Last week DoorDash announced a $17 billion acquisition of Instacart. Modern Retail published a podcast on it on 7 November 2026. Its reporters say DoorDash, Uber and Instacart now describe themselves as “discovery platforms” and no longer only as logistics providers. Shoppers increasingly open these apps to find new products, try new brands and compare deals. The apps have used that shift to move into new retail categories and build ad businesses aimed at brands. The episode also covers how AI is changing these platforms.
The excitement is about who owns discovery
The story being told is that the delivery app is becoming the new shop window. If shoppers start their search there, whoever runs the app decides which brands they see. Add AI-driven shopping and the app doesn’t just list products. It suggests them.
I ask the same question I ask of any launch. Is there a real need? Partly, yes. A shopper with a long list of similar items does want help choosing. Someone has to narrow the options, and an app that sees what people buy is placed to do it.
What changed is where the choosing happens. It used to happen in the shopper’s head, looking at a page of products. Now more of it happens inside the platform, before the shopper sees anything.
What didn’t change is that the platform can only recommend what it can read. The article gives no detail on how the AI features work or how much the ad businesses earn, so I won’t guess at either. The direction is clear enough without that.
Your item data is what gets read
Think about a small shop selling on these apps. A neighbourhood bakery, a health food store, a pet supplies shop. On each app, every product is a record. It has a name, a size, a list of ingredients, a stock status and a price.
That record is the thing being compared and recommended. A shopper asks for a nut-free snack for a child, or the cheapest 2-litre option that is in stock. The system can only answer from the fields it has. If your listing says “Family loaf” with no weight, no ingredients and a stale stock count, there is nothing to match. A competitor with the full details gets picked.
This is where I land on agentic commerce. For smaller businesses the shift is already decided. Get data-centric. Your website is one surface among many now. Wherever your data shows up, it needs more precision and more detail, and it needs to say accurately what you are.
Precision alone isn’t enough. A recommendation goes to the product with enough attributes that are materially and measurably different from the alternatives. “Delicious” and “premium” are adjectives, and engines drop them. “420 g, no palm oil, baked daily, contains sesame” is a set of facts that can be matched and quoted.
The questions are also getting more varied. Everyone asks their agent the same thing in slightly different words. To catch that, your listings need to cover a wider spread of attributes, not a longer description.
Ads are a separate lever
The article makes clear these platforms are building ad businesses for brands. Ads can put you in front of shoppers. They are a different lever from the quality of your item data.
If you pay for placement and your listing is thin, you are paying to show an unclear record to people. If your data is precise, you are readable whether you pay or not. A small merchant with a tight budget should fix the data first, because it costs nothing but attention and it supports every surface, paid or not.
What this has to do with my own work
I work on this on the search side. In GEO the pattern is the same. The AI doesn’t pick the biggest brand. It picks the one it understands most clearly and can quote most precisely, and you can work on that. AeroChat publishes many pages that each target one integration with specific software, and each one catches a single long-tail question.
What a business should do
Start with an audit of what each app shows for your ten best-selling items. Check size, ingredients, stock and price against what is true today. Fix the gaps and the stale fields.
Then look at what makes each item different from the closest competitor on the same app. Write that down as facts with numbers where you have them. Remove adjectives that no one can check.
Keep one master record for each item so every app gets the same facts. Update stock on a schedule, because a wrong stock status is a data problem too.
Only after that, decide whether to spend on ads.
Source: Modern Retail: How delivery apps came to be ‘discovery platforms’