Revenue up 70%after the marketing stackgot smaller.
A business selling connected health hardware in the United States was losing sales while paying for more marketing tools every year. The client is not named here and the figures are given as changes rather than amounts. Both come from the client’s own sales system and cover every channel the team ran, which makes them a business result rather than an attributed one, and also means no holdout separates the work from everything else happening that year.
- +70%
- revenue in the client’s own sales system, across every channel
- -20%
- the cost of acquiring one customer
- One
- data lake under the reporting, in place of manual updates
- No dates
- the summary names no period, so none of this is annualised
More tools every year, and less idea what any of them did.
A stack that grew by addition
Tools arrived one at a time, each solving the problem in front of somebody that quarter, and none of them left. The bill grew, the overlaps grew, and the number of people who understood the whole thing stayed at one or two.
Reporting kept alive by hand
Numbers moved between systems through manual updates, so two reports on the same week disagreed and both were defensible. When the underlying figures cannot be trusted, a planning meeting turns into an argument about whose export is right, and your market goes undiscussed.
Sales falling while acquisition got dearer
Fewer sales at a higher cost per customer is the pairing that ends budgets. The stack made it worse in a specific way: targeting, messaging and measurement were all constrained by what the tools could pass between each other.
Build the floor first. Buy media on top of it.
One place for the marketing data
A data lake went in underneath the reporting, built with an analytics partner, so the numbers stopped depending on somebody remembering to refresh a sheet. Nothing about the media buying changed at this stage. The point was to make the next decision arguable on evidence.
The stack cut back to what earned its place
Overlapping tools came out and the customer data platform already in the business took over moving data between what remained. A smaller stack is cheaper, and more than that it is knowable: fewer joins between systems means fewer places where a number can quietly go wrong.
Testing turned into a routine
Bidding, creative and audience changes went through a structured test-and-learn cycle rather than landing as one-off edits. Results were read on the new reporting, which is the only reason the cycle meant anything.
The internal team trained to run it
The engagement included training on the rebuilt stack. Handing back a system nobody in the business can operate buys you a dependency, not a capability, and the second year costs more than the first.
Two figures, no period, and one of them off the chart.
Revenue appears as an index rather than in currency: the starting point is set to 100, so the size of the change is visible without the amount. Both figures come from the written summary of the engagement, which names no dates.
The 20% fall in cost per customer stays off the chart on purpose. Revenue rose and cost fell, so the two move in opposite directions, and a bar shows length rather than sign. Putting them on one scale would read as two increases. The larger gap in this page is the period: the summary gives no dates, so nobody can tell whether 70% took six months or three years, and those are different businesses. Treat the pair as the direction of travel rather than a rate, and ask for the window before quoting either figure. One more limit matters more than the dates. This is the whole business counted in the client’s own system, over a stretch when the team ran every channel, and no holdout ran alongside it. So it says what the business did while the work happened. It does not separate what the work caused from what a new product, a price change or a good season would have delivered anyway.
Count your tools, then count the people who can run them.
A stack grows by addition because every single purchase is defensible on its own day. The cost of that arrives later and lands somewhere nobody is watching: in the joins between systems, where a number changes shape on the way from one tool to the next. If two of your reports disagree about the same week, you have already paid it.
Order matters more than the tool list. The reporting went in before the media buying changed, so the results of the next change could be read. Do it the other way round and you spend money on tests you cannot settle, which is how accounts end up with strong opinions and no evidence.
One caution about this page. It rests on a written summary rather than an export, and the summary gives no window, so 70% has no rate attached to it. It is here because the sequence transfers to your account even where the figures cannot. Ask for the workings on a call and you will get this same caveat in the same words.
This work was delivered by MTA Group, which Zero Fluff Digital is part of, by the same specialists who would work on your account. Both figures were read in the client’s own sales system, which counts money taken rather than conversions an ad platform claims, and they cover every channel the team ran. What reached this page is a written summary of that reading, with no period attached to it. The client is not named and the figures are given as changes rather than amounts, at their scale rather than in their currency. There is no third-party study to link to here, so treat this as weaker evidence than the cases that carry one, and ask us for the workings on a call. The reviews behind our work are public and verified on Clutch, where clients rate the work rather than the agency describing itself.
Measured across a period the source never names.
Your reports disagree about the same week.
30 minutes, no deck. You leave knowing which joins between your tools change a number on the way through, and which tool you could switch off tomorrow.
Book a callNext case studyAudioteka: 7M organic TikTok views in 12 months →