I spend a lot of my week inside Shopping feeds and Merchant Center accounts, and there's a pattern I see on nearly every ecommerce account I take over: the account treats every conversion of the same value as the same event. £50 in, £50 in, £50 in. Job done, box ticked, ROAS calculated.
It isn't the same event. A £50 order from someone buying from you for the first time and a £50 order from your best repeat customer are two completely different pieces of business — and if your bidding and your audiences can't tell them apart, you're almost certainly overpaying to reach the customer who was going to buy anyway, and underpaying to find the one who wasn't.
The transaction isn't the story
Your Google Ads conversion action sees a purchase. It doesn't see history. It has no idea whether this order is someone's first ever visit to your site or their eighth purchase this year. On paper, identical. In reality, one of them is acquisition and the other is retention wearing an acquisition costume.
That distinction only exists if you're connecting the transaction back to a customer profile rather than just logging it as an isolated event — order history, purchase cadence, whether they've bought before at all. Most small-to-mid ecommerce accounts I audit aren't doing this. They've got the transaction data. They don't have the history attached to it.
Where this actually costs you money
- Prospecting campaigns quietly count repeat buyers as wins. If your "new customer" acquisition campaign's conversion value includes people who were always going to reorder, its reported ROAS is flattered and its real job — finding people who've never heard of you — looks like it's working better than it is.
- Remarketing lists don't exclude people who've already converted this cycle. Someone who buys every two weeks doesn't need a retargeting ad nagging them three days after their last order. You're paying to remind a loyal customer to do the thing they were already going to do.
- Smart Bidding learns from a blended signal. If your conversion value feed can't tell the algorithm "this one's a first-timer, this one's a repeat," it optimises toward whichever pattern is easiest to win — which is usually more of the customers you already have, not the new ones your growth actually depends on.
What to actually do about it
- Split new-customer value in your conversion tracking. Google Ads supports a new customer acquisition goal for exactly this reason — most accounts running Shopping or PMax haven't switched it on.
- Build a suppression list for recent purchasers in your remarketing setup, sized to your actual reorder cycle rather than a generic 30-day window pulled from a template.
- Check whether your feed and CRM are actually talking to each other. If a customer's order history lives in Shopify and their ad interactions live in Google Ads with nothing joining the two, you're bidding blind on the exact distinction that matters most.
- Recalculate your acquisition cost using only genuinely new customers. If repeat buyers are baked into that number, your real cost of acquisition is higher than you think — sometimes a lot higher.
The point
Two £50 orders. One of them is worth protecting with a suppression list. The other is worth chasing with everything the account has. Until you can tell them apart, you're optimising toward an average of two very different customers — and averages, as ever, hide exactly the thing you needed to see.
Russell Anson runs Compulsion, an independent paid-search consultancy specialising in Google Ads, Shopping feeds and Merchant Center. Eighteen years in, pointed at the only metric that matters. compulsion.agency