Merchant Center sale prices for PMax: discount in the feed or keep regular price?

There are two camps on this one. Some say put the sale price in the feed and let the sale annotation do the work, so the ad shows the discount and the bidding sees the real price people pay. Others say keep promos out of the feed, because short discount spikes teach the bidding that a product converts well at a price you won't keep, and then CPA jumps when the feed goes back to regular price.

I lean toward regular price as the default, and sale price only when a discount runs longer than about two weeks. Short flash sales mostly just churn the learning. Has anyone actually measured this on their own store, or is it mostly gut feel?

Two weeks is a reasonable cutoff, but I'd tie it to conversion volume. If a product only gets two conversions a week, a flash sale won't move the bidding much anyway. For the price itself, use sale_price with effective dates, so the price field stays at the regular number and the sale annotation still shows.

Volume matters, but margin is the other half. A 20% promo on a thin-margin product can wipe out profit even if the bidding handles it fine, so the real test is whether the sale price still clears your target CPA. If it doesn't, keep the sale out of the feed, however long it runs.

Theo's margin test works best when you set it as a rule before the sale starts, because that's when people tend to talk themselves into a promo. Before launch, check whether the discounted price still covers your target CPA with room left for profit. If it doesn't, keep the feed at regular price. After the sale, compare conversion value per click against the two weeks before it, not overall CPA. Sale weeks often overlap with seasonal demand, and what you really want to know is whether the discounted product paid for itself.

Depends on whether you bid on value or on conversions. With tROAS, sale orders log lower values, so the target effectively tightens during the promo and bids drop right when you want volume. With Maximize Conversions, it mostly just shows up in your reports.

Agreed on tROAS. The drop hits every sale order, so it affects the whole campaign, not just one SKU. Before blaming the feed, check what your purchase tag actually sends. If it reads the cart total before the discount is applied, or pulls the list price from a data layer variable, tROAS learns from the wrong number and bids tighten even when the feed is correct.

Pull a handful of sale orders from your GA4 purchase events and compare the value field with what the customer really paid. If they differ, fix the tag first, because no feed setting will rescue bad values. If they match, the cleaner setup is a custom label on sale items and a separate listing group or campaign for them, so their lower ROAS doesn't drag down the rest of the account.

Checking the tag first is right, but I'm not sure the separate campaign holds up for most stores. It only works if the sale items can still hit the roughly 30 conversions in 30 days Google suggests for stable Smart Bidding. A two-week promo on a handful of SKUs rarely gets there, so the split just starves both campaigns. My rule of thumb is to split the sale items out only if they clear that volume on their own. Otherwise, keep them in the main campaign with a custom label and accept the short ROAS dip.

Jun, I don't think a sale listing group will isolate the ROAS dip. In Performance Max, listing groups inside one campaign share the same bid strategy and learning pool, so whatever the algorithm learns bleeds across all of them. Only a separate campaign gets you a clean read, which is why your volume test matters.