Marketing budgets for brands that sell through retailers.

The sections below cover which sales to model, how retailers, marketplaces and your own site come together, the controls that matter in consumer goods, how TV and retail media get into the same model, and how to handle several brands and countries.

A brand that sells through retailers rarely sees its consumers buy. What it sees depends on what the retailers share. Sell-out, the units and value sold to consumers through the retailer's tills and website, is the best outcome for the model, because it moves with advertising week by week. Many grocers and electronics chains report it through their data services, often weekly by product and store.

Sell-in, what the retailers order from you, is the fallback. It is always available, but it follows the retailers' ordering cycles: stock is built before a campaign and run down after it, so sell-in lags and smooths the effect of advertising. Where only sell-in exists, the model uses it with a longer lag and wider ranges, and we say so. Where some retailers report sell-out and others only sell-in, we agree with you how to combine them before the model is built.

Your own webshop is the one channel where you see every order. It goes into the same outcome as the retailer sales, or into a separate outcome when the question is how much of the effect lands online. The data guide lists what to collect.

Advertising for a brand does not choose a store. A TV campaign for a coffee brand lifts sales in every supermarket chain that stocks it, and a YouTube campaign for headphones lifts sales at the electronics chains, on Amazon and on your own site at once. Measured one retailer at a time, each piece of the effect is too small to read. Added into one outcome per market, it is.

So the model explains total sales per market across retailers, marketplaces and your own site. The split by retailer stays in the data as a dimension, for reporting and for the questions where it matters, such as how much a retailer's own media lifted sales there. For consumer electronics, where a launch drives sales through several chains and marketplaces in the same weeks, product launches are marked on the timeline, so the model does not credit them to whatever media ran alongside.

In consumer goods, much of the week-to-week movement in sales comes from things that are not advertising. A price promotion in a large grocery chain can double a product's sales for a week. A listing in a new chain adds a step that stays. Leave them out, and the model credits their lift to whatever media ran in the same weeks.

• Price and promotions. Promotion weeks and price changes by retailer go in as controls, from your trade promotion calendar or the retailers' data. Category managers often set the promotion calendar, so it rarely lines up week for week with the media plan, and that is what lets the model tell the two apart.
• Distribution. The number of stores that carry the product, or a weighted distribution figure, goes in because it limits what marketing can sell.
• Launches and events. New products, packaging changes and large seasonal events are marked on the timeline.

We keep the controls few and direct. Every variable added takes explanatory power from the media channels, and anything the model uses to explain the past must also be known when it forecasts. How marketing mix modelling separates price promotions from advertising goes into the detail.

For many consumer brands, TV and out-of-home carry most of the budget, and they live in media plans and spot reports from the media agency. We set up a pipeline for those with you or your agency, so offline media lands in the same structure as digital: market, product, then the media down to each campaign and spot, online and offline together. See how we collect offline media.

Retail media, the ads you buy on the retailers' own sites and apps, goes in as its own channel per network where the spend allows. The retailer's closed-loop report counts sales among people who saw the ad. The model reads retail media against the same total sales as TV and Meta, so all of them are measured on the same terms. How to measure retail media with marketing mix modelling.

TV for a brand also builds the brand over months, not only weeks. The model can carry a brand component that captures sales that arrive later, and connected TV and YouTube can sit next to linear TV as their own channels. How to measure TV, connected TV and YouTube.

Consumer goods companies often run several brands across several categories and countries. We build one model per market on total sales, with brand and category as dimensions in the data. Where brands are large and advertised separately, each can be its own outcome in that market. We start with the largest market. What that model learns about each channel, how fast it saturates and how long its effect lasts, becomes the starting point for the next market, scaled to its size, and that market's own data then corrects it.

Marketing mix modelling for different business models compares the setup for grocery, retail and other models. If you also run your own stores, see marketing budgets for retailers.

Odins is a managed marketing mix modeling platform. We connect the ad platforms and retail media networks, set up the pipeline for TV and out-of-home with you or your media agency, structure the retailer data and your own sales, build and maintain the model, and review every recommendation before it reaches you. A first model is ready 4 to 6 weeks from kickoff, data collection and structuring included. Data flows in daily and the model is retrained once a month, while scenarios and forecasts are yours to run whenever you need them, in the platform or from your own AI assistant through the Odins MCP.

When you already run incrementality or lift tests, their results go into the model. Where the model is uncertain about a channel, we structure a test for it. Odins fits brands with a marketing budget from EUR 1M to several hundred million a year. See the managed service.