As of November 8, 2026, the most useful number in grocery technology is not a delivery time. Instacart reported 1.095 billion dollars of revenue with 147 million dollars of GAAP net income, and its advertising line is the reason that split is possible, because fulfilment stays capital heavy and thin while the data layer on top of it does not. Bruno Digital makes the point plainly: the margin sits in data and advertising, not in moving a basket from a store to a door.
Where the margin sits in a grocery platform
The structure is easy to misread as a delivery business with an ad sideline. It is closer to the reverse. Delivery generates the transaction record, the basket composition and the substitution event, and those three data points are what make a sponsored placement worth buying. A platform that stops investing in fulfilment loses the data; a platform that only invests in fulfilment never earns the margin. That is why the capital keeps flowing into dark stores and slot capacity even while the profit line is defended by advertising.
For a brand, the practical consequence is that the shelf inside an app is rented space with a rent that rises as the platform matures. Retail media is no longer a test budget appended to trade spend. In categories where search results decide the sale, it is the mechanism that determines whether a product is discovered at all, which means the media plan and the distribution plan are now the same document.
Seen as a consumer trend rather than a media story, the shift is simple to state. Shoppers now start a grocery trip inside an app for a third of their baskets, and the app decides which brand is seen first. That is a market trend with a budget consequence, because the money that used to buy shelf space in a store now buys position in a search result.
What a 16 percent online share changes
| Banner type | Online share of sales | Media stack | What the brand still controls |
|---|---|---|---|
| Regional grocery, 504 stores | About 16 percent, online up more than 12 percent | Retailer network built on a platform partner | Assortment depth per store and pack size |
| National marketplace with many banners | Not disclosed per banner | Platform-wide auction plus retailer add-ons | Listing quality and price position |
| Single-banner campaign | Small absolute volume | Retailer-only placements and brand pages | Creative sequence and claims |
The 16 percent figure matters because it is past the point where a brand can treat online as a rounding error and short of the point where it can treat it as the main channel. A store estate of 504 locations with a meaningful online share means the same shopper appears twice in the data: once as a trip and once as an order. If those two records are never joined, the brand reads a stable shelf share while losing the digital shelf, which is where the next purchase is actually assembled.
There is also a cost consequence that rarely appears in a media plan. A retailer network that runs on a platform partner gives the brand two places to spend against one shopper. Campaigns bought inside the platform already reach the retailer's storefront, and a retailer-specific campaign reaches the same shopper again from a second budget. Two reports then answer two different questions, and adding them together stops showing which budget produced the sale. That is a measurement problem before it is a spending problem.
Two budgets, one shopper
The clean way to run this is to separate the two buys by the decision they are supposed to influence. A platform-wide campaign buys reach across banners and should be judged on new-to-brand buyers. A banner-specific campaign buys position inside one retailer and should be judged on share of category within that retailer. When the same team owns both budgets and reports them as one line, the cheaper channel always looks better, because the expensive one is doing the work the cheap one is credited for.
Incrementality before bid
The other habit worth building is to test incrementality before optimising bids. A sponsored placement that appears above an organic listing the shopper would have clicked anyway converts well and sells nothing new. Running a holdout in a matched set of stores, then comparing units per store against the control, costs a fraction of a quarterly budget and tells you whether the placement is doing work. Without that test, the bid strategy is optimising for the report rather than for the shelf.
A four-step check before the next media plan
The reconciliation step is the one teams skip. A media plan can raise search impressions sharply while the item is out of stock in a third of the estate, and the report will still show improving return on ad spend because the clicks that converted were the ones in stores that had stock. Comparing spend with coverage by store turns that into a visible gap, and it usually explains a decline that the media team cannot.
FAQ
Is advertising really more profitable than delivery for a grocery platform?
On the reported split, yes. Advertising revenue is built on data the fulfilment operation already produces, so the incremental cost of serving an extra impression is far below the cost of serving an extra order.
Does a 16 percent online share mean stores matter less?
No. It means the same shopper is recorded twice. Store assortment still decides what can be sold online, so coverage remains the constraint on growth.
Should a brand buy the platform campaign or the banner campaign?
Both, but with separate metrics. Platform campaigns buy reach across banners; banner campaigns buy position inside one retailer. Blending them hides which one worked.
How often should coverage be checked against media spend?
Monthly at minimum, and weekly during a peak. Platforms add, pause and convert stores continuously, so a quarterly check misses the weeks that matter.
