Why is my ROAS dropping?

Before you change anything, prove the fall happened in the orders and not only in the report. Ad platforms count conversions they can see, and what they can see changes when consent banners, browser rules, app versions or tracking tags change.

Compare three sources for the same week: orders in the store admin, revenue in your analytics, and conversions in the ad platform. If store revenue held steady while the ad platform's number fell, you have a tracking problem, not a performance problem. That happens after a theme update, a new consent tool, a change to a tag setup, or a checkout change that moves where the purchase event fires.

Then check the attribution settings themselves. A shorter conversion window, a change from data driven to last click, or a switch in the model will move reported return without a single order changing hands. So will a change to which conversion action is counted as primary, which quietly happens when someone adds a new conversion event.

Also check for double counting in the other direction. If reported return jumped a month ago and has now come back down, the drop may be a correction rather than a decline.

Only when the store's own revenue confirms the fall is it worth looking at causes. Half the panic about return on ad spend is a reporting artefact, and changing bids to chase an artefact makes the real numbers worse.

Return on ad spend is an average, so it moves whenever the ingredients move, even if nothing performed worse.

Product mix is the usual cause. Adding new products, or a stock shortage on a hero line, shifts spend to items with different prices and margins. A campaign that used to spend on a 200 dollar bundle and now spends on a 40 dollar single item will show a lower return while performing exactly as before.

Audience mix is next. Brand searches convert at a rate cold traffic cannot match. If brand demand fell, or you added a prospecting campaign, or a competitor started bidding on your name, the blended figure drops because the share of easy traffic dropped. Split branded from unbranded and look at each separately.

Campaign type mix matters too. Moving budget between a campaign that mostly reaches people who already know you and one that reaches new buyers changes the average without changing any individual result.

New customer share is the honest version of this. If the store is deliberately buying more first time buyers, reported return falls and that can be the right trade, as long as the second order actually arrives. Measure it as cost per new customer alongside return, and the picture stops looking like a decline and starts looking like a decision.

The ads did not necessarily change. The store may have.

Price is first. A price rise, the end of a promotion, or a competitor undercutting you in the Shopping results changes the click through rate and the conversion rate at the same time. Both push return down while the campaign settings sit untouched.

Stock is second, and it is easy to miss. When a best seller goes out of stock, its ads stop and the budget moves to whatever is still available. The account looks the same, the spend is the same, and the return is lower because it is now buying different products. The same happens when a supplier delay pushes delivery dates out and the page still promises the old date.

Site changes are third. A theme update, a new app on the product page, a popup that covers the buy button on small screens, a slower page after adding a review widget or a chat tool. Each one lowers the conversion rate on traffic you are still paying full price for.

Shipping and thresholds are fourth. Raising a free shipping line above the typical order, or adding a surcharge, quietly moves carts into abandonment. Check the cart to checkout rate for the weeks either side of the drop, because that single ratio separates a traffic problem from a checkout problem.

If tracking, mix and the store all check out, the market is the remaining explanation, and it is the one you have least control over.

Auctions get more expensive when more advertisers want the same impression. That happens in a category's peak weeks, during big retail events, when a large retailer starts stocking what you sell, or when a competitor raises its budget. You will see it as a higher cost per click with a stable conversion rate. That combination is the signature of a pricier auction rather than a worse campaign.

Seasonality also changes intent. The same search means something different in a category's quiet month than in its busy one. People browse in the quiet weeks and buy in the busy ones, so the same spend converts differently.

Demand shifts show up in impression share and search volume. If impressions are flat but clicks and orders both fell, interest cooled. If impressions fell and cost per click rose, you are being outbid.

When the auction is the cause, the answer is rarely to bid harder. It is to widen the margin you have to bid with: raise the order value with bundles and thresholds, lift the conversion rate on the pages taking the traffic, or shift budget to product groups where you still hold an advantage. A campaign wins on the economics behind it more often than on the bid in front of it.