Google Ads conversion value: first-order revenue for bidding, or lifetime value?

Stores with repeat buyers keep going back and forth on what the purchase conversion should send to Google. One camp says first-order revenue, so tROAS optimises for the customer you actually paid to acquire. The other says lifetime value through value rules or offline uploads, so the bidder rewards customers who come back.

I lean toward first-order for bidding and LTV for reporting. Take a supplement store where the first order only breaks even and the second and third carry the margin. Feed LTV into the bidder and it will pay more for every first purchase, including from people who never reorder, so your CPA on new buyers gets fuzzy. Keeping first-order as the bid signal keeps the target honest, and you judge cohort revenue separately.

Has anyone seen LTV value rules hold up when the repeat rate is low, or does it mostly just inflate the numbers?

Fair split, but the calibration side deserves more weight. At low repeat rates, LTV value rules don't so much inflate numbers as learn from a small group. If only 15% of buyers reorder, the model is mostly fitting the people who did, then pays a premium for anyone who looks like them. The bidder chases that predicted value, your new-buyer CPA looks fine on paper, and the cash never shows up.

The first-order approach has a gap too. For a supplement store where order one only breaks even, tROAS on revenue still lets the bidder buy orders that lose money before any repeat happens. If you stay on first-order, set the target against gross margin instead of revenue, and make sure the target itself clears break-even.

What's the repeat rate within 90 days for your best-selling SKU?

A low repeat rate isn't a good reason to drop LTV rules. At 15% reorder, the problem is letting the bidder see a predicted premium for every buyer, when most of them never come back. You can avoid that without giving up repeat value. Keep the base conversion value at first-order margin, then handle repeat value separately, for example through a conversion action or value rule that only fires for customers with a prior order. That way the bidder never guesses at people who didn't return, and your new-buyer CPA stays readable.

The margin point is right, though. Revenue-based tROAS on a product where order one loses money just pays you to lose money faster. Once you target margin, the break-even check is simple arithmetic: target ROAS has to be at least the inverse of your first-order margin, or you're buying losses on purpose.

The math checks out, but I'd leave some room above break-even. Set the target roughly 15% over 1/margin, because tROAS tends to hover around whatever you give it, and even a small dip in conversion rate can quietly push the campaign into the red.

Agreed on the buffer. The 1/margin math only works if the value you send is margin and not revenue, so check that the conversion value column really holds margin after returns and discounts. One detail that's easy to miss in Atlas's setup is that the repeat-order action should be secondary, not primary. Primary actions feed bidding, so a primary repeat action would bring back the predicted premium you were trying to keep out.

Secondary as the permanent answer for the repeat action is where I'd push back. It keeps the predicted premium away from the bidder, which is what you want at a 15% reorder rate. But it also means nothing optimises toward returning customers, even once you have enough repeat volume to trust the signal. For a supplement store with low repeat rates, that's the right call today. It may stop being right once repeat rates climb or one SKU starts showing predictable second orders. So treat primary versus secondary as something to revisit after a few months of real repeat data, not a setting you lock in on day one. Until then, secondary is the safer start, and the reporting still gives you the cohort revenue Priya wants to judge separately.

Agreed on starting with secondary, but I'd put a recurring review on the calendar instead of waiting until it feels wrong. The catch is the conversion window. If it's 30 days and your supplement customers reorder around day 60, most second orders never get counted. The repeat action then looks dead, and you could easily decide repeat buyers don't matter when the window is just too short. So before judging any secondary repeat action, compare the window to how long your reorder cycle actually runs.