What phantom ROAS is, and why almost every account has some
When a buyer who already knows your brand searches for your company name, they intend to find you. Google shows your ad above your organic listing. The buyer clicks the ad, converts, and Google credits the conversion to the ad. The claimed ROAS for that brand-search click is enormous because the conversion rate on intent-loaded brand traffic is high. The incremental revenue, the revenue you would not have gotten without the ad, is much smaller. The same buyer would have clicked the organic listing two pixels below if the ad were not there.
This is not Google's fault. The attribution is correct: the buyer clicked the ad, the ad got the credit. The mistake is scanning claimed ROAS as if it equals incremental ROAS. They are different numbers. Most accounts treat them as the same number, optimize against the larger one, and budget against revenue that was already coming in.
The audit below tells you how much of your claimed ROAS is phantom and what to do about it. It takes about an afternoon. The structural fix takes longer.
Step 1. Identify every campaign and channel where brand keywords are bid on
Open Google Ads. List every active campaign. For each one, ask: are brand keywords (your company name and trademarked product names) being bid on, either explicitly through keywords, or implicitly through Performance Max signals?
Common places brand traffic gets bid:
- A dedicated brand campaign. Often labeled "Brand", "Brand - Exact", or by company name. The keywords are explicit. Easy to identify.
- A general search campaign with branded keywords as part of broader keyword lists. Look for branded terms inside campaigns labeled by category, not by brand. The cleanest accounts separate these. Many do not.
- Performance Max without brand exclusions. If you have an active PMax campaign and no brand exclusion list configured at the campaign level, PMax is bidding on brand searches. You can verify this in the Insights tab under search themes; brand terms appear in the high-converting categories.
- Smart Shopping or Standard Shopping with no negative brand keywords. Less common, but Shopping campaigns occasionally serve against brand-related queries depending on feed structure.
Mark every campaign that bids on brand. Those are your phantom-candidate campaigns.
Step 2. Quantify the brand-attributed revenue share
Set the date range to last 30 days. For each brand-bidding campaign, pull spend and conversion value. Add the total. Divide by total Google Ads conversion value across the account.
The result is the percentage of your claimed Google Ads revenue that is sourced from brand bidding.
Common findings on accounts that have not been audited recently:
- Brand campaigns producing a disproportionate share of claimed return while consuming a smaller share of spend.
- Performance Max producing the highest claimed ROAS in the account while brand contribution is not separated.
- A general search campaign that looks like a clean acquisition campaign but contains a hidden brand keyword in one ad group, distorting the campaign-level number.
Write down the percentage. It will not be the incrementality number, but it tells you the upper bound on phantom credit. If brand-bidding campaigns produce 40 percent of claimed revenue, the phantom share is somewhere between zero (every conversion was incremental, which is rare) and 40 percent (every conversion was already organic-bound, which is also rare). The truth is usually in the middle.
Step 3. Run the brand incrementality test with an account-appropriate measurement window
Pick the campaign that bids most heavily on brand. Pause it for two weeks. Compare ground-truth revenue (from Shopify, Stripe, or whichever billing system holds the actual money) for the two weeks before the pause against the two weeks during the pause.
Three patterns show up:
- Revenue holds steady. The brand campaign was claiming credit without creating a material revenue difference. Quantify the gap from the test rather than using a universal incremental ROAS.
- Revenue drops materially. The brand campaign was producing real incremental revenue, usually because a competitor is bidding against the brand auction or because your organic listing has slipped below the fold on mobile. The campaign is doing structural work and should run.
- Revenue drops slightly. Some brand bids were incremental, some were not. The campaign needs surgery, not a full pause; tighten the bid ceiling and narrow the keywords to the specific brand terms where competitor encroachment is happening.
The two-week test is imperfect. Seasonality, paused remarketing audiences, and other variables introduce noise. But the test is directional and an order of magnitude better than scanning claimed ROAS as if it equals incremental ROAS.
Step 4. Add brand exclusions to Performance Max
Performance Max needs explicit brand exclusion lists at the campaign level. Otherwise it bids on brand searches because brand searches convert. Open the PMax campaign, go to Settings → Brand Exclusions, and exclude all variations of your company name, product line names, and any partner brands you do not want PMax pursuing.
A brand-exclusion change can alter delivery and PMax-reported ROAS. Compare a defined baseline with attribution, customer status, orders, refunds, and collected revenue before estimating incremental effect.
Do not reverse or keep the change from the platform number alone. Evaluate the test design, uncertainty, delivery, qualified buyers, and reconciled commercial outcomes before reallocating spend.
Step 5. Restructure brand bidding into a disciplined campaign
Brand bidding is not always wrong. The discipline is structural:
- One dedicated brand campaign. Manual CPC or Target Impression Share. Tight bid ceiling. Narrow keyword list scoped to actual brand terms (not brand-plus-product variants, which are general acquisition).
- Bid the campaign only when a competitor is encroaching or when your organic listing is unstable. If neither condition holds, pause it. The cost of running brand defense should be measured and intentional, not perpetual.
- Performance Max with brand exclusions configured. Verify the exclusions are taking effect by checking the Insights tab two weeks later; brand search themes should drop materially.
- A tracking view (custom dashboard, Looker Studio, or whatever your team uses) that separates brand-attributed revenue from non-brand. The headline ROAS number is one of two: brand or non-brand. Scanning them as a single blended number hides the phantom problem.
The claimed account-level ROAS will be lower after the restructure. The incremental ROAS, the number that actually drives commercial decisions, will be more honest.
What this audit will not tell you
The brand-search audit does not tell you whether your non-brand campaigns are structured correctly, whether your conversion tracking is accurate at the event level, whether your landing pages are converting cold traffic, or whether your bid strategies are right for each campaign's stage. Those are separate marketing services. The bid-strategy audit covers one of them; the 20-minute version is the entry point.
The full account marketing services, the one that produces a prioritized fix list ranked by revenue impact, is the Conversion Marketing Plan. Brand-search auditing is one chapter inside it. The CSO covers the rest.