How should a food brand read GA4's model comparison?

If you sell food and drink, read the comparison on purchase alone, since every key event is added together by default. Judge a launch month on Ad interaction time, once slow buyers have arrived. Then set each channel's Key events change against its Revenue change: the gap usually shows whether it brings trial orders or full baskets.

If you sell food and drink

If you sell coffee, snacks, sauces or drinks, your site probably asks visitors for more than a purchase. A recipe newsletter, a sample request or a subscription start may all be marked as key events. That helps other reports and muddles this one.

Your orders also come in two sizes. A first order is often a trial pack, a sampler or a single bag. A regular fills a bigger basket. One channel can be good at the first and poor at the second, and the comparison can show which.

Launches are the third habit. A new flavour or a seasonal drink gets a burst of ads in one month. Buyers who need a few visits finish after the month has closed.

The export below is one store's, and it does not say what that store sells.

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 days are the useful part. In one store, buyers on 2 to 3 touches took 12.5 days and buyers on 4 to 9 touches took 16.9 days (Journeys sheet). If your buyers take that long, sales from the last fortnight of a launch month land in the month after.

Single-touch journeys hold 79.5% of revenue on the Journeys sheet, at 0.5 days to buy. Those sales almost always land in the same range, whichever reporting time you pick.

What the export cannot show is order size. It holds shares of revenue, so the trial-pack question needs GA4's own columns.

What changes for a food and drink store?

Sign-ups blur the comparison. Google's report selects every key event by default and adds them together. A recipe sign-up then counts like a sale in the Key events columns, and a channel that collects sign-ups looks like a seller. Pick purchase alone in the drop-down at the top left.

Launch months need Ad interaction time. The default, Event time, credits touches before the purchases in your date range. Ad interaction time credits touches inside your range, even when the purchase comes later. For a launch month, pick Ad interaction time and read it two or three weeks after the month ends.

Key events and Revenue can split. Each model gets both columns, with a % Change for each. Say a channel's key events rise more than its revenue under data-driven. Then the purchases it gains are smaller than the ones it had: trial packs, not hampers. If revenue rises more, it opens journeys that end in big baskets.

A sampler channel can look weak and still feed you. A channel that wins first orders at trial prices shows modest revenue under either model. Before you cut it, check in your shop's own reports whether those first-time buyers come back.

What to do this week

1. Read purchases alone. In GA4, open Advertising, then Attribution, then Attribution models, and keep only purchase in the key events drop-down. Pass: purchase is listed and you can select it on its own. Fail: it is missing, so purchase is not a key event and no model can credit it.
2. Rerun your last launch on Ad interaction time. Set Reporting time to Ad interaction time and the date range to the launch month. Then compare data-driven with paid and organic last click. Pass: you read it at least two weeks after the month ended, so slow buyers are in. Fail: you read it on the month's last day, before they arrived.
3. Set Key events against Revenue for two channels. Compare both % Change columns for Paid Social and Email under data-driven. Pass: you can say which one gains small first orders and which gains big baskets. Fail: you judged a channel on key events alone and called a trial-pack channel weak.

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 event attribution models report (Google Analytics Help); Get started with attribution (Google Analytics Help)

Related answers

• All GA4 how-to questions
• How do I read the GA4 attribution models comparison?
• How to read GA4's attribution models report, step by step
• Causal attribution