Promotion Performance Measurement & Incrementality
Executive Overview
Promotion performance measurement is the discipline of determining whether a promotional investment generated genuine, incremental commercial value - or whether it merely moved revenue from one period or price point to another while consuming margin. The distinction matters enormously: a promotion that drives a 20% sales uplift but cannibalises 18% of full-price purchases in the same category has delivered almost no net value while costing significant margin.
Accurate measurement is harder than it appears because promotions interact with seasonality, consumer stockpiling, competitor activity, and the natural purchase rhythm of categories. Gross revenue comparisons - comparing promotional-period sales against the prior period or the same period last year - are the most common measurement approach in retail but the least analytically rigorous. They confuse correlation with causation and systematically overstate promotional value.
The field of incrementality measurement provides the corrective framework. Grounded in causal inference methodology, it answers the only commercially relevant question: what additional value was generated because the promotion ran, compared to what would have happened without it? The IAB and IAB Europe formalised industry guidelines for incremental measurement in commerce media in November 2025, grounded in three principles: credible counterfactuals, control of bias, and separation of signal from noise.
In this article:
- The five measurement errors that most commonly overstate promotional ROI
- The three principal methodologies for incremental measurement and when to apply each
- The core KPI set for promotion performance: what to measure and what to ignore
- How cannibalism, pull-forward, and halo effects distort measurement
- How Enactor's transaction analytics layer and Shannon's effectiveness scorecards support structured promotion measurement
Market Context
Promotion measurement is at an inflection point. A 2024 survey by the Association of National Advertisers (ANA) found that 71% of advertisers now consider incrementality the most important KPI for retail media investments2. Only 26% of in-house marketers currently conduct incrementality testing internally14 (Funnel Research, 2024), leaving the majority making investment decisions without causal evidence of effect.
Skai's 2025 State of Retail Media report found that 44% of marketers express concerns about the accuracy and reliability of their measurement results, 43% struggle with a lack of clear methodologies, and 41% cite difficulty distinguishing organic from driven sales3. These are not abstract analytical concerns - they translate directly into promotional budgets being allocated to programmes that are not creating value.
The eMarketer and TransUnion July 2025 survey found that over half of US brand and agency marketers are already using incrementality testing, with 36.2% planning to increase investment in the methodology over the next year. Marketing Mix Modelling was identified as the most reliable measurement methodology by 27.6% of respondents4. Both signals reflect a fundamental shift: the industry is moving from activity measurement (did the promotion run?) to causal measurement (did it work?).
For UK retailers, the motivation is intensified by margin pressure. With promotional spend constituting a significant proportion of gross margin investment and input cost inflation reducing headroom, the ability to identify and eliminate ineffective promotions is a direct lever on profitability - not merely an analytical exercise.
The risk of inaction is concrete and compounding: retailers without structured measurement frameworks systematically overinvest in promotions that underdeliver, cannot demonstrate commercial return to their board or buying teams, and lose the institutional knowledge of what actually works in their categories.
How It Works
The Five Measurement Errors
Most retailers who report that their promotions "worked" are making at least one of five systematic errors that inflate apparent performance. Understanding these errors is the prerequisite for building a measurement programme that produces actionable insight.
| Error | What happens | Why it inflates results |
|---|---|---|
| Seasonal uplift attribution | Promotional period coincides with a seasonal peak | Sales would have risen anyway; the promotion gets credit for the season |
| Stockpiling / pull-forward | Customers buy more than normal but do not return sooner | Post-promotional trough offsets uplift; total-period incremental is near zero |
| Category cannibalism | Promoted SKU gains volume from a non-promoted category competitor | Category revenue is flat or down; item revenue is up |
| Full-price cannibalisation | Promotion accelerates full-price buyers into the promotional window | Margin is destroyed without any net volume gain |
| Halo misattribution | Non-promoted items in the basket rise alongside the promoted item | Uplift in adjacent items attributed to the promotion rather than natural basket composition |
Stockpiling and pull-forward are invisible in a 7-day measurement window. A promotion on long-shelf-life grocery or household items generates volume during the promotional week, then a significant trough in the following two to four weeks as customers draw down stock. Measuring only the promotional window is the most common cause of systematically overestimated promotional ROI in grocery and FMCG retail. The minimum measurement window for any promotion on a commodity product should be the promotional period plus four weeks.
The Three Principal Incrementality Methodologies
The IAB's November 2025 guidelines for incremental measurement in commerce media identify four approaches along a spectrum of causal rigour and implementation complexity. For retail promotion measurement, three are practically applicable for most UK retailers.
| Methodology | What it measures | Causal rigour | Implementation complexity |
|---|---|---|---|
| Holdout test (A/B) | Absolute lift from promotion vs. no promotion, measured against a withheld control group | High | Medium - requires random group assignment and withholding capability |
| Matched market test | Lift in test markets vs. matched control markets without the promotion | Medium-High | Medium - requires geographic variation and matched market selection |
| Marketing Mix Modelling (MMM) | Statistical decomposition of sales drivers including promotional spend, seasonality, and external factors | Medium | High - requires 18-24 months of historical data and specialist modelling |
Holdout testing is the most directly applicable methodology for individual promotion measurement at POS level. A defined cohort of customers or stores - typically 5 to 20% of the eligible population - is withheld from the promotion entirely. Their purchasing behaviour during the promotional period is compared to the behaviour of the treated group. The gap is the incremental lift. The Agile Brand Guide's 2025 guidance on holdout methodology identifies the key design requirements: randomise at the appropriate unit (individual customer or store), stratify by lifecycle stage or value tier before randomisation, enforce the holdout across all channels, and include a measurement window that accounts for conversion lag8.
Matched market testing is more appropriate for promotions that cannot be withheld from individual customers - such as in-store price reductions that are visible to all shoppers in a location. Test stores or regions run the promotion; matched control stores or regions of similar trading characteristics do not. The incrementality is the gap between test and control locations, adjusted for known differences. Stella's geo-based incrementality research across 225 tests conducted August 2024 to December 2025 found an overall median incremental ROAS of 2.31x5 - meaning for every pound of promotional investment, the typical campaign generated £2.31 in truly incremental revenue - with substantial variance by execution quality.
Marketing Mix Modelling is the most resource-intensive approach but provides the broadest view: it decomposes overall sales into their contributing factors, attributing a proportion to promotional spend while controlling for seasonality, price elasticity, competitor activity, and macroeconomic conditions. The eMarketer/TransUnion survey (2025) found 46.9% of US marketers planning to increase MMM investment. Its primary limitation is reporting lag - models require months of post-period data and are typically run quarterly rather than in real time4.
Holdout tests provide precise, campaign-level causal measurement for specific promotions but cannot guide strategic portfolio decisions. MMM provides strategic portfolio guidance but cannot attribute individual campaign effects with precision. The most analytically sophisticated retail measurement programmes use holdout tests to validate specific promotion mechanics and MMM to guide overall promotional budget allocation - neither alone is sufficient for both purposes.
The Core Promotion KPI Set
Not all promotion metrics are equally informative. The following KPI framework distinguishes metrics that measure genuine commercial outcome from those that measure activity - a distinction that is frequently collapsed in standard promotional reporting.
Tier 1 - Commercial outcome metrics (what actually matters)
| KPI | Definition | Measurement approach |
|---|---|---|
| Incremental revenue | Revenue generated above baseline that would not have occurred without the promotion | Holdout gap or MMM decomposition |
| Incremental gross profit | Incremental revenue minus the cost of the discount offered | Requires margin data per promoted SKU |
| Incremental ROAS (iROAS) | Incremental revenue / promotional cost | Holdout or MMM measurement |
| Second visit rate | Proportion of customers acquired through the promotion who make a second purchase within a defined window | Loyalty transaction data |
| Category net impact | Promoted SKU uplift minus cannibalised volume from competing category SKUs | Category-level transaction analysis |
Tier 2 - Diagnostic metrics (useful for explaining outcomes)
| KPI | Definition | What it diagnoses |
|---|---|---|
| Redemption rate | Promotional offers redeemed / offers issued | Targeting quality; offer relevance |
| Basket uplift | Average basket value in promotional transactions vs. baseline | Spend-to-unlock effectiveness |
| Average discount depth | Actual average discount achieved across all promotional transactions | Whether margin floor is functioning |
| Duplicate redemption rate | Vouchers redeemed more than once / total redemptions | Serial tracking configuration failures |
| Group coverage rate | Customers in targeted group who transacted during the promotional period / total group membership | Group population freshness |
Tier 3 - Activity metrics (necessary but insufficient)
Gross sales volume, total transactions, total vouchers issued, and total redemption count are necessary operational metrics for confirming the promotion executed as configured, but they say nothing about whether it created value. They should not feature in board-level or buying-team promotional performance reviews.
A high voucher redemption rate indicates that customers claimed the offer. It does not indicate that the offer drove incremental purchase - the customer may have intended to purchase regardless, with or without the voucher. Redemption rate is a diagnostic metric for offer targeting quality, not a commercial outcome metric. It should always be paired with holdout-based incremental revenue data before any conclusion about promotional value is drawn.
Measuring Cannibalism, Halo, and Pull-Forward
The three distortions that most commonly mislead promotional assessments require specific measurement approaches beyond simple uplift calculation.
Cannibalism is measured at category level. CrossCap's guide to analysing promotional lift describes the methodology: examine all historical promotional periods for the promoted SKU and assess the impact on competing items in the same category. If Pampers and Huggies are substitutable commodities and Pampers goes on promotion, Pampers unit sales rise but Huggies unit sales fall - the category net impact is the relevant measure, not the Pampers item uplift in isolation. The practical implementation requires analysing category-level transaction data and identifying the degree to which competing SKU sales decline during each promotional event.
Halo effects are measured by examining non-promoted item movement in the same basket as the promoted item. If baskets containing the promoted item also show elevated sales of certain complementary items, this halo value can be attributed to the promotion - provided it is not simply a reflection of the natural co-purchase pattern of those items outside promotional periods.
Pull-forward is measured by extending the analysis window to include the post-promotional period - typically four to six weeks for FMCG and eight to twelve weeks for considered-purchase categories. If the promotional-period uplift is substantially offset by a post-promotional trough, the true incremental value is materially lower than the in-period measurement suggests.
Seasonal Baseline Construction
Seasonal baseline construction is one of the most technically demanding aspects of promotion measurement. The counterfactual question - what would have sold without this promotion? - requires a credible model of baseline demand that accounts for the time of year, calendar events, weather, and category-specific seasonality patterns.
For most UK retailers, the practical approach is an equivalent calendar-week comparison using two to three years of historical data. If the promotion ran in week 14 of 2026, the baseline is constructed from the average of weeks 14 in 2024 and 2025, adjusted for any known trading differences (store estate changes, new competitors, significant pricing changes). This approach is accessible without specialist statistical software and produces a serviceable baseline for most promotional evaluations. More sophisticated approaches, such as synthetic control methods that construct a weighted combination of historical periods to match the pre-promotion trading pattern, produce lower variance baselines but require statistical expertise to implement correctly.
Costs and Considerations
| Cost layer | Notes | Frequency |
|---|---|---|
| Transaction data infrastructure | Granular, time-stamped POS transaction data at SKU level is the foundational requirement - must be accessible and queryable | Ongoing infrastructure cost |
| Analytical resource | Basic holdout test analysis is achievable with Excel; MMM requires specialist data science resource or external vendor | Per measurement cycle |
| Holdout test design and enforcement | Requires system capability to withhold a promotion from a defined customer cohort | Per test |
| MMM vendor or platform | Third-party MMM services range from £30,000 to £150,000+ per annum for a UK mid-size retailer | Annual contract |
| Loyalty capture | Holdout testing at customer level requires identified transactions - minimum 50% POS loyalty capture rate | Ongoing operational investment |
What is free: The measurement framework, KPI hierarchy, and analytical approach described in this article can be implemented using existing transactional data and standard analytical tools. The incremental investment is primarily analytical resource and process design, not platform licensing.
What is optional: Third-party incrementality testing platforms, MMM services, and AI-powered measurement tooling are additive investments for retailers who have outgrown manual analysis. The foundational disciplines - holdout design, cannibalism measurement, pull-forward analysis - are implementable without specialist tooling.
Retailers implementing measurement disciplines for the first time should identify their single highest-spend promotion and design a rigorous holdout test around it first. A single well-designed holdout test on one promotion produces more actionable commercial insight than a poorly designed attempt to measure the entire promotional portfolio simultaneously. The learnings from the first test - sample sizes, window lengths, category cannibalism patterns - inform the design of all subsequent tests.
The Business Case
Core Argument
The business case for promotion measurement is not about reducing promotional spend - it is about redirecting it. Every pound of promotional budget allocated to a programme that does not drive incremental sales is a pound available for a programme that does. A disciplined measurement programme that identifies even 10-15% of promotional spend as non-incremental creates a material opportunity for either cost reduction or redeployment to higher-performing mechanics.
Revenue Upside
The commercial opportunity from measurement-driven promotion optimisation is substantial. Stella's geo-based incrementality research found that iROAS performance ranged from 253% to 1,609% across different advertiser programmes5 - a span that illustrates how dramatically promotional effectiveness varies across mechanics, categories, and execution approaches. Retailers who can identify which mechanics genuinely drive incremental revenue and concentrate investment there capture a substantial share of that variance. Measured's Q4 2025 case study found that reallocating 40% of promotional budget from retargeting to prospecting campaigns - informed by holdout test results - generated a 29% lift in incremental revenue6.
Cost Reduction
The direct cost reduction is the elimination of promotional spend on programmes demonstrated to be non-incremental or sub-threshold on iROAS. The indirect cost reduction is the prevention of margin erosion from misconfigured promotions - those without margin floors, with serial tracking failures, or with overly broad eligibility - which can be identified through the systematic analysis that a measurement programme requires.
Risk of Inaction
Without incremental measurement, all promotional spend appears to generate uplift because baseline behaviour is invisible. Retailers who measure only gross metrics consistently over-report promotional success, over-invest in subsequent promotions based on that false signal, and compound margin erosion over time. The Skai 2025 data suggesting 44% of marketers distrust their own measurement results reflects the commercial cost of this problem: significant investment is being made without confidence in its effectiveness.
Indicative Business Case Model
Figures are illustrative only, based on published industry data and common UK retail patterns.
| Initiative | Illustrative investment | Illustrative benefit | Payback period |
|---|---|---|---|
| Measurement framework design and KPI definition | £5,000-£15,000 | Foundation for all subsequent measurement | Immediate - enables all other benefits |
| First holdout test on highest-spend promotion | £5,000-£20,000 | Identification of true incremental contribution; 10-20% budget reallocation opportunity | 3-6 months |
| Category cannibalism analysis | £5,000-£10,000 | Identification of promotions destroying category margin; 5-10% category margin recovery | 3-6 months |
| Pull-forward analysis (FMCG) | £5,000-£10,000 | Identification of stockpiling-driven false uplift; promotion frequency right-sizing | 3-6 months |
| MMM for portfolio-level budget allocation | £30,000-£100,000 | 10-20% improvement in total promotional portfolio ROI through budget reallocation | 6-18 months |
Key Risks and Mitigations
| Risk | Likelihood | Mitigation |
|---|---|---|
| Insufficient transaction data granularity | Medium | Audit POS data capture - SKU-level, time-stamped, loyalty-linked transactions are the minimum requirement |
| Low loyalty capture rate preventing customer-level holdout testing | High for non-loyalty retailers | Use matched market testing as the alternative; invest in loyalty capture to enable customer-level holdouts |
| Internal resistance to measurement findings | High - results often challenge established beliefs | Secure CFO or commercial director sponsorship before beginning; present findings as optimisation, not audit |
| Seasonal volatility corrupting measurement | Medium | Extend measurement windows; use equivalent calendar-week baselines; avoid peak trading periods for first tests |
| Holdout test contamination (cross-channel leakage) | Medium | Enforce holdout across all channels; audit channel delivery logs for exposure leaks |
Enactor and This Topic
Enactor's transaction analytics architecture provides the data foundation for promotion measurement disciplines described in this article. The combination of granular POS transaction data, loyalty identity capture, and Shannon's effectiveness scoring framework makes Enactor one of the few POS platforms with native capabilities for incrementality-adjacent promotion measurement.
Every Enactor transaction is recorded at SKU level with time-stamp, store identifier, operator, loyalty identity, promotion applied, and discount value. This is the foundational data layer for all promotion measurement - category uplift analysis, basket composition, redemption tracking, and post-promotional trough measurement all depend on this granularity.
Enactor's Estate Manager provides redemption reporting across all active promotions, including redemption count, discount value, and average basket value at time of redemption. This is Tier 3 activity measurement - necessary but insufficient. The commercial measurement layer sits above this.
Targeted vouchers with serial tracking generate an audit trail that identifies duplicate redemption attempts - a critical integrity check for targeted promotional programmes. Duplicate redemption rate (configured as EFF-022 in the Promotion Intelligence Framework) is a diagnostic metric for serial tracking configuration failures.
Shannon's effectiveness scorecards, defined in the Promotion Intelligence Framework (PIF v1.6), include EFF-023 (group-targeted promotion reach rate) and EFF-024 (concentration of redemptions within a small proportion of the group). These metrics diagnose targeting precision - whether the promotion reached its intended cohort and whether redemptions were evenly distributed or concentrated among a subset of repeat redeemers.
Shannon's transaction analytics layer, built around Enactor's three-tier customer model (Anonymous/PAR -> Pseudonymous -> Identified), provides the data infrastructure for seasonal baseline construction. The layer supports equivalent calendar-window baseline calculations using two to three years of historical transaction data per store, enabling the counterfactual demand model required for holdout test analysis.
Shannon's 20 EFF-category rules cover second visit rate for acquired customers (EFF-005), targeted voucher redemption rates against configured benchmarks (EFF-021), and entropy metrics (BER, IPER, ERC) that detect anomalous promotional behaviour patterns. These rules provide an automated first-pass effectiveness assessment for each promotion in the portfolio, flagging those that warrant deeper holdout or cannibalism analysis.
Shannon's PIF-governed effectiveness review is not a replacement for holdout testing - it is the screening layer that identifies which promotions most need rigorous causal measurement. A promotion with a low second visit rate (EFF-005 flag), a low redemption rate (EFF-021 flag), and a high concentration score (EFF-024 flag) is a candidate for a full holdout test. Shannon surfaces the signal; the measurement programme provides the causal attribution. Together they form a scalable measurement stack that operates from automated screening through to rigorous test design.
For scoping discussions on transaction data architecture, Shannon's analytics layer, and the Promotion Intelligence Framework, contact Enactor Professional Services.
References
- IAB and IAB Europe. Guidelines for Incremental Measurement in Commerce Media. November 2025. https://www.iab.com/guidelines/guidelines-for-incremental-measurement-in-commerce-media/
- Association of National Advertisers (ANA). 2024 Retail Media Network Survey. January 2024. Cited in Dataslayer.ai and Infillion, 2025.
- Skai / Path to Purchase Institute. 2025 State of Retail Media Report. 2025. https://skai.io/blog/2025-state-of-incrementality-in-retail-media/
- eMarketer / TransUnion. MMM, Incrementality, and Other Measurement Trends That Will Define 2026. July 2025. https://www.emarketer.com/content/mmm--incrementality--other-measurement-trends-that-will-define-2026
- Stella. 2025 DTC Digital Advertising Incrementality Benchmarks: 225 Tests, August 2024-December 2025. https://www.stellaheystella.com/blog/2025-dtc-digital-advertising-incrementality-benchmarks
- Measured. Holdout Testing: A Marketer's Tool. 2025. https://www.measured.com/faq/holdout-test/
- Fusepoint. Holdout Testing: A Marketer's Secret Weapon. September 2025. https://fusepointinsights.com/blog/holdout-testing-gold-standard/
- Agile Brand Guide. Holdout Campaign Methodology. September 2025. https://agilebrandguide.com/wiki/methods/holdout-campaign/
- CrossCap. Guide to Analysing the Overall Lift of a Retail Promotion. January 2025. https://www.crosscap.com/guide-to-analyzing-the-overall-lift-of-a-retail-promotion/
- Northbeam. The 2024 Guide to Incrementality. 2024. https://www.northbeam.io/blog/the-2024-guide-to-incrementality
- Infillion. The Incrementality Imperative: Why Retail Media's Future Depends on Proving ROI. June 2025. https://infillion.com/blog/the-incrementality-imperative-why-retail-medias-future-depends-on-proving-roi/
- Dataslayer.ai. Incrementality: Top Retail Media KPI for 2025. 2025. https://www.dataslayer.ai/blog/incrementality-becomes-the-primary-kpi-for-retail-media-advertisers
- Measured. Marketing Mix Modelling: A Complete Guide for Strategic Marketers. November 2025. https://www.measured.com/faq/marketing-mix-modeling-2025-complete-guide-for-strategic-marketers/
- Funnel Research. Cited in Infillion. Only 26% of in-house marketers conduct incrementality testing internally. 2024.
Enactor Retail Knowledge - published March 2026. This article draws on publicly available research and platform documentation. Market statistics are sourced from named third-party publications and do not represent Enactor's own research. Pricing figures are indicative based on publicly available information at time of publication and should be verified directly with providers.