A person types the business’s name into search. By the time the ad appears, someone or something has already given them that name. I would want to know what happened before the click, especially if the report is about to credit the advert for the entire sale.
The reported journey can be accurate and still leave the budget question open. If the paid result had not appeared, would the person have found the shop through the organic result? Would a competitor have taken the visit? The conversion column does not show us the version of the afternoon that never happened.
That makes brand-search returns worth examining rather than automatically celebrating or dismissing. Existing intent is part of the result. The business needs to decide what it paid to add.
Credit and additional sales are different questions
Suppose a hypothetical customer hears about a brand from a friend, searches its name and clicks the first relevant result. If that result is paid, the advertising report may record a conversion. We cannot establish from that sequence alone whether the ad created an additional purchase, redirected a purchase from the organic result or helped prevent a competitor taking the visit.
Those possibilities call for different decisions. Simply removing brand-search spending from the report would hide potentially useful activity. Combining it with broader acquisition can hide the difference between serving existing demand and reaching someone who did not already know the business.
Start by separating brand and non-brand reporting, with attention to how searches are classified. Google's brand-exclusion documentation describes controls for selected brand traffic in supported campaigns. The available controls depend on campaign type and should be checked before changing settings. Google's brand-exclusion guidance.
What a well-known experiment can and cannot tell us
Blake, Nosko and Tadelis studied paid search through large-scale experiments at eBay. Their findings challenged the assumption that the sales attributed to search ads necessarily represented additional sales, including for brand queries. The research paper.
I would resist carrying eBay’s answer straight into a smaller company’s account. A large, recognizable marketplace has circumstances another business may not share, and this is an older study. What I would carry into the meeting is its challenge to the report: how did we establish what would have happened without the spending?
Google describes one controlled approach in its Conversion Lift documentation, comparing exposed and withheld groups. Access, sufficient data and suitable study design matter. A quick switch-off followed by a before-and-after comparison can be confounded by seasonality, promotions or changes elsewhere. Google's lift explanation.
An experiment needs a question specific enough to influence action. A hypothetical retailer might ask whether reducing brand-search coverage in a suitable test setting changes total purchases enough to justify the saved spend. That is different from asking whether the paid-search dashboard records fewer conversions when fewer ads run. The latter result would say little about purchases that moved to another route.
Specify the business outcome, the comparison, the observation period and the conditions that would make the result too uncertain to act on. Have someone with appropriate measurement expertise assess the design and the available volume. A test that cannot detect a commercially meaningful difference may leave the business with a result that sounds reassuring but settles nothing.
The study may disappoint the meeting. It may not separate the alternatives clearly enough to justify a change. That is inconvenient when a decision was promised for Friday, but an inconclusive test does not become a useful yes or no because its scheduled final day has arrived.
Review before making the budget decision
Record the brand-query spend separately. Examine the search results, including relevant competitor advertising and the business's organic presence. Check new and returning customer results where the data supports that distinction. Define the decision the team needs to make and whether a credible controlled test is feasible.
If it is not feasible, keep that uncertainty in the recommendation. Explain the plausible reasons for maintaining coverage and the evidence that would justify changing it. High attributed returns are useful information, but they cannot carry the whole argument.
A useful paid visit deserves consideration. The recommendation needs to explain why buying it was worth the money, with the uncertainty still visible. “The dashboard says the return is excellent” leaves the most interesting part of the decision where it started.