The funnel model
Ad spend to sessions to orders to GMV, split by traffic source. A brand-agnostic South African benchmark you can bend: change the rate at any step and see what it is worth in rand.
The funnel by traffic source
Spend → sessions → orders × AOV = GMV → ROAS. Slice width at the top is each channel’s share of sessions; at the orders and GMV rows it is that channel’s share of orders and GMV.
ATC rate × cart CTR × checkout conversion break down each channel’s conversion rate (orders / sessions). Volumes are an illustrative 4-week benchmark. Hover, tap or pick a channel to highlight it.
Move a lever
Each change is relative: +10% on a 5.0% ATC rate takes it to 5.5%. The funnel, the table and the totals all update.
The numbers and how they are built
View the data as a table
Source: illustrative, rounded assumptions for a typical South African ecommerce site, generated 2026-10-26. Not pulled from, or traceable to, any single brand’s analytics — every figure is a round-number benchmark meant as a starting point. “Google Paid” stands in for paid search plus shopping / Performance Max style channels combined. “Other” covers every remaining channel (unassigned, referral, email, display and so on).
ATC rate = sessions with add-to-cart / sessions. Cart CTR = sessions reaching checkout / sessions viewing cart. Checkout conversion = sessions that purchase / sessions reaching checkout. Conversion rate = orders / sessions. AOV = revenue / orders. GMV = modelled order revenue. ROAS = attributed revenue / ad spend.
Ad spend and ROAS are illustrative, built off an assumed benchmark ROAS for paid channels, not real media billing. Paid Social spend (R75,000 over 4 weeks) is back-solved to give roughly 2x ROAS on its attributed revenue; Google Paid is set at 5x. Swap in your own spend when adapting the model for a real brand.
In a live model the three rates don’t multiply exactly into orders: ATC sessions and cart-view sessions can differ, and a session can hold more than one order. Here they are kept simple and consistent so the assumptions are easy to audit. The scenario sliders keep each channel’s bridge ratios fixed, so at 0% every number matches the baseline. Slice and card widths below the top row use a small minimum-width floor so the smallest channel stays readable, which can very slightly compress the largest channel relative to its true share.
A personal reference kept in the open. The benchmark rates will be tightened as the Metrics definitions get pinned down. Nothing here is a forecast for any specific business.