Every July, the same screenshots reach me from every direction. Clients share them in reviews, prospects lead pitches with them, and my feed fills with them. Sale-week ROAS up. Conversion rates up. Revenue records broken. After nearly a decade advising brands and retailers on paid media, I can predict what happens next: the deck travels to leadership, the team collects congratulations, and next year's promotional calendar quietly gains another event.

The question I ask in those rooms is harder: how much of that performance belonged to the media, and how much belonged to the price cut?

Across the organizations we advise, very few can answer that with confidence. And that gap matters more right now than at any point in recent years.

One thing to state clearly before going further: this is not an argument against discounting. Promotions are often commercially rational, and later in this piece I describe the cases where they clearly are. The problem sits in how their success gets measured, with metrics that cannot separate the effect of a price cut from the effectiveness of the marketing around it.

Why Discounting Became the Default Growth Lever in Retail and DTC

Promotional intensity has been climbing across retail and DTC for years, and the scale is easy to underestimate. In many CPG categories, 28 to 50% of retail sales now happen on promotion, and companies spend up to 20% of gross revenue funding those promotions. Consumers have adapted accordingly. Recent Prime Day data reveals that shoppers are increasingly promotion-sensitive, with more than half comparing prices across retailers before buying. Brands trained their customers to wait for deals, and the customers learned.

At the same time, the reporting layer never adjusted. Platform dashboards measure promotional weeks with exactly the same logic as normal weeks. Ads that run during a 30% off window get judged on the same ROAS metric as ads that run at full price, and they win that comparison almost every time. Leading brands are starting to acknowledge the problem: brands are increasingly moving beyond platform-reported ROAS toward incrementality testing, precisely because platform reporting alone cannot explain true business impact.

Until that shift completes, the result is a systematic illusion. Discount-window performance looks like marketing excellence, so budget and calendar space keep flowing toward it. The numbers reward the behavior, the behavior intensifies, and very few people in the room are asked to separate what the media earned from what the margin paid for.

Three Reasons Promotional ROAS Overstates Marketing Performance

The illusion has a mechanical explanation. Three effects stack on top of each other during any promotional window.

The price effect gets credited to the media: When conversion rates jump during a sale, the discount often did most of that work. The ads were present, so the ads get attributed. Platform measurement reflects platform-defined attribution, which means it cannot distinguish between a customer persuaded by creative and a customer persuaded by 30% off. Revenue primarily driven by the price cut gets booked as media performance.

This pattern is well documented beyond promotions. When eBay paused brand search advertising in selected markets, the majority of the “lost” revenue simply arrived through organic channels instead (Blake, Nosko & Tadelis, 2015). Attribution had been crediting ads with journeys that were finishing anyway. Discounts amplify exactly this effect, because they attract precisely the customers who were closest to buying.

The calendar pays for the spike: Promotions pull demand forward. A meaningful share of sale-week buyers would have purchased anyway, at full price, in the weeks before or after. McKinsey's work on promotion analytics makes the same point: promotional impact can only be judged after controlling for stock-up behavior, cannibalization across products, and cross-period effects. In many cases, the promotional window borrows revenue from the weeks around it, books it at a lower margin, and the softer weeks rarely make it into the campaign report.

ROAS counts revenue the margin already paid for: This is the quiet one. Return on ad spend measures revenue, not profit. During a promotion, every conversion arrives with a discount attached, which means the business paid twice for it: once in media and once in margin. The aggregate consequence is stark: McKinsey's analysis of CPG trade promotions found that 59% of them lose money globally, and in the US the figure rises to 72%. A campaign can post its best ROAS of the year while delivering its worst contribution of the year, and the standard dashboard will present that as success.

There is a fourth effect that shows up later. Customers acquired on discount behave differently. Across many of the cohorts we analyze at Crealytics, deal-acquired buyers repeat less often, wait for the next promotion, and deliver materially lower lifetime value than full-price buyers. That pattern will not hold for every brand, but it recurs often enough to deserve testing. The acquisition looked cheap. The customer was expensive.

How Inflated Promotion Metrics Distort Budgets and CFO Trust

None of this would be urgent if promotional reporting stayed inside the marketing team. It doesn't.

Sale-week screenshots travel upward. They shape how boards perceive marketing effectiveness, they anchor next year's targets, and they inform capital allocation decisions that take quarters to reverse. When the flattering version of promotional performance becomes the official version, organizations systematically over-invest in demand harvesting and under-invest in demand creation. That trade-off is invisible in any single quarter and very visible after three years of it.

The prize for correcting it is measurable. McKinsey finds that companies applying advanced analytics to pricing and promotion decisions can increase revenue by 3 to 5% while simultaneously improving profitability, largely because so many promotions fail to generate profitable growth in the first place. The improvement doesn't come from spending more. It comes from finally seeing which spend was doing the work.

There is also a credibility cost. CFOs increasingly read marketing dashboards with a skeptical eye, and they are right to. The moment finance discovers that record ROAS coincided with declining contribution margin, every future number marketing presents gets discounted too. Trust, once spent that way, is expensive to buy back.

And the timing is specific: Q4 budgets are being drafted now. Plans locked in August will carry whatever assumptions this summer's reporting installed. If those assumptions came from unexamined promo-window metrics, peak season inherits the illusion at the highest-stakes moment of the year.

How Marketing Leaders Can Measure Promotions on Profit, Not Revenue

The response does not require new technology. It requires a decision about which numbers are allowed to govern. In the measurement framework we published at Crealytics, the principle is simple: evaluate channels on contribution, not attribution, and give each measurement method the decision layer it can actually answer.

Report contribution, not revenue, during promotional windows. Standardize on metrics that reflect business impact: incremental revenue, contribution margin, cost per incremental conversion, customer lifetime value. Profit-adjusted ROAS changes promotional conversations immediately. Some events survive that lens. Others reveal themselves as expensive ways of buying the appearance of growth.

Measure the full window, not the spike: Judge every promotion across a period that includes the weeks before and after, controlling for stock-up and cannibalization effects. Pull-forward is only invisible when nobody looks for it. This is exactly the question marketing mix modeling answers well: it sees diminishing returns and cross-period effects that campaign dashboards structurally cannot.

Run holdouts during sales, not only in normal weeks: A geographic or audience holdout during a promotional event answers the question dashboards cannot: how much of this revenue needed the media at all? Experiments are the causal layer of the measurement stack for a reason. These are the weeks when incrementality testing earns its keep, and the weeks when teams are most reluctant to run it.

Track cohorts by acquisition context: Separate deal-acquired customers from full-price customers and follow their repeat behavior for twelve months. This single view has changed more promotional strategies than any argument I've made in a meeting.

Agree the governing metric with finance before the event: Decide together, in advance, which number will define success. When marketing and finance judge a promotion by the same measure, the post-event conversation becomes a planning session instead of a negotiation.

Deliberate Discounting vs Default Discounting: The Strategic Difference

I want to be clear about what I am not arguing. Discounting is a legitimate commercial tool, and some of the strongest results I've seen came from promotions designed with full knowledge of the math.

Clearing seasonal inventory ahead of a range change is rational. Using a sharp introductory offer to acquire customers in a category with proven repeat behavior is rational, provided the cohort economics have been tested rather than assumed. Defending share during a category moment when a competitor stumbles can be worth planned margin sacrifice.

The difference between those cases and the July screenshots is that someone did the contribution math before the event, decided the trade-off deliberately, and measured the outcome honestly afterward. Deliberate discounting is strategy. Default discounting, reported through metrics that hide its cost, is an illusion with a budget line.

The organizations that get this right don't promote less in every case. They promote knowingly. And their marketing leaders walk into budget season holding numbers that finance already trusts, which turns out to be worth more than any sale-week record.

A discount can buy you a revenue spike. Only the margin math tells you what you paid for it.

Key Takeaways for Marketing Leaders

· ROAS measures revenue, not profit. During promotions every conversion carries a discount, which is how 59% of trade promotions globally, and 72% in the US, end up losing money.

· Discounts attract the customers closest to buying, inflating attributed media performance. The eBay brand search study showed how much attributed revenue survives when the ads switch off.

· Judge promotions across the full window, controlling for stock-up and pull-forward effects, so borrowed demand can't masquerade as created demand.

· Run holdout tests during sale events, since those are exactly the weeks when attribution flatters the most.

· Agree the governing metric with finance before the promotion. Companies that bring analytics discipline to pricing and promotion decisions grow revenue 3 to 5% while improving profit.

---

Unsure whether your promotional performance is really driving profitable growth? Reach out to us.

Relevant Articles:

· Article: The Zero-Click Customer Journey: What It Means for Performance

· Article: How to Optimise for Generative AI Search: GEO vs. SEO

· Report: The State of Search 2026 and Beyond: How AI, Automation, and Commerce Are Reshaping Discovery

About Crealytics

Crealytics is an award-winning full-funnel digital marketing agency fueling the profitable growth of over 100 well-known B2C and B2B businesses, including ASOS, The Hut Group, Staples and Urban Outfitters. A global company with an inclusive team of 100+ international employees, we operate from our hubs in Berlin, New York, Chicago, London, and Mumbai.