Marketing11 min read
Performance Max ROAS dropped? 7 checks before you change anything
admin
It is Monday, Performance Max ROAS fell from 4 to 2.6 last week, and the founder wants an answer by lunch. Cutting tROAS (your target return on ad spend), slashing budget or rebuilding the campaign can each make it worse.
When Performance Max ROAS dropped this hard, the cause usually sits in one of 7 places, and some are not real drops at all. PMax spends one budget across Search, YouTube, Display, Gmail, Discover and Maps, so check in this order before you touch a setting.
Check conversion lag first, because a recent ROAS dip is often conversions that have not arrived yet. Google says conversions can be reported up to 90 days after the click, while spend shows up in full straight away.
Google Ads reports a conversion against the date of the click, not the date of the purchase. So the last 3 to 7 days always look weaker than they will a week later. Google's own help page on conversion lag reporting says this can make CPA look inflated and ROAS look deflated.
What to check:
Say your brand's typical buyer takes 3 days from click to order (an illustrative number). Judging yesterday's ROAS means judging a day whose revenue Google Ads has not finished counting.
If ROAS fell overnight and spend dropped with it, suspect tracking before strategy. Google says a sudden drop or spike in tag fires for the conversion action your campaign bids on can reduce serving and spending.
That line comes from Google's page on Performance Max conversion tracking and delay. On Shopify, check anything that touched the storefront that week: a theme change, a checkout app swap or a new consent banner. A broken tag does two kinds of damage: your reported ROAS drops, and Smart Bidding learns from bad data.
What to check:
We covered the Meta side of this in Meta pixel not firing on Shopify. The same theme update can break both tags on the same morning.
Look at change history next, because Google lists recent setting changes as the first reason Performance Max performance moves. Bid strategy, budget, conversion goals, audiences, location, language and ad scheduling edits can all shift results.
Google's Performance Max troubleshooting page lists 10 reasons for fluctuation. Changes you made sit at the top. Pull up Change history, set it to the 14 days before the drop and write down every edit with its date. Then lay the dates next to the daily ROAS line and see which edit, if any, lines up.
No. Google says changes to budgets, targets, creatives or targeting never remove historical data or trigger a "Learning" status. That status appears only for structural changes, and it never lasts more than 7 days.
According to Google's page on Smart Bidding learning, 3 things trigger learning:
| Trigger | Example |
|---|---|
| New strategy | The bid strategy was created or reactivated |
| Composition change | Campaigns added to or removed from a portfolio strategy |
| Goal change | The conversion goals or actions you optimise for changed significantly |
That does not mean a target change is harmless. The same page says it can take 1 to 2 conversion cycles for results to settle at a new ROAS goal. Google also advises against several target changes inside one conversion cycle. So the forum classic, where an agency switches bid strategy 3 times in a month and ROAS slides, is a stability problem, not a learning-reset problem.
Check your top sellers in Merchant Center, because a product that is out of stock or disapproved stops serving. When PMax loses its best converters, it moves spend to products that convert worse, and ROAS falls.
Google's Shopping troubleshooting page also warns that editing IDs, titles, descriptions or product identifiers in the feed can cause fluctuations. Its suggested check works for PMax with a feed too:
Stock is where a lot of spend leaks quietly. A size that sells out on Thursday can keep ads running on the variants nobody wants, or keep the feed showing a product the warehouse cannot ship. We wrote a full playbook on stopping ad spend on out-of-stock products for both Meta and Google.
Open the channel performance report, because a ROAS drop often comes from spend moving to a weaker channel rather than from every channel getting worse. Google built this report to show where PMax spends across Search, YouTube, Discover, Gmail, Display, Search partners and Maps.
Google announced channel performance reporting in April 2025, with a table of clicks, conversions and cost by channel plus diagnostics for channel-specific issues. Search terms reporting is now available for all Performance Max campaigns, per Google's August 2025 visibility update.
What to check:
Sometimes, and it can hide a drop rather than cause one. Brand searches convert well, so when PMax picks them up, its ROAS looks strong while it takes credit for buyers who were already coming.
If your brand Search campaign ROAS fell at the same time PMax ROAS rose, or the reverse, look at the search terms report for brand queries. Google added an option in January 2025 for retailers to apply brand exclusions to Search text ads only, while keeping branded traffic for Shopping ads. When you exclude brand, expect PMax ROAS to fall. That is the honest number showing up.
If none of the checks above explains it, look outside the account. Google lists auction dynamics and market shifts among the 10 reasons, and they are the ones you cannot fix with a setting.
Google's page on auction dynamics gives the example of a competitor lowering its own target and pushing your ads off the first page. With a tROAS target, your campaign may serve less until you adjust. Check impression share trends and Auction insights where you have them, and Google Trends for your category terms.
Big sale days move conversion rates fast, and the weeks after can look like a crash. Google's seasonality adjustments let you tell Smart Bidding to expect a conversion-rate change, and they work for Performance Max.
Google says they suit short events of 1 to 7 days and may not work as well beyond 14 days. You also do not need a negative adjustment after the sale ends. Indian festive demand now runs roughly from July to New Year, according to the Clevrr D2C Festive Report 2026. So use them for the actual sale days, not the whole season.
Fix the cause you found, change one thing at a time, and wait 1 to 2 conversion cycles before judging. Google's guidance is to evaluate over 2 conversion cycles or 2 weeks, whichever is longer.
A short order of moves:
Do not rebuild the campaign as a first move. A new campaign means a new bid strategy, which is one of the 3 structural changes that does trigger learning.
Use this table to match what you see to where to look first.
| What you see | Likely cause | Where to look |
|---|---|---|
| Last 3 to 7 days look worst, older days fine | Conversion lag | Conversion lag report, bid strategy report |
| Spend and conversions fell overnight | Broken tracking or policy issue | Conversion tracking troubleshooter, Shopify orders |
| Drop starts right after an edit | Setting change | Change history, last 14 days |
| A few top products stopped getting clicks | Out of stock, disapproved or feed edit | Shopping Item ID report, Merchant Center |
| Spend shifted to Display or YouTube | Channel mix | Channel performance report |
| Brand Search and PMax ROAS move in opposite directions | Brand overlap | Search terms report, brand exclusions |
| Impression share falling, nothing changed | Competitors or demand | Auction insights, Google Trends |
The hardest causes are the ones that sit between systems: a size sold out in the warehouse while PMax kept spending, or a tag that broke after a theme update. Google Ads sees its own data. It does not see your stock levels or your delivery failures.
This is the gap Clevrr AI, an AI performance marketer for D2C brands, is built to cover. Klev, Clevrr's AI agent, watches Google Ads and Meta Ads next to Shopify orders, inventory and deliveries every day, and names the cause of a change in rupees. ShaperCult, for example, had ~₹1.6L run into ads for out-of-stock sizes over 4 days; spend now auto-pauses on stock-outs. If COD is a big share of your orders, also check how much of that PMax revenue survives delivery with RTO adjusted ROAS.
When Performance Max ROAS dropped, the fastest route to the answer is order, not instinct. Rule out conversion lag and tracking first, then recent changes, then feed and stock, then channel mix, and only then blame competitors or demand. Change one thing, then give it 1 to 2 conversion cycles.
If you would rather have the cause waiting for you before the standup, start free with 500 credits and let Klev check your accounts every morning.
Google says the "Learning" status never lasts more than 7 days. Only structural changes trigger it: a new or reactivated bid strategy, campaigns added to or removed from a portfolio strategy, or a significant change to conversion goals. After a new ROAS target, results can still take 1 to 2 conversion cycles to settle, so judge over 2 conversion cycles or 2 weeks.
No. Google's Smart Bidding documentation says changes to budgets, targets, creatives or targeting settings never remove historical data or trigger a "Learning" status. Budget changes take effect immediately. Large jumps can still move results, because the campaign starts bidding into more or fewer auctions, so raise budgets in steps and watch ROAS over a full conversion cycle before the next change.
Only after you rule out lag, tracking, setting changes and feed problems. Lowering tROAS tells Google to chase more volume at a lower return, and Google notes it can increase conversion volume. It does not fix a broken tag or an out-of-stock best seller. If you do change it, move in small steps and avoid several target changes within one conversion cycle.
Most likely conversion lag. Google reports conversions against the click date, and conversions can arrive up to 90 days after the click, while spend shows in full immediately. That makes recent ROAS look lower than it will end up. Check the conversion lag report and compare the same dates again a week later before you change any setting.
Join hundreds of D2C brands using Clevrr AI to automate their growth and efficiency.
Join 500+ D2C founders using Clevrr AI to automate their operations and scale faster.
Get Started Free
