Meta's New Creative Diversity Metric: What 'Low' Means for Your D2C Ad Sets

What Meta's Creative Diversity metric actually measures

If you opened Ads Manager in the last few weeks and spotted a new column grading your ad sets "Low," you're not imagining it. Meta rolled out a Creative Diversity metric in early August 2026, and Search Engine Land documented it on 26 August. It sits quietly alongside your usual columns and puts a one-word verdict on each ad set: Low, Medium or High.

The metric's job, in Meta's own framing, is to estimate how visually varied the images and videos inside an ad set are. Not how many you've uploaded. Not how well they're performing. Just how different they look to Meta's systems.

That distinction trips people up immediately, so it's worth sitting with. You can have six creatives in an ad set and still get rated Low if those six are basically the same photo with different captions.

Where to find it

You won't see it unless you go looking. Open Ads Manager, click the Columns dropdown, choose Customize columns, and search for "Creative diversity." Add it, save the view, and the rating appears at the ad set level. It's a good idea to bake it into a saved column preset you actually use, because a metric you have to hunt for is a metric you'll forget exists.

What Meta will and won't tell you

Here's the honest part. Meta has not published the exact criteria behind the score. Nobody outside Meta knows the precise thresholds that separate Low from Medium from High, and the company hasn't shared them. Media buyers testing it have reported ad sets rated Low even when they were running a genuine mix of formats, which tells you the scoring is stricter, or at least fuzzier, than it first appears.

So read this metric the way you'd read a smoke alarm, not a thermometer. It's a signal that something may be worth checking, not a precise measurement you can calibrate against. Anyone selling you a guaranteed formula to hit "High" is guessing.

Why this showed up now: Andromeda and the creative-hungry algorithm

This metric didn't appear in a vacuum. Meta's ad system has spent the last two years moving toward machine-led delivery, and its retrieval engine, Andromeda, is built to sift an enormous pool of creative and match individual people to the specific ad most likely to move them. That only works if you give it variety to sift.

Think about what the machine is doing. For every impression, it's trying to pick the one creative, from everything you've handed it, that fits this particular person in this moment. If everything you've handed it is a variation on one idea, the machine has nothing to choose between. The Creative Diversity metric is Meta telling you, out loud, when your ad set has starved its own algorithm.

Volume was never the point

Indian D2C teams often respond to "make more creatives" by producing twenty versions of the winning ad: same hook, same model, new background, new discount sticker. It feels like productivity. To Andromeda, it reads as one idea wearing twenty costumes. The diversity score is, in part, a correction to that habit. More assets is not more information.

"Low" isn't an insult, it's a map

The most useful reframe is to stop treating the rating as a grade and start treating it as a diagnosis. Each level is pointing at a different likely problem and a different fix.

Notice that even a "High" rating comes with a warning. As you scale a winner, Meta shifts budget toward it, your other concepts starve, and an ad set that started diverse quietly collapses back to Low. Diversity isn't a box you tick once.

The D2C diversity checklist: the axes that actually move the score

When brands try to diversify, they almost always reach for the wrong lever, colour and copy, because those are the cheapest to change. Real diversity lives on axes that cost more effort but carry more signal. Here's where to spend it.

Hook

The first two to three seconds are the single biggest lever. A problem-agitation open, a founder talking to camera, an unboxing, a before-and-after, a customer testimonial, a "three reasons" list, these are structurally different hooks, and they pull structurally different people. If all your creatives open on a beauty shot of the product, you don't have diversity, you have a catalogue.

Format

Static image, single video, carousel, and collection are genuinely different surfaces, and they behave differently across Feed, Reels and Stories. A brand running only 9:16 video is invisible to a chunk of Feed inventory. Mixing ratios and formats is one of the fastest, most legible ways to broaden an ad set.

Messenger

Who delivers the message changes everything. A polished brand edit, a raw UGC clip from a real customer, a creator in their own voice, and a founder explaining why they built the thing, are four different messengers. For Indian D2C especially, a regional-language UGC creator can reach an audience your studio-shot English video never will. This is where thoughtful performance marketing earns its keep: the creative pipeline, not just the media buying, is what feeds the algorithm.

Offer and angle

Same product, different reason to buy. One creative leads with price and a festive discount, another with a trust signal like reviews or free returns, another with a specific use-case ("gifting for Diwali"), another with an objection-buster ("machine washable, actually"). Angle diversity is invisible in a thumbnail but very real to the person deciding whether to tap.

What Indian D2C brands get wrong about "diverse" creative

Two mistakes show up again and again, and both quietly keep ad sets stuck at Low.

Reskinning is not diversifying

The template trap is seductive because it scales. You build one high-performing layout in your design tool, then spin out fifteen SKUs on the same frame. Efficient, yes. Diverse, no. If a stranger scrolling past couldn't tell your fifteen ads apart in half a second, neither can Meta's model in any way that helps. Keep the template for catalogue and retargeting, but your prospecting ad sets need concepts that don't share a skeleton.

The COD and WhatsApp angles you're ignoring

D2C brands sitting on India-specific trust levers rarely put them in creative. Cash on delivery, WhatsApp support, easy returns, and same-city dispatch are exactly the objections that stop a first-time buyer, and each one is a distinct creative angle you're probably not running. A creative built entirely around "Pay only when it arrives" is a genuinely different message from your hero product video, and it broadens both your diversity score and your actual audience.

How to work the metric without chasing a vanity score

The trap now is obvious: teams will start optimising for "High" the way they once chased a perfect relevance score, and lose the plot. Don't. The metric is a means, not the end.

A simple weekly routine

Keep it light. Once a week, add the Creative Diversity column to your view and scan your prospecting ad sets. Any that read Low or Medium go on a shortlist. For each one, ask a single question: do these creatives differ by hook and messenger, or only by colour and caption? If it's the latter, you've found your next creative brief. Then close the column and get back to the numbers that pay the bills.

Because here's the order of operations that matters. Incremental sales come first. Cost per acquisition and contribution margin come next. Creative diversity is a supporting input that helps those numbers over time by giving the algorithm room to explore, but a "High" score on an ad set that isn't selling is worth nothing. If an ad set is Low but printing profitable orders, leave it alone and go find the next winner.

Meta added this metric because its own systems run better on varied creative, and telling you the score is cheaper for them than watching you feed it the same ad twenty times. Use it in that spirit. Let it nudge your creative planning toward real difference, hooks, formats, faces and offers, and let your sales numbers, not a one-word rating, decide what actually stays live.