Why is my Google Ads ROAS dropping?

By Ahmed Imran · Updated June 2026 · 7 min read

Most ROAS drops are not performance drops. The first things to rule out are broken measurement and a shift in your brand vs non brand mix, because both move the reported number while actual revenue holds. Here is the order I check causes in on real accounts.

Did performance actually drop, or did measurement break?

Start here because it is the most common cause and the cheapest to confirm. Tracking degrades quietly. A theme update strips the purchase tag, or a consent banner stops passing signals to Google. Ad blockers and iOS privacy features trim another slice on top. If you sell into the EEA or UK, Consent Mode enforcement that tightened in mid 2025 has been silently cutting reported conversions on incomplete setups, and Google's modeling recovers only part of the gap.

Platform attribution also undercounts by design. On SwimOutlet, in platform ROAS showed 1.75x while Northbeam, a third party attribution platform, showed the true 3.39x. Same spend and revenue, roughly half the credit. I manage to the verified number, because cutting budget on the platform number would have starved campaigns that were paying for themselves twice over.

The quick test: compare Google Ads conversion value to backend revenue over the same window. If that gap widened recently, you have a measurement problem, not a performance problem.

Did your brand vs non brand mix shift?

Blended ROAS is a weighted average of cheap branded clicks and expensive prospecting clicks, so a shrinking branded share drags the blend down even when every campaign performs exactly as before. A growing brand can flatter the blend the same way while prospecting quietly decays underneath.

On Aliava, a premium women's fashion brand, brand search held a 27 percent CTR at 8.5x while a full rebrand ran. Separating brand from non brand kept the picture honest through the transition, and the account finished at $1.88M on an 8.49x blend. Read the two lines separately before you decide anything is broken.

Is Performance Max hiding what actually changed?

PMax folds brand and non brand into one reported ROAS. When it absorbs branded queries it posts a strong number while your true prospecting return sits hidden underneath. Apply brand exclusions and keep a dedicated branded Search campaign. The channel level reporting Google added in 2025 shows where conversions actually come from.

Product skew hides drops the same way. On Autobuffy, an auto parts retailer, Google Ads scripts surfaced zombie products that spent without selling and overindex products soaking up budget inside PMax. Rebalancing that spend carried the account to $1.5M at 6.89x. If a handful of bestsellers were propping up your blend and they slipped, that is your drop.

Are you paying more for the same clicks?

Sometimes the auction just got more expensive. Benchmark data put average CPC up roughly 12 percent year over year in 2025, with most industries rising. Check Auction Insights for new competitors, and compare against last year rather than last month to rule out seasonality. Watch broad match drift too: under Smart Bidding it expands into looser queries, and your cost per qualified visitor climbs without any single setting looking wrong. If conversion rate held while CPC rose, this is your cause.

Has creative or product fatigue set in?

Assets that have run unchanged for months earn weaker engagement, and Smart Bidding pays more to win the same conversion. On the product side, a bestseller going out of stock or a wave of feed disapprovals removes your most profitable Shopping traffic with no warning in the campaign view. Check Merchant Center diagnostics weekly. A disapproved bestseller does more damage than most bid changes.

On Texas Hill Country Olive Co, the feed and the tracking both needed a cleanup before PMax could optimize toward real purchase value. Once it could, the account ran $467K at 5.36x.

Did a bid strategy change choke your volume?

Raising target ROAS reads like a demand for efficiency, but the algorithm hears permission to buy much less traffic. Set the target above what the account has ever delivered and volume collapses while the surviving spend looks great. Google's guidance is clear that an aggressive target limits traffic, and moving a target more than roughly 15 to 20 percent at once can push the strategy back into learning. Step targets gradually, then hold for two to four weeks of stable data before judging the result.

Is ROAS flat while profit drops?

ROAS measures revenue, not margin. If the sales mix shifted toward discounted or low margin products, or shipping and supplier costs rose, profit falls while the dashboard stays green. I push ecommerce clients toward POAS, profit on ad spend, by feeding margin adjusted conversion values back into Google. A 5x ROAS on a 15 percent margin product can lose money while a 3x on a 60 percent margin product earns plenty.

How do I diagnose a ROAS drop in order?

This is the sequence I run on every audit, cheapest check first.

• ›1. Reconcile platform conversion value against backend revenue for the same window. Fix tags and consent first.
• ›2. Split brand from non brand and read each line over 90 days.
• ›3. Open PMax with brand exclusions plus channel and product level reports.
• ›4. Review Auction Insights and the search terms report for CPC pressure and broad match drift.
• ›5. Check asset age, then stock and feed status in Merchant Center.
• ›6. Read the change history for target and budget edits in the last 30 days.
• ›7. Rebuild the margin math. If profit is the problem, fix conversion values, not bids.

If the drop still resists explanation, send the account my way. I run this exact pass as a free Google Ads audit and send back findings, not a pitch. I have managed US accounts for over eight years on a flat monthly fee, never a percent of spend, so the only number that matters to me is the verified one.