Does Meta Robyn work for a food and beverage brand?

It can, if you sell food or drink and have at least two years of weekly sales, with ad spend that moved over time. Put weather, holidays and promotions into the model, since they can move a snack or drinks brand's sales on their own. What shoppers buy in stores that stock you usually never reaches your Shopify data.

If you sell food and beverage

If you sell coffee, snacks or drinks, many of your orders may come from people who already know you. They run out, and they reorder. In Robyn, steady reorders like that mostly sit in the baseline: Meta's name for the trend, season, holidays and intercept, the parts nobody can steer.

That is not a flaw. An ad gets credit for the sales it adds in its week, plus some carryover into later weeks. Meta's docs say adstock covers only media's direct effect on sales, not its effect on the baseline. So a reorder that a first order set off months ago is not the ad's, as far as the model knows.

Seasons can matter as much as ads. If iced drinks sell in a heatwave or soup in the cold, the weather belongs in the model. Robyn's guide names temperature as the common input for businesses that swing between summer and winter.

Promotions are the other big mover. If multibuys, bundles and price cuts come and go, the model needs to know when. Robyn's guide says a weekly discount percentage tells the model more than a yes-or-no promotion flag.

What one store's data shows

One store's anonymised GA4 export, 1 January 2024 to 21 August 2026. It holds shares of revenue only: no ad spend, no order counts.

The export does not say what Store A sells. It is one store, and not a benchmark for food.

On the Channels sheet, Direct holds 57.7% of revenue in every view. On the Journeys sheet, journeys with 1 touch hold 79.5% of revenue, and those buyers took 0.5 days to buy.

If you sell food or drink, that is roughly what a reorder looks like in GA4: one typed visit, same day, done.

Now picture the same weeks in Robyn. The model never sees Direct as a channel. It sees total weekly sales, your spend and your context columns. Sales like these would land in the baseline, unless they rise and fall with your spend.

So the two reports will disagree, and they should. GA4 gives a typed reorder to Direct. Robyn may give part of it to recent ads, through carryover, and the rest to the baseline.

What the export cannot show is how much of that one-touch revenue an ad started. It holds no spend either, so it cannot feed Robyn on its own.

What changes for a food and beverage brand?

Model the weather if it moves you. Add a weekly temperature column as a context variable. Prophet handles the usual yearly season, but a hot June is not an average June.

Enter promotions as numbers. Put the average discount percentage for each week in its own column, rather than a flag that only says on or off. Give a promotion its own column when you want it measured separately.

Leave tiny activities out. Robyn's guide uses a small sampling run in one region as its example of activity a model is unlikely to measure. A tasting tour of three shops probably belongs in your notes, not your model.

What to do this week

1. Check how steady your base is. In Shopify, go to Analytics > Reports, set the Category filter to Customers and open New vs returning customers, grouped by week. Pass: you can see how the split between first-time and returning customers moves week by week. Fail: you skip it, and the size of Robyn's baseline comes as a surprise.
2. See if your peaks repeat. In Shopify, open Total sales over time, group it by week, then click Compare to and choose Comparison to past for the previous year. Pass: the seasonal peaks line up year on year, which Prophet's season term can learn. Fail: they shift with the weather, so add a temperature column.
3. List your promotions by week. In Shopify, go to Discounts, click Export and choose All discounts. Pass: every promotion has dates you can turn into a weekly discount column. Fail: undated sales, which a model may credit to whatever channel spent most that week.

Check the homework. Your GA4 Attribution paths export already holds the evidence. Causality Engine reads that one file and shows what each channel caused next to what last-click gave it, in 1 to 2 minutes, for €99 once (excluding VAT), refundable within 30 days. Check the homework

Sources, 1 October 2026: Key Features (Meta); An Analyst's Guide to MMM (Meta); Customers reports (Shopify); Setting and comparing time ranges for your reports (Shopify); Exporting discounts (Shopify)

Related answers

• All MMM & measurement questions
• What is Meta Robyn?
• How to run your first Meta Robyn model, step by step
• Time series analysis, defined