
Costs and pricing
Part of How to plan shopping feeds costs and pricing for a multichannel shop
Shopping feeds return on investment: how to measure it
Learn how to measure shopping feeds return on investment using contribution margin, reconciled sales data, a worked example and monthly reviews.
What to take away
Shopping feeds return on investment is the money left after feed tooling, staff time, platform charges and the goods themselves, measured against the orders the feed influenced.
- Feed software is usually the smallest line. Merchandising labour and platform fees decide the answer.
- Use contribution margin, not revenue, or every channel looks profitable.
- Reconcile three sources (feed platform, ad platform, order system) before you believe any figure.
- VAT treatment can move margin on some product categories, so check the rules before you promise a number.
- Review monthly, because platform and policy changes reset your assumptions.
Define the return before you measure it
Pick the margin you will use
Start with contribution margin per order: selling price minus VAT, payment fees, delivery subsidy and the cost of goods. A £60 order is not £60 of value if £35 goes out the door before advertising. Write the number down once and reuse it, so month two is comparable with month one.
Contribution margin per order
- £60Selling price
- £35Costs before advertising
- £25Contribution margin
Feed software is rarely the deciding cost. Merchandising hours, image work and category mapping usually outweigh the subscription. Our shopping feeds costs and budget guide for England sets out those lines in order of size, which is the sensible starting point before you attribute any return.
Decide what counts as an influenced order
A feed does not always close the sale. Someone may see a product listing, then buy through a branded search or a marketplace. Decide in advance whether those count, and say so in the report. If you count everything, the return looks generous and the next budget round is built on sand.
A simple rule beats a perfect one. Count an order when the customer saw or clicked a feed-driven listing inside the window your ad platform already uses. Label everything outside that window as untracked, not free.
Build the calculation step by step
Collect the three sets of numbers
Source impressions, clicks and cost from the ad platform; feed health, disapprovals and product-level performance from the feed tool; and net revenue and returns from the order system.
Three-source reconciliation
- Ad platformimpressions, clicks, cost
- Feed toolhealth, disapprovals, product performance
- Order systemnet revenue and returns
- Reconcile one month before scaling
- Pull refunds into the sale month
Reconcile them for one month before you scale any figure up. Google publishes feed and policy changes in its Merchant Centre announcements change log, which is worth checking when a metric moves without explanation.
Pull refunds and cancellations into the month the sale happened, not the month the money went back. A period that looks strong can weaken once refunds settle.
Work a labelled example
Suppose a shop spends £900 a month on feed software and merchandising labour, and £2,400 on shopping ads. The feed-influenced orders produce £12,000 of revenue at a 40% contribution margin, so £4,800. Calculate ROI as the contribution margin from feed-influenced orders minus feed and ad costs, divided by feed and ad costs. Subtract the £3,300 of cost and the monthly net return is £1,500. Every figure here is illustrative, not a benchmark.
Track it month by month in a simple sheet. Our shopping feeds budget template in England gives you the columns and the formulas to keep that comparison honest.
Handle VAT and data requests properly
Some product categories are exempt or partially exempt, which changes the margin you calculated. Check VAT exemption and partial exemption guidance before treating a category as standard-rated.
If a customer asks what data you hold on them, subject access request guidance from the ICO explains the timeline you must meet. Put the admin time in your cost line.
Choose the right measurement approach
Match the method to your situation
Choose
- Under 500 SKUs, one channel
- Platform-reported revenue with a monthly margin check
- 500 to 5,000 SKUs, two or three channels
- Reconciled three-source report
- Marketplace plus own site
- Channel-level contribution margin
- Heavy returns category
- Net-of-returns revenue
- New market, first 90 days
- Cost per influenced order only
Avoid
- Under 500 SKUs, one channel
- Full attribution modelling
- 500 to 5,000 SKUs, two or three channels
- Trusting one dashboard
- Marketplace plus own site
- Blended return figures
- Heavy returns category
- Gross revenue reporting
- New market, first 90 days
- Return on investment claims
Set the review rhythm
Review the return monthly, and recheck assumptions whenever a platform changes its reporting or fees. A figure that was accurate in spring can be misleading by autumn. Keep the definition stable and the inputs current.
Fix a date each month to rebuild the figure from source. Keep a short note of what changed: fees, return rates, feed disapprovals or platform reporting. Without that note, a movement in the return is a guess.
Common questions
Is feed software the main cost in the calculation?
No. Subscription fees are usually smaller than the labour behind merchandising and the platform charges on the ads themselves. Count all three or the return will look better than it is.
Should I use revenue or margin in the return figure?
Margin. Revenue ignores VAT, delivery subsidies, payment fees and returns, so it flatters weak channels and hides strong ones. Contribution margin per order is the number that survives scrutiny.
How often should the return be recalculated?
Monthly for the headline figure, and immediately after any platform reporting or fee change. Keep the method fixed so month-on-month comparisons mean something.
What if the feed influences a sale on another channel?
Decide the rule in advance, state it in the report, and apply it consistently. Mixed rules across months make the trend unreadable.



