Everything Everywhere All at Once split the critics. Peter Bradshaw panned it in the Guardian and just a few days later Mark Kermode called it “multiverse madness with meaning” in the Observer. The Academy went on to give it seven Oscars. I’m with Kermode. I also spent the commute home wondering which version of me is still brand-side somewhere, halfway through a replatform.
That hectic day in the film is closer to retail than I’d like. Nobody is fighting a tax auditor across parallel universes, but every retailer I speak to is running several worlds at once (stores, site, marketplaces, social, the inbox), and shoppers expect each one to recognise them.
Most retailers are already everywhere. The hard bit is making it all work at the same time. And the Golden Quarter has already started. Between Black Friday and the January sales, UK shoppers will spend at least £90 billion.
For Cloudinary’s 2026 Global Retail Survey, run by Researchscape International, we asked 539 retailers in 14 countries what their top priority was this year. They gave us 13. Mobile shopping topped the list at 43%, with high-volume video a point behind. Social commerce, personalisation, sustainability and omnichannel all made the cut. None of them is winning and that’s the problem.
More content, same plumbing
The response, almost everywhere, is more content. In our 2024 video survey, 8% of ecommerce teams hosted 1,000 or more videos. In this year’s retail survey it’s 41%, and 6% now sit on libraries of 100,000 videos or more. Three quarters accept user-generated content in whatever form it arrives.
The systems underneath haven’t kept pace. 45% told us site and app performance is now their biggest operational headache, and 42% lose time manually reformatting video for every device and channel.
Making more content is the easy part. The job for the next 90 days is managing, adapting, moderating, and moving all of it across every channel without the site slowing down.
The plumbing behind the content has always been critical, it’s just increasingly so. The retailers getting from photoshoot to purchase fastest have stopped buying one tool per problem. They stack: one place where images and video are stored, tagged, cropped, personalised and delivered, so each capability makes the next one cheaper.
On its own, speed is a line item in someone’s budget. Stacked with metadata and automation it becomes an operating layer, and operating layers are much harder to rip out.
I’m a headless person (Cloudinary has sat in the MACH Alliance since it launched in 2020), so I’d still buy best of breed for the layers that need it. The stack argument is about what sits underneath them.
Big-company results, small-team resourcing
AG Jeans is one of the best examples I have. The Californian premium denim brand shoots every week and used to park thousands of files in storage with no metadata, then hand-build asset packs for its wholesale accounts. Now every image carries 20-odd points of structured data (who’s in it, which products, when the model rights expire) and flows into its two Shopify stores through Cloudinary’s AssetLink. Milosh Nikodijevic, who looks after ecommerce technology there, says a photo now “knows a great deal about itself”. Automating the wholesale requests off the back of that metadata took 10 to 18 hours of manual work a week off the studio team. The number everyone quotes is the speed one. A second off page load took conversion up 20% year on year, which came to about $1m in additional revenue. Three Black Fridays and counting, no image or video downtime.
Personalisation is the other stress point (35% named it). Bogg, the US brand behind the EVA-foam beach bag that has sold more than 7 million units, is a useful corrective to the idea that it needs a big team. The very small lifecycle marketing team there created a flawless personalised birthday email campaign: one approved Figma design, rendered as a different image for every recipient at send time, with the expiry date baked into the artwork rather than typed into a template. It took under five minutes to set up and lifted click-through 19% in its first three weeks.
Then there’s AI. 84% of the retailers we surveyed already use it somewhere in the workflow, spread fairly evenly across 14 different jobs, and workflow automation alone is at 56%. The number that surprised me was 71% planning to put more AI concierges and digital stylists in front of shoppers this year. I’d go carefully with that one. A generative model isn’t ready to stand in front of a customer and be the brand, least of all in luxury.
I put a question to a room of luxury houses over two dinners in Paris in September: if agents end up doing the buying, what’s the website for? The answer round the table was that the website gets to be the brand again. That changes what it’s judged on, which until now has been conversion. Kermode’s other point about the film was that its best special effect is the emotional wallop, the one most franchise blockbusters never land. Once an agent handles the transaction, that’s the website’s job.
Stack attack
Nobody has this fully worked out, me included. Alpha Waymond’s line to Evelyn is that the universe is so much bigger than she realises. So is the content pile.
What I’d do with the next 90 days is pick the one place where everything already piles up (for most of you, that’s the image and video pipeline) and stack it properly: storage, metadata, automation, delivery, in that order.
AG Jeans’s answer was 20-odd data points on every image and a second off page load. Start there.
Aaron Smedley
Aaron Smedley is GM of the Shopify Ecosystem at Cloudinary and a digital commerce operator with 15 years of experience across brands like Apple and Arc'teryx. He works at the intersection of retail operations, customer experience, and Shopify transformation, helping brands scale without sacrificing speed, consistency, or performance.
In his current role at Cloudinary, Aaron works with merchants and partners to turn visual content into a performance asset across their storefronts and channels, making the “visual layer” of commerce as measurable as the media that drives traffic to it.
