Nest Creative AI Intelligence – what 10,000+ ads reveal
Nest Creative AI Intelligence
The creative patterns that actually drive performance aren’t necessarily the ones most brands are testing.
The creative patterns that actually drive performance aren’t necessarily the ones most brands are testing.
Most creative briefs are built on instinct. The photographer knows what looks good. The creative director knows the brand. The performance team knows what ROAS they need. But nobody is systematically asking: across hundreds of ads and dozens of brands, which visual patterns actually correlate with results?
We can now answer that question. Nest Creative AI, our end-to-end performance and creative solution, connects to ad accounts and analyses every creative at scale, surfacing patterns no human analyst could find manually. It breaks down top-performing ads into Winning Patterns: AI-extracted insights into which compositions, lighting, colour palettes, subject positioning and messaging treatments are driving performance. From there, it synthesises these into Macro Trends that can be used to shape the next brief for Creative Production inside the tool.
We normally run this per brand. For this report, we ran it across our entire portfolio. Over 10,000 live ads across fashion, health, subscriptions, and homeware. The question: which creative signals hold across categories, not just within a single account?
Nobody is systematically measuring which visual details drive performance across an entire portfolio. Until now.
Six patterns emerged. Some confirm what good creative teams already sense. Others challenge assumptions most brands take for granted.
Six things we found that should change how you brief creative.
Show the hands. It’s the most common pattern we found.
Nest Creative AI analysed over 10,000 ads from our portfolio of ecommerce brands. Hands-with-product emerged as one of the most persistent patterns in every category we looked at. Across the set that meant a model adjusting a cuff, a hand running across wood grain, a case being opened, a book held at reading angle. In the vast majority of cases it correlated with stronger CTR, CVR and ROAS.
The likely reason is that hands create a sensory bridge. They turn a product image into a tactile experience, helping the viewer imagine owning the item before they buy it.
It was a consistent winner across the portfolio. If your product shots don’t put it in someone’s hands, that’s an easy gap to close.
Grid layouts are outperforming single-hero shots.
The standard ecommerce ad is a single product, hero-shot. But across nearly half the brands we analysed, multi-panel and grid layouts outperformed single-image compositions in the accounts where both ran. The strongest result was a +272% ROAS lift, and the pattern was positive everywhere it appeared.
Grids likely work because they reduce the pressure on any single image. They show range, context and detail at once, giving the viewer more reasons to stop and more ways in.
If your creative still defaults to the single hero shot, the grid is the obvious thing to test against it.
Earth tones kept winning. So did bold colour.
Warm, earthy palettes (muted greens, terracotta, sand) showed up as a performance pattern across 39% of brands, correlating with ROAS improvements in 88% of cases. But vibrant, saturated palettes drove ROAS too. What consistently lost was washed-out, default colour.
What seems to matter is committing to a palette at all. A deliberate one, warm or bold, gives the creative a point of view; indecisive colour gives the viewer nothing to hold onto.
Warm tones were the most frequent winner in our data, but the broader pattern is that brands committing to a clear palette outperformed those without one.
Texture sells. Close-ups are an underused lever.
Close-ups of product texture and material quality correlated with conversion improvements across nearly half the brands we analysed, holding across fashion, cashmere, homeware, flooring and jewellery.
Every online buyer wants to know what a product feels like, and a screen can’t tell them. A macro shot of weave, grain, stitching or finish gets closer than any line of copy can.
Macro texture is cheap to produce and quick to test.
Eye contact stops the scroll. But looking away might close the sale.
Direct eye contact appeared as a positive signal in a third of brands, primarily driving conversion rate improvements. It works because it creates an immediate personal connection: the model is looking at you, not past you.
But there’s a counter-signal in the data. In three brands, strong performance came from models who weren’t looking at the camera at all. Walking away, gazing off-screen, mid-movement. These candid compositions drove higher ROAS and CVR in the instances where they appeared.
Our read: direct gaze is a scroll-stopper that commands attention. Candid composition signals authenticity, and audiences that feel like they’re observing a real moment rather than being sold to convert at higher rates.
The takeaway: both work, but for different reasons. If your creative testing is only producing one type of model direction, you’re missing half the picture. Test both deliberately and let the data tell you which converts for your brand.
Soft lighting kept outperforming hard lighting.
Soft, diffused lighting appeared in 28% of brands, a moderate frequency. But it was the most consistent pattern in the dataset. Where it appeared, it appeared prominently, and it correlated reliably with ROAS.
The reason is familiar to any photographer. Soft light flatters both product and skin and reads as premium, while hard light is far easier to get wrong.
It was one of the most consistent patterns in our data. On a shoot it’s also one of the simplest things to control, which makes it a low-effort lever to get right.
From analysis to production
This report is Nest Creative AI’s analysis, run across our whole portfolio rather than a single account. It does the same thing for any brand we work with, reading every ad in the account and finding the patterns a human couldn’t spot at that scale, down to which angles and formats are performing.
Analysis on its own is still just a report. The difference is what Nest Creative AI does with it. The same platform turns those patterns into new creative, with your brand guidelines fed in before a single ad is made and human sign-off at every stage. It’s how a brand goes from 50 ads a month to 500 without losing what makes it recognisable.
Keen to see what Nest Creative AI can do for your ads?
Get in touchMethodology
This analysis used Nest Creative AI’s Creative Intelligence module, which connects to ad accounts and uses AI to extract visual attributes from top-performing creatives: composition, lighting, colour palette, subject positioning, format and messaging.
We ran it across our portfolio covering over 10,000 ads active during May 2026. 229 distinct creative patterns were identified. We grouped these into 32 standardised categories and mapped them across brands. A pattern was considered cross-client if it surfaced independently in at least four brands spanning two or more product categories.
Metric associations are directional. Frequency-based claims (e.g. “44% of brands showed this pattern”) are derived from normalised cross-client data and are stable.
