The marketing budget is shrinking. The marketing work is not.
A six-month investigation into how mid-stage app companies are reorganising their marketing functions in response to a quietly brutal change in the unit economics of paid acquisition. The work has not disappeared. It has, in many cases, moved somewhere the spreadsheet does not see.
The investigation began, as these things often do, with a chart. In November 2025, a small analytics firm based in Berlin published an aggregated dataset showing the marketing spend, as a percentage of revenue, of a sample of around two hundred mid-stage app companies in Europe. The dataset covered the period from 2018 to the third quarter of 2025. The chart that summarised it showed an unmistakable trend: marketing spend as a percentage of revenue had peaked in early 2022 and had, since then, been declining steadily. By the third quarter of 2025, it was roughly forty per cent below its 2022 peak, and the curve was still pointing downwards.
This was, on the face of it, surprising. The mid-stage app sector has been, in revenue terms, broadly healthy. The customer-acquisition costs in most of the relevant categories have, by every public account, been rising rather than falling. The conventional response, in a high-CAC environment, is for marketing budgets to rise rather than fall. The chart was saying something different. It was saying that the companies in the dataset were, collectively, choosing to spend less on marketing — at exactly the moment when conventional wisdom would have suggested they should be spending more.
I spent the following six months trying to understand why. I spoke, on the record, to seventeen marketing leads at companies broadly matching the dataset's profile. I spoke, off the record, to several more. I read the published quarterly reports of the half-dozen public companies in the relevant size band. The picture that emerged is, I think, more interesting than the headline number suggests.
What the chart is actually measuring
The first thing to understand about the trend is what it is and is not measuring. The chart shows marketing spend, narrowly defined, as a percentage of revenue. The narrow definition includes paid acquisition, conventional brand marketing, sponsorships, and the salaries of staff explicitly categorised as marketing. It does not include, in most of the relevant accounting standards, several activities that have historically lived at the boundary between marketing and other functions.
The activities at that boundary turn out, in the data I collected, to be the most important part of the story. Content production — when done in a way that genuinely supports the brand and the product narrative — is increasingly classified as either product or editorial rather than marketing. Lifecycle messaging that runs through the product itself is, in most companies, owned by product engineering rather than by marketing. Community work, partnerships, integration work with adjacent products: all of these are, in the modern app company, increasingly likely to live in functions other than marketing.
What the chart is measuring, in other words, is the shrinking of the part of the marketing function that the accounting categories still recognise as marketing. The work that the older definition of "marketing" used to include is, in many cases, still happening. It is just, increasingly, happening somewhere else in the org chart.
"The marketing budget went down because half of what marketing used to do is now owned by other functions. The work didn't disappear. The line in the spreadsheet did."
Where the work has gone
I want to be careful not to overstate this point. There is a real, substantive reduction in some kinds of marketing spend across the sector. Paid acquisition spending has, in particular, fallen sharply at the companies I looked at — not because the companies have given up on paid acquisition, but because the marginal unit economics have, in most categories, turned negative at the spend levels that were previously normal. The companies are spending less on paid because the math no longer works at higher spend.
What is more interesting is what has happened to the work that the old marketing function used to do beyond paid acquisition. In rough order of how often I saw the change made:
Content has, in many companies, moved into product or editorial. The articles, guides, and reference material that used to be produced by the marketing content team are now produced, in many cases, by an editorial function that reports either to product or directly to the founding team. The change is not, on the surface, dramatic — the articles still look like the articles always looked. The reporting line, however, is meaningfully different, and the editorial decisions are made closer to the product than they used to be.
Lifecycle messaging has moved into product engineering. The onboarding emails, the activation prompts, the dormant-user reactivation campaigns — most of which used to be the responsibility of a marketing operations team — are now, in many of the companies I looked at, owned by the product engineering team. They are still being shipped. They are now being shipped by the same team that ships the rest of the product, in the same release cycle, against the same metrics.
Brand has, more often than I expected, been recentralised under the founding team. The brand work that used to be done by a dedicated marketing brand team is, in several of the companies I looked at, now done directly by one of the founders, often working with a small number of trusted freelance designers. This is, I want to acknowledge, a change that does not scale beyond a certain company size. At the sizes I was looking at — broadly, between thirty and two hundred employees — it is, however, more common than the conventional wisdom would suggest.
Partnerships and integrations have moved to business development. The work of building relationships with adjacent products, which used to be partly a marketing responsibility, is now almost entirely owned by business development teams that report into either product or commercial. This is, in my view, the change that is most clearly an improvement. Partnership work is, structurally, closer to product than to marketing.
What this means for the marketing function
The narrower marketing function that has emerged, at the companies I looked at, is doing a substantially different job from the one it was doing in 2022. The narrower function is responsible for: positioning, messaging, the marketing-site copy, the marketing-site design, the small amount of paid acquisition the company still runs, PR, events, and the analytics work that supports all of the above. It is no longer responsible for content production, lifecycle, brand identity, or partnership work.
This narrower function is, by the numbers, smaller. The marketing teams I encountered were, on average, somewhere between forty and sixty per cent smaller than they had been at the same companies three years earlier. The work they were doing was, however, considerably more focused. The people in the narrower function were, in most cases, more senior than the people in the broader function had been, and were doing more of their own thinking and less management of contracted-out execution.
This is, in my conversations with the marketing leads themselves, a change that they regard as broadly positive. The narrower function is, by their own account, more strategically valuable to the business than the broader one had been. The reorganisation that moved content, lifecycle, brand, and partnerships out of marketing was, in most cases, painful at the time. The function that emerged on the other side of the reorganisation is, by the accounts of the people running it, a better function.
What this means for the rest of the industry
I want to draw two implications from the pattern I have described above, with the appropriate caveats about generalising from a sample of seventeen companies.
The first implication is that the conventional benchmarks for "marketing spend as a percentage of revenue" are, in the mobile-app sector at least, in the process of becoming meaningless. The benchmark of "ten to twenty per cent" that was widely used in the 2018-to-2022 period assumed a particular definition of marketing that no longer matches what marketing departments are actually doing. A company in 2026 that is spending six per cent of revenue on marketing, narrowly defined, may well be doing more total marketing-relevant work than a 2020 company that was spending fifteen per cent on the same line. The benchmark, on its own, does not tell you anything useful.
The second implication is more strategic. The reorganisation I have described — moving marketing-adjacent work into the functions that own the underlying capability — is, I think, going to continue. The mid-stage companies that have made the change are, on most measures, in better operational health than the ones that have not. The change is, I expect, going to keep spreading across the sector over the next two to three years.
What this leaves us with is a marketing function that is smaller, more strategic, and more clearly bounded than the one we are used to thinking about. The marketing budget will continue to look like it is shrinking. The marketing work will continue not to be. The interesting question, for anyone trying to understand how an app company in 2026 is actually doing, is no longer how much it is spending on marketing. The interesting question is which function owns each of the activities that used to be marketing, and how well that function is doing the work.