Most Price Promotions Are Still Judged Before Their Cost Arrives

A price promotion produces an immediate and unusually persuasive signal: sales move. That visibility helps explain why discounts retain such a powerful hold on consumer-goods marketing, particularly as seasonal trading approaches.

But a new survey of 250 senior UK marketing and insight leaders raises a more awkward question. Three quarters said they could not estimate what proportion of their promotions had been profitable over the preceding year. The research, presented by Les Binet and Kantar’s Dom Boyd, does not show that promotions are inherently wasteful. It shows that many organisations are treating an observed sales uplift as if it were a commercial answer.

That is a research problem before it is a pricing problem. A promotion can recruit a genuinely new buyer, prompt a useful trial, defend distribution or clear stock. It can also subsidise a purchase that would have happened anyway, train customers to wait for the next deal or shift demand from one week into another. Those outcomes can look remarkably similar at the till.

Key takeaways

• Measure incrementality, not movement. A sales spike matters only after the likely baseline without the promotion has been established.

• Use a longer observation window. Include the period after the offer ends, when stock-up behaviour, reduced repeat purchasing and margin effects become visible.

• Separate promotion objectives. Recruitment, retention, range trial and clearance require different success measures; one headline ROI obscures those differences.

• Build learning into the calendar. Reserve comparable stores, audiences or periods wherever possible, rather than attempting to reconstruct evidence once the campaign is over.

The sales spike is not the answer

The weakest question in promotion analysis is often the simplest: did sales rise while the discount was live? Of course they may have done. The harder question is whether the brand sold more than it would otherwise have sold, at a contribution worth having.

This requires a counterfactual. Depending on the category and available data, that might mean matched-store comparisons, holdout audiences, regional tests, historical models adjusted for seasonality, or econometric analysis that accounts for media, distribution, competitor activity and price. None is flawless. Together, they are much more useful than comparing one promotional week with the week before it.

The distinction is particularly important where shoppers can stock up. A strong promotional week may be followed by a weaker one, leaving little net volume gain. The same applies when a discount simply switches buyers from a full-price pack, another product in the portfolio or a retailer’s own channel.

Margin needs equal weight. Revenue is easy to see and easy to celebrate. The cost of funding, retailer terms, operational disruption, cannibalisation and the value of demand brought forward are harder to assemble. Yet these are the elements that decide whether a promotion created profit or merely made a dashboard look busier.

A promotion should be judged by the demand it adds, not by the demand it temporarily makes visible.

Stop treating promotions as one category

The most productive response is not to ban discounts. It is to stop grouping very different commercial actions under the single heading of “promotion”.

A launch offer for a product with low household penetration has a different job from a multibuy mechanic aimed at loyal shoppers. A retailer-exclusive event may be designed to secure visibility or distribution. A short clearance price may protect cash flow rather than build the brand. Each can be reasonable, but each needs an explicit hypothesis before it goes live.

For a trial-led promotion, the relevant evidence includes the proportion of buyers who are genuinely new to the brand or category, and what they do on their next purchase. For a retention offer, the test is whether it changes future value among customers at risk of leaving, rather than rewarding customers who were already likely to return. For a volume event, the business needs to know whether the rise is incremental after substitution and post-event decline.

That framing changes the role of customer insight. Instead of supplying a generic view of price sensitivity, research teams can identify which buyer groups are responsive, what purchase occasion is being targeted and whether the offer answers a real barrier. It also makes creative work more consequential. A price message can simply announce a saving, or it can make a product more relevant to a meal, a household routine or a moment of use. Those are not equivalent interventions.

Make the next event a learning programme

The timing matters. The IPA Bellwether report published this week found UK marketing budgets rising across every tracked category in the third quarter, with sales-promotion budgets among the strongest increases. That means more activity to evaluate, not a licence to rely on familiar mechanics.

Teams planning promotions for the final quarter should begin with a short audit of the largest recent events. For each, assemble total cost, incremental volume estimate, post-event performance, effect on full-price sales and any evidence of new-buyer recruitment. The purpose is not retrospective blame. It is to identify which mechanisms deserve repetition, redesign or retirement.

Then choose a small number of upcoming offers to measure properly. Predefine the objective, the comparison group or model, the observation period and the commercial threshold for success. Where clean experiments are impractical, record the assumptions openly rather than allowing a sales spike to stand in for proof.

The new research is a useful warning because it exposes a gap between confidence and knowledge. Consumer-goods businesses do not need fewer promotions as a matter of principle. They need fewer promotions whose apparent success has never been tested against the demand, margin and customer behaviour they were meant to change.