Problem
Google Ads ROAS dropped: how to fix it
Return on ad spend is revenue divided by cost, so it falls when revenue falls, when cost rises, or when the measurement of either changes. Before adjusting bids, work out which of the three moved. A measurement change and a genuine performance decline look identical in the reports and need opposite responses, and acting on the wrong one usually makes the account worse.
By CartKernel · Last reviewed 2026-09-07
Does this look familiar?
- Reported return fell while store revenue over the same period barely moved
- Spend is flat but revenue in the account fell across every campaign at once
- Conversion volume held while the average conversion value dropped
- The decline started on a specific day rather than drifting over weeks
- One campaign fell sharply while the rest of the account was steady
- Return looks worse in a recent view but improves when the date range is extended
Causes, ranked
Why it happens, most common first
Check them in this order. The first two account for most cases we open.
- most common
Conversion measurement changed underneath the reports
A consent banner update, a checkout change, an expired tag, a new attribution model or a switch in the conversion actions counted in the goal column will all move reported revenue without anything happening in the market. This is the first thing to rule out because it is common and it invalidates every other reading.
- most common
The traffic mix shifted away from high intent queries
Broader matching, a new campaign, a feed rewrite or the end of a brand campaign changes which searches you buy. More top of funnel traffic converts less often and at lower value. Reported return falls even though performance on each type of traffic is unchanged.
- common
Discounting or product mix moved the average order value
A promotion, a shift toward lower priced lines, or best sellers going out of stock reduces revenue per order. Because bidding is priced against value, a lower average order value cuts return directly. Checking revenue per order alongside conversion rate separates this from a traffic problem.
- common
Costs rose in the auction
Seasonal competition, a competitor expanding, or a bid target loosened to chase volume raises the price of the same click. Return falls as the denominator grows. Auction insights and the cost per click trend by product type will show whether this is the driver.
- common
The site or the offer changed
A slower page, a checkout error, a shipping threshold moved, a payment method removed or an out of stock landing page all reduce conversion rate for traffic that was already good. Advertising is charged for the click regardless, so return falls without any campaign change.
- occasional
Campaign learning was restarted
Editing bid strategies, merging campaigns, changing conversion goals or making large budget moves puts campaigns back into a learning state. Performance is unstable for a period afterwards, and stacking further edits during that window extends it.
The fix
In this order
Each step is something you can do today. Do them in sequence; skipping ahead is how a review fails twice.
Prevent it next time
- Keep a shared release log covering the site, the tag setup and the ad account with dates
- Alert on daily conversion counts and average order value, not only on spend
- Review the top revenue products for stock and approval status every week
- Change one bidding variable at a time and record when the learning period started
Reconcile against store revenue first
For the weeks before and after the drop, compare reported ad revenue with store revenue attributed to paid search in your own reporting. If store revenue held while the account fell, treat this as a measurement problem and stop making bid changes.
Read the change history for both accounts
Check the Google Ads change log and any recent releases on the store, including theme edits, app installs, consent tool updates and checkout changes. Match dates against the first day of the decline. Most drops have a cause sitting in one of those two logs.
Split the metric into its parts
Chart clicks, cost per click, conversion rate and revenue per order separately over the same period. Return is the product of those, and only one or two will usually have moved. That tells you whether the problem is traffic, price, site or basket size.
Segment by campaign type, brand and product group
Look at whether the fall is account wide or concentrated. A single product group or a single campaign falling points at feed, stock or landing page issues. An account wide fall points at measurement, seasonality or the auction.
Check availability and price on the products that carry revenue
Identify the products that produced most of the revenue before the drop and confirm they are in stock, approved, priced as expected and landing on a working page. Best sellers going out of stock is a frequent and easily missed cause.
Repair measurement before touching bids
If tracking is implicated, fix the tag, the consent configuration and the conversion action settings, then allow a clean week of data. Bidding decisions made on broken data compound the problem, because the models learn from the same wrong numbers.
Make one bidding change and hold it
If performance genuinely declined, adjust the target in a single step toward what the traffic can now deliver, or reduce budget on the weakest product groups. Then leave the account alone long enough for learning to settle before making a second change.
Set the baseline you will judge against
Write down the current figures for cost per click, conversion rate, revenue per order and blended acquisition cost. Recovery is judged against those, not against the peak, which may have included conditions that no longer exist.
When to get help
It is time for help when the drop persists after measurement has been verified and the obvious site and stock causes are ruled out, particularly if margin is tight enough that a few weeks of weak return matters. At that point the useful work is modelling contribution per product, deciding which lines can carry advertising at current prices, and rebuilding campaign structure around that answer. It is a merchandising and finance exercise as much as an advertising one.
Free. We reply within one business day.
How long should I wait before reacting to a fall in return?
Long enough to separate noise from a trend, which for most stores means a full week compared against the same weekday pattern, and longer for accounts with low daily order counts. Reacting to a two day dip usually restarts learning and creates the instability you were trying to fix.
Does raising the target return fix a declining return?
It raises the reported figure by cutting volume, which can be the right trade if the account is unprofitable. It does not address why performance fell. Use it as a deliberate decision about how much revenue you are willing to give up, not as a diagnosis.
Could the drop be caused by an attribution model change?
Yes. Moving between attribution models or changing the conversion window redistributes credit across campaigns and days, which can make one campaign look much worse and another much better with no change in reality. Check when the model was last changed before investigating anything else.
Should I compare year over year or against last month?
Use both. Month over month catches recent breakage, year over year catches seasonality that month over month makes look like a decline. If the two disagree, the year over year view is usually the more honest read for a store with a strong seasonal pattern.
Services that fix it for good
Related problems, answers and terms
Stop the leak.
A free Growth Analysis finds what is broken and ranks it by the revenue it costs you.