The Line Item Nobody Questions

I sat in a Q4 budget review a few weeks ago, watching a media team defend their plan to a room full of executives, line by line. Paid social got questioned. CTV got questioned. The affiliate program got questioned twice. Then they reached brand search, and the room went quiet. Best return on the page. Nobody asked it a single thing.

I have sat in enough of these to know that silence is not confidence. It is a blind spot wearing a good number.

Brand search looks unbeatable because of what it is actually measuring. Someone types your company's name into a search bar because they already know who you are and they have already decided you are worth a click. Paying to show up next to your own name does not create that decision. It arrives after the decision is made and takes credit for the last few seconds of it. Something else did the convincing: a recommendation, a previous purchase, an ad they saw three weeks ago on a channel nobody in that room was defending. Brand search just happened to be standing there when the decision turned into a search.

Calling that your best channel because it wins every dashboard comparison is like calling the cashier your best salesperson because she rings up every sale in the store. She touches every transaction. She closes none of them. The comparison flatters the wrong person, and the budget follows the flattery every time.

That part, once you put it to someone plainly, usually lands. The part that took me longer to see clearly is why nobody questions it in the room in the first place. Brand search does not just perform well. It reports beautifully, and a number that reports beautifully survives scrutiny that a messier, more honest number never gets. Nobody defends brand search in these meetings because nobody has to. It defends itself, with a number that was never measuring what the room assumed it was measuring.

Here is the harder problem, and it showed up in that same meeting an hour later.

Someone pulled up two vendor reports side by side. A remarketing partner and a display partner had both bid on the same cart abandonment inventory that week. One customer saw an ad from each, bought once, and both platforms logged the sale. Multiply that across a full quarter of overlapping inventory and the vendor totals add up to more revenue than the company actually did. Everyone in the room had seen some version of this before. Nobody had a clean answer for it.

Neither vendor was lying. Each one reported exactly what it saw: an impression or a click, followed by a purchase, inside its own window. Asking those two reports to reconcile is asking two people who each watched one minute of the same movie to agree on how the whole thing ends. They are not wrong about their minute. They simply never saw the rest of it, and neither platform is built to.

This is where most teams make the expensive move. Someone gets assigned to reconcile the numbers, loses a quarter chasing a spreadsheet that will never balance, and the room quietly concludes that measurement itself is broken. Measurement is not broken. The expectation that platform reports should sum to the truth was never reasonable to begin with. Two honest reports can each be accurate about what they individually observed and still be incompatible with each other, because each one is only watching its own slice of the customer's path.

If you are in that room and someone on the finance side asks the question that actually ends these meetings, take it seriously: if two of your numbers cannot both be true, why should I believe either one? It is a fair question, and defending the discrepancy is the wrong response to it. Name it instead. Tell them no single platform can see the whole path, by design, and that reconciling platform reports against each other will always fail, for a structural reason rather than a tracking one. Then tell them what you are doing about it: measuring the outcome from a vantage point outside any single platform's view, rather than asking the platforms to agree with each other.

That reframing changes what you do with brand search too, and it points to the one action worth taking out of this issue. Run a holdout. Turn brand search off in one market, or for a few weeks, and watch what happens to total conversions, not to the brand search line by itself. If the number barely moves, you have just learned what that line item is actually worth, and it is probably less than the dashboard has been telling you. If you cannot run a holdout yet, do the smaller version first: separate brand from non-brand in your reporting so the two stop getting defended as if they were a single channel earning a single number.

The channel that survives every budget review without a single question is usually the one that most deserves one.

Talk soon,
Jeff Greenfield
CEO Provalytics

Jeff is the co-founder and CEO of Provalytics. He'll mention it occasionally. He'll also tell you when it's not the right tool.