How Trade Promotions Inflate Media ROAS And How to Separate the Two Effects

Every quarter, the same situation plays out across CPG and omnichannel retail. A brand runs a media campaign alongside a seasonal promotion, reports a strong ROAS, and sets that number as the new benchmark. The following quarter, without the promotion, performance drops and the media team is asked to explain a gap that was never a media problem to begin with. This post covers the mechanism, the financial cost, and the only measurement approach that resolves it cleanly.

Each of the methods above fails at the same point: none can see the promotional variable, so none can report media performance without it. 

Why Your Strongest ROAS Quarter Is Often a Budget Planning Trap

The strongest ROAS numbers of the year almost always coincide with the heaviest promotional periods, and that coincidence is the planning trap. Your Q4 marketing campaign is running alongside a seasonal promotion. The platform reports a whopping 4.2x ROAS, a significant jump over your average ROAS. Finance is happy with the number and immediately approves a larger Q1 budget. When Q1 launches without the promotion to back it up, ROAS dips down to 2.1x. The ultimate conundrum? The media team is now being asked to explain the 50% drop in performance that no one saw coming. 

Here is what actually happened: The Q4 number was not a media success driven by the campaign nor was the Q1 number a media failure. Both were the result of a combined effect that no one separated before making the budget call. The promotion added demand tailwind to Q4. The media ran without promotion in Q1.  

This is an all-too-familiar pattern that plays out in budget meetings every quarter across omnichannel retail, CPG, and any brand running promotional activity alongside paid media. What makes this inflated ROAS illusion dangerous is not the single misread, rather the fact that the 4.2x becomes the new benchmark. Finance expects it again and the media team is held to a number that they cannot measure up against. 

So, what’s to be blamed here? Is it the attribution model? No. Attribution models work exactly as they should; they see revenue rise when the promotion and the media runs together and credit the media. They have no mechanism to see the promotion’s contribution to the combined ROAS being reported. 

Why Attribution Cannot Separate These Two Effects

What Attribution Models Actually Measure

Attribution models trace the paths near the conversion touchpoints: ad impression, click, purchase. That is the problem they were built to solve, and they solve it reasonably well within those boundaries. 

What they cannot measure are external demand factors, let’s say, a price cut or an in-store promotion. This is what makes trade promotion misattribution structural rather than incidental. When a consumer clicks an ad and purchases a product, the attribution model records: the ad caused purchase. The fact that the trade promotion may have tipped the scales in the ad’s favor remains invisible to the model. 

The Concurrent Event Problem

Trade promotions and media campaigns almost never run in isolation from each other. In practice, promotional windows are chosen precisely because they align with high-traffic moments. The signals are structurally entangled before the media campaign even launches.  

In fact, industry estimates suggest that in CPG categories up to 70% of total volume is sold during promotional periods. If most of your media runs when most of your promotions run, your media ROAS data is almost entirely plagued with media attribution contamination you have no way of separating.  

The Feedback Loop This Creates

The contaminated ROAS number enters the next budget cycle as a benchmark. Future campaigns are evaluated against an inflated baseline making them appear to underperform relative to a standard they were never capable of replicating.  

Over time, the brand’s media effectiveness measurement benchmarks drift further from reality, and every reallocation built on them compounds the error one quarter after another. One wrong number teaches the system the wrong lesson, and the system acts on that lesson every time. 

The Scale of What This Is Costing You

The financial stakes here are larger than most brands account for. Approximately $500 billion is spent globally on CPG trade promotions annually. Industry consensus suggests that 35 to 40 percent of that spend is wasted, in part because the feedback loop between trade investment and media attribution has never been made visible. The two are managed separately and never asked to account for what they drove together versus independently. You cannot optimize what you cannot isolate.  

What media ROAS inflation does at the channel level is equally damaging. When a platform gets credit for lift that a discount actually drove, budget migrates accordingly. Brands over-invest in channels the model rewarded for catching a promotional tailwind, moving budget from channels that were actually delivering efficient media returns during non-promotional periods. Budget reallocation is being built on a distorted signal. 

Another consequence is brands falling into the repeatable performance trap. A campaign may have returned 4x during a promotion, but that doesn’t make it a 4x campaign. Presenting it to the board as a repeatable baseline is a consequence of the architectural limitation where promotional lift attribution blends into media reporting. When Q1 comes back at 2.1x, Finance concludes that the media team cannot accurately predict performance. This creates a dent on credibility that costs more than one budget cycle; it reshapes the entire CMO-CFO relationship for the next twelve months. 

Four Scenarios Where This Distortion Is Most Severe

Trade promotion misattribution causes the most damage in four commercial scenarios that most omnichannel and CPG brands encounter regularly. 

1. Seasonal Campaigns Running Alongside Promotional Windows

A brand that sets its media effectiveness standard from a Black Friday campaign ROAS is setting a benchmark it will spend the rest of the year failing to meet. Holiday, back-to-school, Black Friday, these are some moments when media budgets are at their highest, and promotional depth is at its greatest. They are also the moments when ROAS inconsistency is severe and likely to be extracted as annual benchmarks. 

2. Retailer Co-Op Media That Requires Promotional Placement

Many co-op arrangements such as a featured listing or a sponsored position require a promotional offer for a media placement. The media and the promotion are commercially inseparable; one does not exist without the other. Attributing the resulting lift to the media alone happens when the model has no variable for the promotional condition that makes the placement possible. 

3. New Market or Channel Launches With Introductory Pricing

When introductory pricing normalizes, performance declines and the media is blamed, even though the discount was the primary trial driver. The media is blamed, in reality, the discount was the primary trial driver, and the media ran on its tailwind without any model separating the two. 

4. Performance Marketing During Category-Wide Promotional Events

When a retailer runs a category-wide event, irrespective of whether the brand initiated it, baseline demand rises across the shelf. A campaign active during that window captures organic promotion-driven sales lift and reports it as incremental media ROAS. When the promotion ends, the channel underdelivers against its own inflated benchmark, with no reasonable explanation that satisfies Finance. 

What Clean Measurement Actually Requires

Clean separation of trade and media effects requires a model that was designed to see both simultaneously, not an upgraded attribution stack. Trade and media effects can only be separated if they are modelled together with independent variables. Any model that treats ROAS as a function of media spend alone, without factoring in promotional depth, timing, and retailer overlap, will always misallocate credit between the two. 

What the model needs as inputs: A complete trade and promotion calendar covering dates, discount depths, retailer, and SKU-level data where available; media spend by channel and market for the same period; and POS-level sales data as the dependent variable 

Why this cannot be done inside an attribution tool: Attribution operates at the user-journey level and has no mechanism to register promotional variables. The separation requires a marketing mix modelling trade promotions architecture that treats trade as a first-class variable alongside media spend, not as noise to filter out. 

How Marketing Mix Modeling Separates Trade and Media Effects

How MMM Includes Trade as a Variable

A well-built marketing mix modelling trade promotions model does something attribution cannot: it separates media and trade using a promotional variable for every period in the dataset. Was a promotion running at the time? For which markets or channels? What was the discount depth? These variables sit in the model alongside media spend as independent variables that measure the contribution of the promotion to sales lift separately from media. 

What does this do when a media campaign and promotion run at the same time? The model attributes the share of sales lift to the correct channels, rather than combining it into a single ROAS. 

What Decomposition Looks Like in Practice

The result of this separation is sales decomposition. The model can now show what share of the sales lift was media-driven, what share was promotion-driven, and what share was baseline. One blended trade promotion media ROAS figure becomes two separate, trustworthy ones.  

Here is what that decomposition looks like in practice.: You start a campaign during a 15% off promotion. Attribution reports a blended ROAS of 3.8x. After MMM decomposition, the media ROAS is 2.1x and the promotional contribution accounts for the remaining 1.7x. The media performed well but not at 3.8x which the attribution would have otherwise noted.   

The difference between 3.8x and 2.1x stops being a mystery the moment a model can see both effects separately. This decomposition produces something more valuable: a media-only ROAS coefficient that works as a capital allocation signal making the next board presentation defensible and giving the CFO a credible number to build a budget plan on. 

Why This Changes the Promotion Calendar Too

Clean separation does not just produce better numbers. It drives different decisions and consistently better ones. 

When trade contribution is visible in isolation, the brand can evaluate promotional mechanics such as discount depths, durations, retailer co-op types, with the same rigor as media channels. Which promotions drive genuine trial vs. pantry loading? Where does promotional ROI hit diminishing returns? Only by separating the two does trade spend stop being a cost center and becomes an optimizable lever. 

This changes the entire CMO-CFO dynamic as instead of defending media performance against an unrealistic benchmark, the CMO can now present both media and trade efficiency in the same view. And when the next planning cycle comes, the revenue forecast holds. Finance’s question moves on from “why didn’t you replicate Q4’s success?” to “where do we put the next dollar?”  

LiftLab’s Agile MMM architecture treats trade and promotion as separate variables in the model rather than noise to be absorbed, because the real question is not just whether the promotion worked, but whether the media worked without it. 

Find out how the online-to-offline attribution gap compounds media attribution contamination across your full measurement stack. 

Three Signs Your ROAS Has a Promotion Problem

1. Your strongest campaigns always coincide with promotional windows, and no one knows why. If top-quartile media ROAS consistently clusters around promoted periods and lower-quartile results fall in clean ones, the variable your model is missing is the promotion not media volatility.  

2. Your media performance looks unpredictable, quarter over quarter. When similar budgets, channels, and creative produce widely different results, an uncontrolled trade calendar is frequently the underlying cause. What reads as ROAS inconsistency is usually promotional overlap changing the demand environment. 

3. Your Q4 or peak season ROAS becomes your year-round benchmark. Peak periods typically carry the heaviest promotional activity. Using that ROAS as the standard for evaluating Q1 or Q2 campaigns sets a benchmark no clean-period campaign can structurally match. What looks like underperformance is the difference between a promoted environment and a clean one. 

How to Start Separating Trade from Media Effects

The starting point is not a measurement overhaul, but a complete data audit.  

Pull up your trade and promotion calendar for the last 24 months alongside your media spend. Ask how many weeks both ran simultaneously. For many omnichannel brands, the overlap between active trade promotions and live media campaigns exceeds the majority of calendar weeks. That overlap is how most current CPG media benchmarks were set. 

What needs to be done? Bring both datasets into a single model. With independent variables for trade and media, decomposition is now possible. What was one untrustworthy number becomes two reliable ones. The next campaign forecast doesn’t rely on an inflated ROAS driven by trade promotion, rather it is built on what media independently drives. The board? You no longer need to explain the promotional calendar, thereby driving cleaner capital allocation decisions. 

Written by Sushant Ajmani

VP of Product Marketing at LiftLab, helping omnichannel retailers and CPG brands operationalize Marketing Mix Modeling (MMM) for smarter planning and investment. With 25+ years of experience across analytics, product, and go-to-market leadership, he translates causal measurement into clear decisions, balancing short-term efficiency with long-term brand growth that leaders can trust.

More posts