If you make supermarket ads for a living, you already know the asymmetry. One film a year gets remembered and gets all of the plaudits. The few hundred creatives that went out last month got about a second and a half each in a feed, and those were the ones actually carrying the offer.
They aren't the same job. The famous work buys recognition that you'll be able to take advantage of for months to come. The weekly work carries the volume and the deadlines. Almost everything written about supermarket advertising is about the first kind.
This is about both.
The short version. The best supermarket ads repeat rather than reinvent: five of Kantar's ten most effective festive ads in 2025 were reruns of earlier work. They come in ten formats, each serving a different purpose, so a brand banner and a weekly offers ad shouldn't be judged by the same metric.
Grocery is one of the few categories where somebody publishes the marks. System1 and Kantar test the field every festive season, so "best" can mean something narrower than "I liked it": it moved share, and it kept moving it.
The first thing that shows up in the data is repetition, which is roughly the opposite of what a creative review rewards. Kantar found five of 2025's ten best-performing festive ads were reruns rather than new work.
The best supermarket ads aren't one kind of ad at all: a brand banner and a weekly offer look alike and do opposite jobs. So here are 10 of the best supermarket ads, grouped by the job each one does and the metric each should answer to.
For each one we look at: what it is, why it works, what tends to go wrong.
| Format | The job it does | Judge it on |
|---|---|---|
| Brand version | Buys memory. No price, no product | Recognition and share over months, not clicks |
| Price promotion | Shifts a named SKU before the week ends | Sales lift on the promoted line |
| Recipe version | Reaches the cook decision before the shop decision | Engagement and assisted visits |
| Recipe feed ad | Serves a whole recipe library from one template | Coverage: how much of the library actually ran |
| Weekly offers ad | Corrects prices everywhere at once | Accuracy and speed from change to live |
| Compare slider | Two messages in one placement | Interaction rate, then which side won |
| In-banner search | Takes a request instead of asking for attention | Quality of the landing, not click volume |
| Offers gallery | Many SKUs per placement | Which SKU earned the click |
| Category grid | Lets the reader self-segment | Click distribution across the four tiles |
| Omnichannel master | One template across every channel | Production hours per campaign |
It's the only unit in the plan doing memory work, and the only one with nothing in it to date, which is why a good one runs for years while the offers around it turn over weekly. The hard part was never the creative; it's the budget meeting, because you can't tie a brand banner to a basket.
Aldi's decade of Kevin the Carrot is the strongest answer to that objection anyone has produced. Ten years in, Aldi held 10.8% of the UK grocery market in the twelve weeks to 12 July 2026, against 6% when the carrot first appeared.
Take the brand template and flip the hierarchy. Price becomes the biggest thing on the canvas, and a badge carries the mechanic, whether that's two-for, was-and-now, or a percentage off.
Nobody reads a promotional banner. They scan it for a number and a mechanic, which is why the unit works by keeping two things legible and refusing a third. A starburst says "offer" before a single word has been processed.
It isn't a sideline, either. A quarter of all FMCG volume sells on promotion, on NIQ's August 2026 figures, so the unit carrying your offers is carrying a quarter of the business. Design it fresh every week, though, and price accuracy quietly becomes a design problem.
A value idea can carry a whole campaign on its own. Lidl put one into all 31 of its markets in September 2025 with "Lidl. More to Value.", the first time it had run the same line everywhere.
Light ground, food leading, no price anywhere near it.
This is the only retail format that earns attention instead of interrupting for it, and that comes down to timing. Recipe intent runs ahead of basket intent. Someone deciding what to cook on Thursday has not chosen a shop yet, so you decide before price becomes the question. Put a price on it, and it becomes a promotional ad with better photography.
Waitrose has been doing this for decades and took 5.9 stars from System1 with it in 2025.
Here, the title, the cook time, and the image come from columns in a data source instead of being typed into the creative. Small change, and it's the whole point: the creative becomes a frame, the library becomes the content, and one unit carries hundreds of recipes without a design cycle. A content library you already own turns into media inventory.
Product name, pack size, current price, previous price. Four fields, refreshed weekly, multiplied by every size, region, and language you run.
Boring and repeatable, and both are virtues. Shoppers see this one every week for years, so a familiar layout beats a clever one. What makes it good is correctness at scale: with price as a feed field rather than baked artwork, a correction is updated everywhere at once. A wrong price isn't embarrassing; it's a compliance problem, which makes this the one format where how it gets produced is part of whether it's any good. If someone has to reopen a template to change a number, that is where errors begin.
Two images, one draggable divider. Breakfast against dinner, own-brand against the brand it's imitating.
The interaction is the factor for this ad; it's not just a design element. Dragging makes the comparison something the reader did, and a conclusion you reached yourself is harder to dismiss than one handed to you. You also get two messages into one placement without buying another. Only worth doing where the two images genuinely compare.
Put a working search field in the ad. Somebody types an ingredient, and the click lands on a result.
That inverts the whole transaction. Instead of asking for attention, the ad takes a request, so what arrives is self-selected and comes with a stated need. The queries are worth having on their own too, since they read as free demand data. It does need a real result page behind it.
Pull the weekly offer feed into a carousel and rotate it inside a single placement.
It means relevance stops being a gamble on a single product. A single-product banner bets the whole impression on one SKU happening to interest whoever sees it, where a gallery gets three or four chances without buying more inventory. If you're running hundreds of weekly offers against a fixed number of placements, that maths is the whole argument. Rotate it more slowly than feels necessary.
Four tiles in one square: fruit and veg, bakery, dairy, frozen, each with its own offer.
The reader picks their own category, so one unit covers four audiences and the landing page can match whichever tile got pressed. It's the cheapest relevance available, because it needs no personalization data at all. The reader does the targeting for you. Four near-identical discounts, though, and there's nothing to choose between.
Coop Denmark runs it across 900 stores and three retail brands: SuperBrugsen and Kvickly, Dagli'Brugsen, and 365discount. The same template reaches six touchpoints: social, YouTube takeovers, display, its own app, email, and the digital screens in its stores. Designers build templates that adapt to each format, so one concept reaches somebody in a feed, in the app and at the shelf edge without three separate production jobs.
"When we create a template, it's easy to produce it across different channels. We don't have to produce it for each channel separately," says Klaes Simonsen, Senior Digital Strategist at Coop Denmark.
The precondition is flexibility. In Coop's own words: "We have a new concept more or less every week, so we need flexibility to constantly evolve and do something new."
Creative is the biggest lever in any of this, and it is the one factor a lot of teams prioritize the least.
EMARKETER and Perion surveyed US marketers and agency staff managing over $5m in annual media spend: 89.2% said creative matters, 3.6% actively optimize it. More than half get creative feedback slower than they get media signals, and 41.4% see nothing on creative until two to four weeks after launch. In grocery, the offer has changed multiple times by then.
Turnaround expectations are shifting under all of this too. Walmart Connect reported cutting median creative production time in its Creative Hub by roughly 80% in beta, on its own 2025 data. UK retail media reached £3.8bn in 2025, up 18% year on year, and every new placement needs its own set of sizes.
So the ten sort into groups. One to three are versioning problems, where one template covers several campaign jobs. Four, five and eight are data problems, where creative gets populated from a feed instead of typed by hand. Ten decides what both of those cost. Coop's answer was movable templates plus one connected system for design, media and brand, so approvals stopped queuing behind production. The output shows what that unlocks: more than 31,000 creatives and over 6,600 creative sets built from shared templates, delivering upwards of 5.6 billion impressions. Read the full Coop case study for more insights on their approach.
If your week looks anything like Coop's, the formats above are a fair map of where the hours go. See how Bannerflow handles it.
A recognizable device, whether that's a character, a sound or a line, repeated for years, plus specific rather than vague price messaging. Kantar's 2025 analysis found five of the ten most effective festive ads were reruns.
On System1's testing, Waitrose's "The Perfect Gift", Asda's "A Very Merry Grinchmas" and M&S Food's "Traffic Jamming" all scored 5.9 stars, roughly 2.5 stars above the supermarket norm, with Aldi's "The Proposal" on 5.8.
Not all on the same metric. A brand banner is buying recognition and should be judged on share over months, while a weekly offers ad should be judged on price accuracy and speed to live. Aldi is the clearest illustration: ten years into Kevin the Carrot, it held 10.8% of the UK grocery market in the twelve weeks to 12 July 2026, against 6% when the campaign began.
More than most teams act on. A 2026 EMARKETER and Perion survey of marketers managing over $5m in media found 89.2% say creative matters while only 3.6% actively optimize it, and 41.4% see no creative insight until two to four weeks after launch. In grocery, where offers turn over weekly, that feedback arrives after the campaign it describes.