Blog: Your Marketing Looks Profitable. Your Revenue Disagrees. Here’s Why.
Key Takeaways
- Platform metrics and business results can both be accurate and still tell opposite stories:
When cost per lead is flat and ROAS is climbing but revenue is declining, the problem usually isn't bad data. It's that the metrics are measuring the right thing at the wrong level of detail. - Aggregate reporting hides the numbers that matter most:
Rolling daily data into weekly, weekly into monthly, and channel-level into topline summaries is designed to simplify, but simplifying is the same operation as making smaller problems invisible. A real decline in a specific slice of the business can vanish inside a healthy-looking average. - The gap between "marketing looks profitable" and "revenue disagrees" is testable:
You don't have to guess whether a channel is generating new revenue or just capturing demand that was already there. A structured test—splitting markets, holding one steady, changing one variable—can confirm a diagnosis before you move the budget.
Table of Contents
There’s a particular kind of confusion Jordan Crawford recognizes immediately, because she’s seen it from both sides of the table. Two sets of numbers that should tell the same story, but they don’t.
On one side: the marketing reports. Cost per lead, flat or improving. ROAS, climbing. By every metric the marketing team’s been trained to report on, things are working.
On the other side: the P&L. Revenue, flat or sliding.
“Our business is either flatlining or maybe even declining a little bit, whereas our board expects us to be growing at a 15% year-over-year clip,” Jordan says, describing a version of this conversation she’s had more times than she can count. “And I, as the client, am frustrated and also confused because look at the data the ad platforms are saying. It says CPL is either flat or even getting a little better. ROAS is up.”
This isn’t a story about bad data. The platform numbers are accurate, in a narrow sense. The problem is what those numbers are measuring and what they’re not.
Why Summarizing Data Can Make Real Problems Disappear
It’s not that anyone’s avoiding the data. The team in this scenario is looking at their numbers. Regularly, carefully, in good faith. The problem is what those numbers are doing before anyone sees them.
Every report is a rollup. Daily numbers roll into weekly, weekly into monthly, channel-level into account-level, account-level into the topline summary that lands on an executive’s desk.
Each layer of rollup is a layer of averaging, and averaging is exactly the operation that can make a real problem disappear.
A decline in one specific, important slice of the business can be small enough, relative to the whole, that it simply doesn’t move the aggregate number. The aggregate looks fine. The report says things are fine. And the slice that’s actually in trouble never gets its own line because nobody asked for one.
This is different from the goal-setting and attribution issues that create the initial gap between platform metrics and business reality. This is what happens after that gap exists: once a misleading number is baked into the reporting, the normal process of summarizing data for an executive audience doesn’t surface it.
If anything, summarizing makes it less visible because summarizing is, by design, the process of making smaller things invisible inside bigger things.
The result is a report that isn’t lying, exactly. It’s just operating at a level of resolution where the thing that matters most doesn’t show up.
An Example: How Healthy Aggregate Numbers Can Hide a Declining Business
Let’s dive into a real client example.
Organic search performance — in aggregate — looked steady, even improving. Most teams would read that as reassuring. But aggregate organic is a blend of branded and non-branded traffic. And here, the company had built up non-branded organic visibility.
That growth was masking the fact that organic branded search clicks (people searching for the company by name, no ad involved) were “rapidly declining.”
Nobody was looking at that number in isolation, because the aggregate looked healthy.
The business had a strong brand and great reviews. People were searching for it by name. But paid spend kept growing in branded search and Performance Max, so increasingly, the business was paying for those searches instead of capturing them for free.
The decline in free organic brand visibility was the signal, hidden inside a number that looked fine.
Why the Warning Sign Came From Outside the Marketing Function
The most telling detail in this story is who raised the alarm.
By the time Jordan’s team got involved, the business had already let go of its in-house marketer. For six months, the reports told a consistent story: paid media performing well, path to growth is more budget. The business followed that recommendation, but revenue didn’t follow the increase in spend.
It was the company’s private equity firm — looking at the business from the outside, focused on the P&L — that called it. And, Jordan says, “They were right that that budget wasn’t being spent in the right way.”
This is the core of why this matters: the gap between “marketing looks profitable” and “revenue disagrees” can be completely invisible from inside the marketing function.
Every number that function watches can look fine at the same time the business is struggling.
How to Test a Budget Hypothesis Without Risking the Whole Budget
Jordan’s team had a clear hypothesis: too much budget flowing to channels capturing existing demand instead of creating new demand. But moving a large share of a client’s budget on a theory carries real risk if you’re wrong.
So before changing anything, they tested it. Split the business’s markets into two groups. In one, Performance Max spend stayed put. In the other, it was turned down. Nothing else in the marketing mix changed during the window. Then they watched revenue in each group, focused on measuring the real-world outcome.
The result confirmed the hypothesis on both fronts. Revenue in the reduced-spend markets didn’t decline, which meant that spend hadn’t been generating revenue that wouldn’t have come in anyway.
And organic brand search activity, tracked through GA4 and Google Search Console, picked back up in those same markets, just as predicted: when paid stopped capturing those branded searches, the free version of those same searches reappeared.
What Happened Next: 14% Increase in Revenue
With the test confirming the diagnosis, Jordan’s team reallocated roughly a third of the client’s budget out of Performance Max and branded search and into Meta and YouTube, channels better positioned to reach people who weren’t already looking for the business by name.
The result: media spend stayed flat. The business didn’t need to spend more to fix this. The money was already there, just pointed at the wrong thing.
Top-line revenue increased 14% year-over-year.
The Question Every Marketing Budget Should Be Able to Answer
This isn’t “Performance Max is bad” or “branded search is wasted spend.” Both are too simple, and the right multiplier for a channel depends entirely on the business. Sometimes branded search really is driving incremental revenue, and defunding it would be a mistake.
The real question is underneath the whole story: is this spend creating revenue the business wouldn’t otherwise have, or is it capturing revenue that was coming in regardless?
Platform-level reporting can’t answer that. It can only tell you whether the platform thinks it deserves credit.
Answering the real question means looking somewhere the platform can’t see: the business’s own revenue, measured against itself, with and without the spend in question.
When marketing looks profitable and revenue disagrees, revenue is usually right. The work is figuring out why the marketing numbers didn’t see it coming, and increasingly, that work is testable. Not something a business has to hope someone notices before it’s too late.
FAQs
How can marketing metrics improve while business revenue declines at the same time?
Ad platforms measure performance within their own ecosystem—clicks, attributed conversions, reported ROAS. These numbers can genuinely improve while the business as a whole struggles, because the platform can only see what flows through it. If a channel is capturing demand that already existed (customers who would have found you anyway), it looks efficient on every metric the platform tracks while contributing nothing to actual growth. The disconnect is that the metrics and the business outcome are measuring different things.
What does it mean when a marketing channel "captures demand" versus "creates demand"?
Demand capture means spending money to reach people who were already planning to buy from you—people searching your brand name, returning customers, high-intent visitors. Demand creation means reaching people who didn't know you, weren't looking for you, or hadn't considered you yet. Both matter, but they're not interchangeable. A channel that's almost entirely capturing existing demand will look very efficient in platform reporting and contribute very little to revenue growth. The only way to tell which a channel is doing is to test what happens when you pull back spend and watch whether revenue follows.
How do you know when to test your budget allocation versus when to just reallocate?
Test before you move significant budget, especially if the hypothesis involves defunding a channel that currently reports strong efficiency numbers. Moving budget based on a theory alone carries real risk. If the theory is wrong, you've disrupted a channel that was actually working. A structured test (holding one set of markets steady while reducing spend in another, with no other variables changing) can confirm or disprove the hypothesis at lower risk. If the business doesn't have enough geographic or customer data to run a clean test, a directional approach—making a smaller change and watching whether the signal moves as predicted—is a reasonable starting point.
