The short answer
Build product category dimension into the Marketing Mix Model. Keep the product category advertised separate from the category purchased. Map campaigns and sales to those dimensions, then use MMM to estimate effects across the categories the data can support. Check the causal evidence and count each sales outcome once, so a campaign's value can include purchases beyond its advertised product.
This article is part of Asked by Marketers, a series answering real questions from marketing leaders.
Sellforte's team holds more than 1,450 meetings each year with marketing leaders in Ecommerce and Retail about Marketing Mix Modeling and incrementality testing. Each week, we anonymize at least one question from those conversations and answer it in depth, based on what marketers are actually struggling with, not what keyword tools suggest. About the series →
Why marketers ask this
A campaign can advertise one product while contributing to sales elsewhere in the business. That matters when the advertised product has a different margin or customer value from the product eventually purchased. A report organized only around the promoted category may leave the team with an incomplete view of what the campaign contributes.
How should we map the advertised and purchased categories?
Create separate labels for what the campaign promotes and what the sales record contains. Keeping both dimensions allows the analysis to ask how advertising for category A relates to outcomes in categories A, B, and beyond.
Campaign data need a usable product hierarchy. A campaign for one category can carry that category's tag. A campaign promoting the whole brand may need a broader label. Forcing every campaign into a single product category can create a false level of precision before the model even runs.
Apply a consistent hierarchy to sales. Check whether each outcome contains product revenue, orders, subscriptions, or another business measure. The same customer can buy in several categories, but the category totals must have clear definitions if they are going to be combined.
| Dimension | What it describes | Example |
|---|---|---|
| Advertised category | The product or offer promoted by the campaign. | A campaign promoting running shoes. |
| Purchased category | The product category in the measured sales outcome. | Footwear or clothing purchased during the analysis period. |
The mapping makes a cross-category question possible. It does not establish that the campaign caused every purchase in either category.
Does this require matching every shopper to an ad?
A cross-category MMM does not require a complete identity match between each ad exposure and purchase. It uses aggregate outcomes and marketing inputs, with category relationships defined from the business and available data.
In Marketing Mix Modeling, the setup can use analysis of advertised and purchased categories to identify plausible relationships before estimating their effects. A relationship may be stronger for some category pairs than others. Those judgments should be grounded in the data; they are not universal weights that can be copied from another business.
A customer-level match can supply useful information when it exists, but it does not prove that the purchase would have been lost without advertising. Likewise, two categories moving together can reflect a shared promotion, seasonality, or another demand driver.
The model therefore needs the relevant business context as well as category tags. Keep the preliminary mapping distinct from the estimated incremental effect. One says which relationships the analysis should examine; the other estimates how much the modeled outcome changed because advertising changed, under the model's assumptions.
How can we tell whether a cross-category effect is credible?
Check whether the data can distinguish the campaign from other activity affecting the purchased category. Each category estimate needs enough supporting data to inform a decision.
Start at a category level that matches the decision and the available variation. If many campaigns move together or a category has little sales volume, a separate estimate for every pair can be unstable. Grouping related products may support a more useful conclusion than reporting a precise-looking result for every small category.
Incrementality testing can add evidence when the intervention and outcomes match the question. For example, a campaign test can examine both the promoted category and a relevant secondary category. Keep the outcome definitions and measurement period explicit when using that result to inform the model.
Promotions require particular care. A discount on one product may change the basket independently of advertising. The retail MMM RFP guide treats promotion halo and other demand drivers as capabilities to assess. A model of advertising spillover needs to account for those effects rather than automatically crediting them to media.
How should cross-category effects influence the budget?
Evaluate the campaign using the supported sales contributions across the relevant categories, with the campaign spend counted once. Then consider the economic value of what was sold.
A campaign can have a modest return in its promoted category and a meaningful contribution elsewhere. If the secondary category has a different margin or customer-value profile, revenue alone may not be enough for the budget comparison. Keep those value assumptions visible and consistently defined.
Check that the sales outcomes are mutually exclusive before adding them. Order revenue and product-line revenue can overlap. An analysis that reports the same basket in two places should not sum those figures as though they were separate sales.
Use the resulting view to assess the campaign as a whole. The relationship between advertised and purchased categories does not mean a fixed share of every future budget increase will flow to the secondary category. Historical contribution and the expected return from changing spend remain different questions.
How this looks in practice
Consider a hypothetical retailer running a $40,000 campaign for running shoes. Assume the analysis produces credible incremental estimates for two mutually exclusive product-revenue categories over the same period. These figures illustrate the calculation and are not customer results.
| Purchased category | Modeled incremental revenue |
|---|---|
| Footwear | $80,000 |
| Clothing | $40,000 |
| Combined measured contribution | $120,000 |
A footwear-only view gives an iROAS of 2.0: $80,000 divided by $40,000. Including the supported clothing contribution gives a combined measured iROAS of 3.0. The campaign's spend appears once in that calculation.
The team should still review uncertainty and category margins before deciding whether to increase investment. If the clothing estimate cannot be separated from a simultaneous clothing promotion, the table needs that qualification instead of presenting the full $40,000 as settled evidence.
How Sellforte helps
Sellforte can use campaign and sales-category hierarchies to analyze advertising effects across product categories. The category relationships and reporting detail are established from the available data during the measurement setup. Book a demo.
Authors
Lauri Potka is the Chief Operating Officer at Sellforte and has over 15 years of experience in Marketing Mix Modeling, marketing measurement, and media spend optimization. Before joining Sellforte, he worked as a management consultant at the Boston Consulting Group, advising some of the world's largest advertisers on data-driven marketing optimization. Follow Lauri on LinkedIn, where he is one of the leading voices in MMM and marketing measurement.