Mapping the retail funnel from impression to repeat purchase

Two dashboards, one retailer, the same month: the media team reports a 4.1% conversion rateconversion rateThe percentage of visitors or prospects who complete a desired action (purchase, sign-up, contact form), calculated as conversions divided by total opportunities.View full definition →, the e-commerce team reports 1.9%. Neither is lying. One divides orders by paid clicks, the other by every session including organic, email and direct. Nobody had agreed where the funnelfunnelThe customer journey from awareness to purchase, typically Awareness, Interest, Consideration, Decision, Action, with prospects narrowing at each stage.View full definition → starts, so the two numbers describe different funnels and the meeting is unwinnable. That agreement is what a funnel mapmapUsing software to automate repetitive marketing tasks and campaigns, enabling personalisation at scale across channels like email, web, and social.View full definition → is: which stages exist, what counts as entry to each, who owns the hand-off between them, and which drop-offs are structural rather than fixable. Every other metric in this module plugs into it.

The retail funnel, stage by stage

An omnichannelomnichannelAn integrated approach connecting all customer touchpoints (physical, digital, mobile) into a seamless experience, with shared data and consistent context across channels.View full definition → retail funnel has six stages, not five. Most diagrams stop at a vague "retention" box; the two additions below are where money quietly leaves.

1. ImpressionImpressionThe total number of times an ad or piece of content is displayed, regardless of clicks. Each display counts as one impression, even to the same person.View full definition →: ad or organic content served and, sometimes, seen
2. Traffic: store visit, website session or app open
3. Engagement: product view, size selected, cart add
4. Conversion: order placed
5. Net purchase: order kept, after returns and cancellations
6. Second purchase: a repeat order inside the category's natural cycle

Each transition has a rate. They are not equally diagnostic, and they are not measured on the same population, which matters at stage 5.

Stage 1 to 2: click-through and footfall

CTRCTRClick-Through Rate (CTR) is the percentage of people who click a link, ad, or call to action out of those who viewed it.View full definition → = clicks ÷ impressions. Retail display and social CTR benchmarks sit around 0.5% to 1.5% (estimate, varies heavily by format and platform, as of 2025). Read them off the Meta and Google dashboards rather than a third-party average.

The trap is the numerator. An impression is not a fixed unit: Google's Active View standard treats a display impression as viewable when at least 50% of the ad's pixels are in view for one second, while Meta counts a video view at three seconds. Two platforms can serve the same shopper the same creative and report impression counts that are not comparable. The top of the funnel is a platform definition, not a fact about human attention.

Footfall conversion is the store analog: the share of passers-by who enter. Retail analytics firms estimate 10% to 20% for mall apparel (rarely disclosed, varies by location and season). Door counters count entries, not shoppers. A family of four registers as four, staff stepping out and back register too, and a store on a two-entrance corner double counts anyone cutting through. A footfall conversion rate that jumps the week the sensor was recalibrated did not improve.

High CTR from clickbait creative drags in traffic that dies at stage 3. Judge stage 1 on what survives to stage 4.

Stage 2 to 3: engagement

Cart add rate = cart adds ÷ sessions. General retail lands around 8% to 12% (estimate, Baymard Institute and industry reports, 2024-2025). Whether that figure is honest to compare against your category's median is the benchmarking lesson's problem, and it turns on the denominator.

Sessions are not people. A shopper who browses on a phone at lunch and adds to cart on a laptop that night creates two sessions and one cart add, so the rate reads lower than the behaviour. In stores, this transition is barely instrumented at all: fitting-room entries, associate conversations and product handling need dedicated tracking most chains do not have, so the funnel usually jumps from door count straight to till.

Stage 3 to 4: conversion

CVR = purchases ÷ sessions online, purchases ÷ visitors in store.

Sector estimates (2024-2025, sources including Adobe Digital Economy Index and Statista):

• E-commerce overall: roughly 2% to 3%
• Physical apparel retail: roughly 20% to 25%, since walking in is already high intent
• Grocery e-commerce: often above 10%, mostly planned repeat baskets

Cart abandonment rate = 1 minus (purchases ÷ cart adds). Around 70% of online carts are abandoned (Baymard Institute, ongoing research, see their published data). A few points recovered here often beats more top-of-funnel spend.

Worked example: where's the leak?

A mid-size online apparel retailer runs one month:

• Impressions: 5,000,000
• Site sessions: 50,000 (CTR = 1.0%)
• Cart adds: 5,000 (cart add rate = 10%)
• Purchases: 1,000 (conversion rate = 2%, cart-to-purchase = 20%)

Cart-to-purchase of 20% means 80% of carts abandoned, worse than the ~70% estimate. That is the leak, not the CTR. Pushing more traffic into a funnel that loses four carts in five multiplies the loss. Surprise shipping costs, forced account creation and thin payment options are the usual culprits, and each is cheaper to fix than a bigger media plan.

Stage 4 to 5: returns deserve their own stage

Online apparel return rates commonly run 20% to 30%, higher in size-sensitive lines (industry estimates). Free returns plus size uncertainty produce bracketing: three sizes ordered, two shipped back. This is where funnel optimisation can turn on itself. A checkout nudge that encourages multi-size orders lifts stage 4 and lowers stage 5 contribution at the same time, because reverse logistics, restocking and write-offs on unsellable returns eat the extra margin. If the team is bonused on conversion rate, the funnel pays them for a loss.

There is a timing consequence too. With a 30-day return window, last month's conversions stay provisional for another month, so any comparison of this month's CVR against last month's puts a gross number next to a settled one.

Stage 5 to 6: the second purchase, dated

Repeat purchase rate = customers with 2+ orders in a period ÷ customers in that period. Fashion and specialty retail cluster around 20% to 30% within 12 months; grocery and consumables often exceed 60% because the cycle is weekly.

The window does more work than the metric. Measure repeat rate at 12 months in a category whose replacement cycle is two or three years (mattresses, large appliances) and you report churn that has not happened. Measure at 12 months in grocery and you have thrown away eleven months of signal. Take the window from the category's purchase cycle, which is the calculation the lifetime valuelifetime valueLifetime Value: the total revenue (or profit) a customer generates throughout their entire relationship with your business.View full definition → lesson runs, then hold it fixed so cohorts stay comparable.

Stage 6 is also where the funnel breaks in half. Stages 1 to 4 are a cross-section: everything that happened in March, different people at different points. Stages 5 and 6 only mean anything longitudinally, following one cohort forward. Multiply a March CTR by a March cart add rate by a March repeat rate and the product describes nobody.

The value side set against the cost side (the ratio commonly cited as healthy at 3:1 or better) belongs to the lifetime value and acquisition costacquisition costCustomer Acquisition Cost (CAC) is the total sales and marketing spend divided by the number of new customers gained in a period. It measures how efficiently you grow.View full definition → lessons. What the funnel map owes them is a clean, dated stage 6 to count. Worth knowing that the cost side has drifted: paid acquisition got dearer after Apple's App Tracking Transparency rollout and Google's phased cookie changes, so a ratio that was comfortable in 2020 can be marginal on the same product today.

Why retention beats acquisition as a health signal

• CTR and impressions are exposure. Easy to inflate, easy to buy, weakly tied to revenue.
• Conversion rate connects intent to outcome, but it is one transaction seen once.
• Net purchase and second purchase show whether the model works. A viral month can produce excellent CTR and CVR and still leave nothing behind.

Nike shows the stakes of owning the tail. After accelerating direct selling from 2020, Nike Direct (own stores, nike.com and the app) grew to roughly 40% of revenue, and membership gave the brand stages 5 and 6 on named customers: what came back, what was kept, who bought again. The wholesale half of the same business ends at stage 4. Once product ships to a partner, Nike sees a sell-in order, not the buyer, the return or the repeat. That is the structural cost of handing off the funnel early, and it is why the rebuilding of wholesale relationships in 2024 and 2025 is a genuine trade rather than a reversal: reach at stages 1 and 2 bought at the price of visibility at stages 5 and 6.

Cohort analysis (tracking customers who joined in the same period and watching what share still purchase in month 3, 6, 12) is the standard way to read stage 6 honestly, instead of being reassured by a top line that is only adding new customers over a leaking base.

🎬 [VIDEO: "Cohort Analysis Explained" - youtube.com/results?search_query=cohort+analysis+retention+explained - search for a walkthrough of building retention cohort tables, a core technique for separating real retention health from new-customer growth noise]

A quick note on attribution

Multi-touch attribution (splitting credit across the ad impression, the store visit and the final order) is unsolved at scale, and post-cookie signal loss made it worse. The default windows differ by platform: Meta reports on a 7-day click plus 1-day view basis, Google Ads on a 30-day click window, so both can claim the same order. Summed platform conversions routinely exceed the order count in the finance system. Reconcile to orders first, then allocate, blending last-touch reporting with media mix modelling (statistical estimation from aggregate data). Treat any single-touch number as a hypothesis.

Key Takeaways

• Agree the stage definitions and denominators before comparing any two conversion rates; most funnel arguments are definition arguments in disguise.
• Impressions and footfall are platform and sensor artefacts. Judge the top of the funnel by what reaches stage 4.
• Cart abandonment sits near 70% industry-wide (estimate), and cart-to-purchase is usually the cheapest leak to fix.
• Give returns their own stage: apparel returns of 20% to 30% mean a CVR win can be a contribution loss, and reported conversion stays provisional for the length of the return window.
• Stages 1 to 4 are a cross-section, stages 5 and 6 are cohorts. Do not chain the rates together, and set the repeat window from the category's purchase cycle.