself: the lowest step conversion is the weakest link in the chain.

4. 4

   Track cumulative drop-off

   Multiply the step conversions together to get the share of viewers who reach purchase. This shows how a few moderate drops compound into a large overall loss across the full funnel.

A worked example: 20,000 view a product, 8,000 add to cart, 2,400 begin checkout, and 600 pay. The step conversions are 40 percent, 30 percent, and 25 percent. The 25 percent from checkout to payment is the weakest transition, so a one-point improvement there returns the most. The cumulative path conversion is three percent, which is what compounds across every step. Funnel analysis stops you from polishing the 40 percent view-to-cart step that already works while the checkout-to-payment step bleeds.

A metric tree maps the funnel onto its sequence of transitions and then decomposes each transition into the friction points that govern it. The headline outcome, end-to-end purchase conversion, sits at the root. The first level is the set of stage-to-stage transitions, because the product of those transitions is the whole funnel.

Each transition then breaks into the specific reasons people fail to advance. The cart-to-checkout branch, for example, decomposes into unexpected shipping cost, forced account creation, and payment friction. This is the level where an intervention becomes concrete: you are no longer trying to fix the funnel, you are fixing the shipping line that surprises 40 percent of carts.

KPI Tree gives each transition a RACI owner, so merchandising is accountable for the view-to-cart step while the payments team owns the checkout step. When a step conversion drops, the platform pushes the change to the owner of that specific transition rather than to a store-wide dashboard nobody owns. The verified impact loop then confirms whether the fix to that step actually lifted purchases downstream, so a local win is not mistaken for a funnel-wide one.

Benchmarks for funnel flow depend heavily on the category, the traffic source, and the price of the item. The ranges below give realistic step conversion expectations for common ecommerce funnel stages, useful as a sanity check rather than a target to chase.

Treat these as bands, not goals. The more important comparison is your own trend over time and the gap between adjacent steps. A single transition far below the others is the signal, regardless of how the absolute number compares to an industry average. Segmenting the funnel by device or traffic source often reveals that mobile checkout, or one channel, drags the whole funnel down.

Improving funnel flow starts with finding the weakest transition and understanding why shoppers leave there specifically, rather than spreading effort evenly across a funnel that is mostly healthy. The discipline is to fix one leak, verify the downstream effect, then move to the next.

KPI Tree connects each transition to the team that owns it and to the downstream outcome it feeds. Merchandising owns the view-to-cart steps, the checkout team owns cart-to-payment, and the payments team owns authorisation. When a step conversion drops, the accountable owner of that transition is notified directly, and the verified impact loop checks whether their fix flowed through to completed purchases, so you can tell a genuine funnel improvement from a leak that simply relocated.

1. 1

   Measuring totals instead of transitions

   Watching only the store-wide conversion rate tells you the funnel is leaking but never where. The diagnostic value lives entirely in the step-to-step conversions.

2. 2

   Mapping the ideal path, not the real one

   Shoppers abandon and return, revisit the cart, and exit through unexpected steps. Analysing only the path you designed misses the routes where people actually leave.

3. 3

   Counting events instead of distinct shoppers

   Using event counts lets one person reloading the cart inflate that stage and distort the step conversion. Count distinct shoppers at every step.

4. 4

   Ignoring device and channel segments

   A funnel that looks healthy in aggregate can hide a broken mobile checkout. Segment the flow by device and traffic source before concluding the funnel is fine.

5. 5

   Changing stage definitions mid-stream

   Redrawing where one stage ends and the next begins breaks every comparison. Fix the stage boundaries before measuring, and keep them stable across periods.