How to Audit Brand Search in Agency-Managed Accounts

When Brand return on ad spend (ROAS) looks strong but blended customer acquisition cost (CAC) is rising, the answer is not to second-guess the agency or defend the existing budget on attribution alone. Brands and agencies must audit underlying query intent, eliminate cross-campaign cannibalization, map competitor impression share, and execute geo-isolated incrementality tests. That creates a shared basis for deciding how much paid Brand creates or protects and what the business should pay for it.

Why Brand Metrics Look Great While Blended CAC Rises

Brand search can report excellent ROAS while the business pays more to acquire each customer. Brand campaigns capture people who already know the company, so strong attributed performance does not tell you how many of those customers actually needed a paid Brand ad to convert.

Brand search is supposed to look good. Someone searching for the company by name already knows the brand and may simply be returning to buy something they had already decided they wanted.

They click the paid ad and purchase. Google Ads credits the Brand campaign according to the account's attribution settings. Cost per click (CPC) can look efficient, conversion rate is strong, and Brand ROAS ends up near the top of the monthly report.

But that does not answer the question a Director of Ecommerce actually needs answered: How many of those customers would have purchased anyway if the paid Brand ad had not appeared?

Consider a scenario where Brand ROAS looks strong month after month while blended CAC (total acquisition spend relative to customers acquired) has risen for three consecutive months.

Both numbers can be accurate. Google Ads is showing where conversion credit went. Blended CAC is showing what the business is paying overall to acquire customers.

That is where the client and agency can end up arguing past each other. The agency points to strong Brand performance. Finance points to rising acquisition costs. Ecommerce asks whether those customers would have clicked the organic result anyway. Competitor bidding may make some paid coverage genuinely valuable, but ROAS alone cannot tell you how much.

The account structure can muddy the picture further. Non-brand demand can enter Brand through broader matching, while branded demand can also appear in Performance Max (PMax) or Non-Brand campaigns. The campaign labeled Brand may not represent all and only paid Brand demand.

So do not start by cutting Brand because CAC increased or defending it because ROAS looks good.

First establish what the account is actually buying and where Brand demand is being captured. Then clean up material overlap and measure incrementality. Only then should the client and agency decide how much Brand budget the business needs.

What Belongs in a Joint Brand Audit Pack

A useful brand audit pack should answer four questions in order: what are people actually searching, is non-brand demand getting into Brand, is brand demand getting into other campaigns, and how much competitive pressure exists on those branded searches. Each step builds on the previous one, so the result is one view of brand demand rather than four disconnected Google Ads reports.

Those answers come from four places: the search terms report establishes the underlying queries, match type analysis shows how the Brand campaign is expanding, PMax and Non-Brand Search reveal where branded demand is appearing elsewhere, and Auction Insights shows who is competing for that demand and how consistently.

Use the same 90-day window for the first three checks so the findings line up. For Auction Insights, use at least a full quarter and break it into monthly views so temporary competitor activity does not get mistaken for sustained pressure.

1. Search Terms: What Are People Actually Searching?

Start with the raw evidence. The search terms report establishes what people searched, which keywords brought those searches into the Brand campaign, what they cost, and what conversion credit they received.

In Google Ads, go to Campaigns > Insights and reports > Search terms and pull the Brand campaigns for the previous 90 days.

Include search term, campaign, ad group, keyword, match type, impressions, clicks, cost, conversions, conversion value, and the account's primary efficiency metric. Add the Keyword column if it is not already visible.

Do not confuse the keyword with the search term.

If the keyword is brand shoes, that does not mean the customer searched brand shoes. The search term is the query Google reports for what the customer actually searched. The Keyword and Match type columns help explain how that search got into the campaign.

For example, Acme Running might have the keyword Acme running shoes, while the search terms report shows Acme Running, Acme trail shoes, Acme returns, and best trail running shoes.

Start classifying those visible searches as core brand, brand + product, support/service, competitor, brand-adjacent, or non-brand category. Do the classification once. The next steps reuse it to see whether demand is sitting in the right campaigns and being reported under the right labels.

There is one limitation: Google does not expose every individual query. Some lower-volume search-term data is withheld or grouped, so this is not a complete ledger of everything the account bought.

What you get from Step 1: a classified baseline showing what visible demand sits inside the Brand campaign. Now the audit can start asking whether that demand is sitting in the right place.

2. Match Types: Is Non-Brand Demand Getting Into Brand?

Take the classifications from Step 1 and roll them up by match type. The question now is whether Phrase and Broad Match are expanding the Brand campaign into demand the business would normally consider non-brand.

Break out spend, conversions, and conversion value across Exact, Phrase, and Broad Match for the same 90-day period. Then use the search-term classifications from Step 1 to understand what sits underneath those totals.

Suppose Acme's Exact Match traffic consists overwhelmingly of core brand searches, while Broad Match increasingly contains searches such as best trail running shoes.

The issue is not simply that Broad Match exists. The issue is that a campaign reported to executives as Brand may now be buying a different type of demand.

A simple rollup makes that visible:

Then compare the mix with an earlier, comparable period:

Brand campaign is now spending a larger share of its budget on searches that are less clearly branded. That can make Brand performance harder to interpret because the campaign is no longer buying the same mix of demand it was in the prior period.

What you get from Step 2: an answer to what is getting into Brand and which match types are responsible.

3. Brand Leakage: Is Brand Demand Getting Into Other Campaigns?

Now reverse the analysis. Step 2 looked for non-brand demand entering Brand; Step 3 looks for branded demand showing up in campaigns the business considers non-brand.

Pull the Performance Max search terms and landing pages report for the same 90-day period and look for the company name, misspellings, domain variations, product brands, sub-brands, and other searches classified as branded in Step 1. That report segments by ad format, so you can see whether a branded query was served a text ad or a Shopping ad, which is the fastest way to confirm what a brand exclusion is actually blocking.

Repeat the exercise across Non-Brand Search campaigns, particularly those using Broad Match.

Suppose Acme's PMax campaign shows Acme Running shoes and Acme trail runners, while Non-Brand Search is also picking up Acme product searches. Some branded demand is now receiving spend and conversion credit outside the Brand campaign.

Next, check why that overlap is possible. Review the brand exclusions, negative keywords, and other relevant controls actually applied to PMax and Non-Brand Search.

When an agency says brand is excluded, check which control was actually applied. Brand exclusions work by attaching a brand list to the campaign. Negative keywords in Performance Max are added at the campaign level or the account level, not inside an asset group.

Check the Shopping setting as well. Performance Max brand exclusions include an option to allow Shopping ads to keep serving on searches that mention excluded brands. With that box checked, a campaign can be accurately described as having brand exclusions applied while Shopping ads continue to appear on branded queries.

Scope matters too. Performance Max negative keywords apply to Search and Shopping inventory only. Brand exclusions apply to Search, Shopping, and YouTube search inventory. Neither control covers every place a Performance Max campaign can serve.

For Search campaigns, brand settings began moving into the AI Max panel on May 27, 2025. Existing brand exclusions continue to work outside AI Max, but adding a new brand list to a Search campaign requires turning AI Max on for that campaign. An agency reporting that brand exclusions are unavailable on a Search campaign may be describing that constraint rather than an oversight.

A campaign called "Non-Brand" does not prevent Google from matching it to branded demand.

Bring the findings into one cross-campaign view:

The distinction matters: PMax spent $85,000, but $7,200 of visible spend is being associated with branded searches. The audit is isolating the portion of PMax and Non-Brand performance that appears to be benefiting from demand already carrying brand intent.

That matters when comparing campaign efficiency. Strong PMax or Non-Brand ROAS means something different if part of the reported performance is coming from customers explicitly searching for Acme.

What you get from Step 3: a view of where branded demand is showing up outside Brand, how much visible spend is associated with it, and whether that overlap is intentional or needs to be addressed.

4. Auction Insights: How Much Competitive Pressure Are You Actually Defending Against?

Once the team knows what brand demand looks like and where it is being captured, Auction Insights adds the final piece: whether competitors are consistently fighting for that demand and whether your own visibility is being affected.

Pull Auction Insights for the Brand campaigns covering at least the previous quarter. Look at the full quarter first, then break it into monthly periods.

Google does not populate the report when impression share is below 10%, so a small or budget-capped Brand campaign can return nothing even while competitors are active in the auction.

Two practical questions matter here. Who consistently overlaps with us, and is that competition affecting our visibility?

Start with competitor overlap:

Then put that next to Acme's own visibility:

StrideCo's overlap spikes in Month 2 and then falls back, while Acme's visibility remains relatively strong. TrailPro is different: its overlap rises steadily through Month 3 as Acme's impression share and top-position visibility decline.

That does not prove TrailPro caused the decline, but it gives the team a specific competitive trend worth investigating and a much stronger basis for defensive Brand coverage than simply saying "competitors are bidding on us."

Auction Insights turns "competitors are bidding on our brand" into something measurable: who is competing, whether that pressure persists, and whether Acme is actually losing visibility.

What you get from Step 4: a read on whether competitive pressure on branded demand is real, sustained, and affecting your visibility.

Together, the four checks show what Brand is buying, where that demand is showing up across the account, and how much competitive pressure surrounds it. If the audit also finds generic traffic inside Brand or branded traffic leaking into PMax and Non-Brand, clean up those boundaries before testing incrementality. Otherwise, the test may measure traffic moving between campaigns rather than what happens when paid Brand coverage is actually removed.

Clean Up Brand Boundaries With Your Agency Before Testing

Once the audit is complete, review the findings with the agency and agree on what needs to change before testing incrementality. The goal is not to audit the agency's past decisions but to get both sides working from the same definition of Brand, the same campaign boundaries, and the same understanding of any overlap that remains.

Go through the audit findings together before changing the account. Some overlap may be intentional, and the agency may have context that is not obvious from the reports. For each finding, ask three questions: Is this intentional? Should it remain? If not, how are we going to fix it?

Frame these as joint optimization decisions, not corrections the agency has been ordered to make. Finding Brand traffic in PMax does not automatically mean the setup is wrong. Ask whether the overlap is intentional, whether it still makes sense, and whether it should remain when Brand incrementality is measured.

Document anything both sides agree to leave in place so there is no confusion later about what the test includes.

Once the changes are made, verify them together. Rerun the search-term, match-type, and leakage checks to confirm Brand demand is flowing where expected and any remaining overlap is intentional.

Then allow the account to establish a usable baseline before testing. Conversion lag, volume, bidding behavior, seasonality, and the size of the changes will determine how long that takes.

Then the client and agency can move to the next question: how much additional business does paid Brand coverage actually create or protect?

Measure Brand Incrementality Before Changing the Budget

Once Brand demand is reasonably isolated, measure what paid coverage actually adds. Agree with the agency on the method, the business outcome that matters, and what result would be strong enough to change the Brand investment decision.

Google Ads attribution cannot answer that. If someone searches Acme Running, clicks the paid ad, and purchases, Google Ads can credit Brand with the conversion. It cannot tell you whether that customer would have purchased anyway.

There are two practical ways to investigate: a geographic holdback when the account has enough scale, or a paid-versus-organic comparison when a controlled test is not practical. In either case, keep the competitive pressure identified in Auction Insights in view.

Option 1: Run a Geographic Holdback Test

A geographic holdback provides the stronger evidence. Keep Brand coverage running in comparable treatment markets, reduce or withhold it in holdback markets, and measure what happens to total business outcomes, not Brand campaign conversions.

For ecommerce, that outcome might be total orders, new customers, revenue, or contribution margin.

Consider Acme Running. After cleaning generic traffic out of Brand and addressing obvious leakage into other campaigns, Acme reduces Brand coverage in selected holdback markets while maintaining normal coverage in comparable treatment markets.

Paid Brand conversions will fall. That is expected. The useful question is what happens to the business overall.

If customers move to Acme's organic result and total orders or revenue remain relatively stable, paid Brand may have been receiving credit for demand that would have converted anyway. If business outcomes decline relative to the treatment markets, that is evidence that paid Brand was adding incremental value.

Do not pick a few convenient states, turn Brand off, and call the difference incrementality. Markets can differ in baseline sales, promotions, customer mix, competitor activity, seasonality, and other media exposure.

Before testing, the client and agency should agree on the treatment and holdback markets, the business outcome being measured, and what result would actually change the Brand budget decision. Define that threshold before seeing the result; otherwise the argument simply moves from "Is Brand incremental?" to "Is that amount of incrementality worth paying for?"

Geo holdbacks are generally more practical for larger ecommerce accounts with enough geographic distribution and transaction volume. With too little volume, the test may be too noisy to support a confident decision.

Google also offers geo-split testing directly. Conversion Lift based on geography supports Search, Shopping, and Performance Max campaigns, and the Experiment Center runs controlled experiments split by user or by geography. Availability depends on account eligibility, and Conversion Lift requires working with a Google account representative.

Option 2: Compare Paid and Organic Behavior

When a controlled holdback is not practical, compare paid and organic behavior for the Brand query groups identified in the audit. The goal is to see whether paid coverage appears to add traffic or mainly shifts clicks from the organic result to the paid ad.

When Google Search Console is linked with Google Ads, use the paid and organic report to examine paid clicks, organic clicks, and combined paid-plus-organic clicks across the same query groups established earlier.

The report covers text ads only. Shopping ad clicks are not included, so a brand running branded Shopping will be comparing an incomplete paid side against a complete organic one.

Organic data begins on the date the Search Console link was created and is not backfilled. A link created specifically for this analysis will not support a before-and-after comparison until enough time has passed.

The question is whether paid coverage adds traffic. When paid clicks increase, does total paid-plus-organic traffic increase too, or do clicks mostly move from organic to paid?

Suppose Acme's paid clicks rise on core Brand searches while organic clicks fall by a similar amount and combined traffic barely changes. That is consistent with click substitution.

But suppose combined traffic increases on Brand + product searches where Auction Insights also showed sustained competitor pressure. That signal is different. Paid coverage may be adding more value where Acme is competing for the click.

Organic results are not limited by campaign targeting. A campaign targeted to one country can still register organic impressions from another, which means combined paid-plus-organic totals can move for reasons that have nothing to do with paid coverage.

Organic rankings, SERP layouts, Shopping results, competitors, promotions, and devices can all affect these patterns. Treat paid-versus-organic analysis as supporting evidence, not proof of incrementality. It can show where substitution appears more or less likely, but only a well-designed controlled test gets closer to answering what would have happened without the paid ad.

Turn Incrementality Results Into a Brand Budget Decision

Once the client and agency have measured what paid Brand actually adds, use that evidence to make the budget decision. The question is no longer whether Brand ROAS looks good. It is how much incremental revenue or margin paid Brand creates or protects and what that value is worth.

Review the results against the decision criteria agreed before the test.

If reducing Brand has little effect on total business outcomes, discuss where coverage or spend can be reduced or tested further. If business outcomes decline meaningfully, decide how much Brand coverage is worth maintaining and at what cost.

Bring the Auction Insights findings into that interpretation. If incremental value is stronger where competitors are consistently active, that may support maintaining more defensive coverage there. If relatively uncontested Brand searches show little incremental effect, those searches may warrant a different approach.

And if the test is inconclusive, treat it as inconclusive. Low volume, conversion lag, geographic differences, promotions, and other noise can prevent a reliable result. An inconclusive test is not evidence for keeping or cutting the existing budget.

Brand ROAS still matters as an operating metric. But the audit and incrementality work give ecommerce, the agency, marketing leadership, and finance something more useful for allocating budget: a shared view of the demand being bought, the competitive pressure around it, and the incremental value the business is receiving for its spend.

That view will change. Broad Match can expand into different queries, PMax can find new paths into Brand demand, and competitors can enter or leave the auction. Make the Brand audit pack a standing part of the agency review and revisit incrementality when those conditions change materially.

The Brand conversation can then stay focused on the question that matters: How much incremental revenue or margin does paid Brand create or protect, and is that value worth what the business is spending to capture it?

Red Flags That Warrant a Deeper Look

Some patterns are worth auditing before the next budget review rather than after it. None of them proves the account is being managed badly, but each one changes what the Brand numbers mean.

Match type creep is the first. When Exact Match falls as a share of Brand spend while Broad Match rises, the campaign reported as Brand is buying a different mix of demand than it was last quarter.

The second is a brand control that is described rather than verified. Brand exclusions applied with Shopping ads still allowed, negative keywords added at the wrong level, or a Search campaign whose brand settings were never moved into AI Max all produce the same reporting line and very different serving behavior.

The third is Auction Insights that is cited but never shared. If competitive pressure is the stated reason for Brand spend, the quarter-level export should arrive with the monthly reporting rather than on request.

The fourth is inconsistent definitions of Brand. If sub-brands, product brands, misspellings, and domain variations are counted differently across reporting and campaign controls, the leakage analysis cannot be reconciled.

The last is reluctance to discuss incrementality testing at all. A holdback is disruptive and not every account can support one, so an agency may have good reasons to push back. The signal is refusing to discuss the question, not declining a specific test design.

Frequently Asked Questions

What should my agency include in a brand audit pack?

A brand audit pack should include search-term data, match type distribution, visible brand overlap in Performance Max (PMax) and Non-Brand Search, and at least a quarter of Auction Insights data. Make it a recurring deliverable so changes in query mix, leakage, and competitive pressure are caught before the next budget review.

How do I review brand without second-guessing my agency partner?

Treat the brand audit as a joint diagnostic exercise rather than an investigation of the agency. Request the data together, agree on what counts as Brand, review the findings before drawing conclusions, and treat gaps as optimization opportunities. The objective is a shared view of the account before either side recommends changing spend.

What are red flags in brand reporting?

Brand reporting deserves a deeper audit when Broad Match is pulling generic demand into Brand, branded searches appear in PMax or Non-Brand without clear controls, Auction Insights is cited but rarely shared, or the agency resists discussing incrementality testing. None proves poor management, but each deserves explanation and verification.