The conversation happened at a shared workspace in west London on a Thursday morning in April, over coffee and a plate of small pastries we mostly did not eat. The programmatic lead — she asked to be called Astrid — works at a mid-market UK DTC brand doing roughly £28m of annual direct-to-consumer revenue. She has been in role for three and a half years and made the decision to exit open display in Q2 2024, two years before this interview.
The decision
Ravi Chandrasekhar: Let's start with the simplest version. When and why did you stop buying display?
Astrid: Q2 2024. We had been running roughly £340,000 a year of open-display programmatic — mostly retargeting, some prospecting, a small amount of contextual — through two DSPs. The blended reported ROAS looked decent, in the 3-4 range. The MMM-attributed ROAS was closer to 1.3. The gap had been visible for eighteen months before I acted on it. What made me act was a specific piece of internal analysis showing that the display line's contribution margin per pound of spend was the weakest of any line on our marketing P&L, and that the gap was widening quarter on quarter.
I ran a six-week holdout — turned display off entirely in three of our six sales regions, kept it on in the other three, measured the difference. The difference was, on any read of the data, within the noise band. Turning display off did not measurably reduce revenue in the holdout regions. It reduced spend by a substantial margin. The economics were, at that point, indefensible.
I brought the analysis to my CMO in late April 2024. She authorised the shutdown at the end of that meeting. We paused the display programme at the end of Q2. We have not resumed it.
What she replaced it with
Ravi: What did the £340,000 go on?
Astrid: Not much of it went anywhere. Perhaps £80,000 went into a small expansion of our paid social prospecting — an area where we had, at that point, defensible incrementality data. About £40,000 went into a curated-marketplace CTV allocation, which we treated as a test rather than a commitment. Perhaps £30,000 went into a partnership with a specific publisher whose audience matched ours particularly well.
The rest — about £190,000 out of the original £340,000 — was returned to the CFO as spend saving. This is the part most industry conversations about "reallocating from display" tend to miss. The category assumes that spend leaving one line must go into another line, because the total marketing budget is fixed. In our case, the spend leaving display largely did not need to be replaced. It had not been producing enough incremental revenue to justify its own existence. Removing it improved the overall profitability of the marketing function.
Ravi: How did the CFO react?
Astrid: Better than I had expected. She was, at the time, in the middle of a broader push on marketing efficiency, and my willingness to hand back £190,000 of unproductive spend — rather than defending it or finding somewhere else to put it — was, in her framing, a signal that the marketing function was capable of governing itself in ways that reduced her workload. Our budget conversation the following year was easier than any I had had with her before.
"The industry-wide assumption that a mature marketing programme must be running open display is, on our two-year experience of not running it, unsupported. There is no line in our results — none — that indicates the absence of display has cost us anything material. There is a substantial line — the £190k of recovered budget — that indicates the presence of display had been costing us."
What she watches for
Ravi: Are there conditions under which you would resume display?
Astrid: Yes, but they are quite specific.
The first is if I saw evidence that curated marketplaces, in our category, had matured to a point where the delivery quality was verifiably comparable to the walled-garden alternatives. We have run small tests in this space over the last twelve months. The tests have been encouraging but not, so far, conclusive. If in the next twelve to eighteen months I saw a specific curated-marketplace path that produced incremental ROAS above our threshold, on a repeatable holdout basis, I would open a small budget line for it and grow from there.
The second is if I saw a substantive change in the walled-garden pricing dynamics that made our current allocation less efficient. If Meta CPMs went up 30% overnight, or LinkedIn's cost per qualified lead deteriorated meaningfully, the relative economics of open display would improve and might, at some point, justify re-entry. I do not consider this scenario particularly likely in the next twelve months, but I watch for it.
The third is if my MMM partner produced evidence of a specific brand-tracking effect that display had been providing and whose absence was now measurable. So far, our brand-tracking survey work shows no measurable change in awareness or consideration attributable to the display exit. If that changed, and the change was large enough to matter, I would revisit.
What she'd tell a peer
Ravi: Last question. If a peer running open display asked whether they should exit, what would you tell them?
Astrid: I would tell them to run the holdout. Not the audit — the holdout. Turn display off in a defined region for six weeks, run it in the control, compare. This is the single piece of evidence that will settle the question for them, one way or the other, and it costs meaningfully less to run than most people expect.
If the holdout shows display is producing incremental revenue at a defensible ratio, keep running it. My analysis was our analysis, on our category, in our specific market conditions. It does not generalise to every brand, every category, every stage of a marketing programme. There will be brands for whom the display line is defensibly incremental. Those brands should keep spending.
If the holdout shows what our holdout showed — a difference within the noise band, with a spend cost that is not defensible against the measured effect — then the honest read is that the line is not doing what the industry's default narrative says it does. In that case, the right move is to exit. Not to reduce, not to optimise, not to renegotiate. To exit. The category, in my experience, does not reward incremental improvement when the underlying incrementality is close to zero. The value is in stopping.
Most of my peers, when I have made this argument in private conversation, tell me that their internal politics would not survive a decision to exit open display. This may well be true. It is not, on the underlying commercial evidence, a good reason to keep spending. It is a reason to have a difficult internal conversation about what the display line is actually producing. That conversation is worth having, even if it does not lead to an exit. My advice, for anyone reading this, is to have the conversation.