Why creative is now the biggest lever in paid media

For most of the last decade, paid media was won in the settings. Tight audiences, clever exclusions, careful bid strategies. That era is over. On both Meta and Google, the algorithms now do the targeting, and the single biggest input you still control is the creative you feed them.

That is not a soft, "brand people" argument. It is how the platforms are now engineered. Meta has rebuilt its ad delivery system around reading what is actually inside your ads. Google's automated campaign types live or die on the quality and range of the assets you upload. If your creative is thin, repetitive or tired, no amount of account structure will rescue performance.

The good news is that the cost of producing great creative has fallen dramatically. The latest AI video models mean brands of any size can now produce the volume and variety these platforms demand, even when a full production shoot is not in the budget.

Meta Andromeda: your creative is now your targeting

Andromeda is Meta's ad retrieval engine, the first step in deciding which ads anyone sees on Facebook and Instagram. According to Meta's own engineering team, retrieval narrows tens of millions of eligible ads down to a few thousand candidates for each person. Only those candidates go on to the auction.

"Andromeda improves performance of Meta ads system by delivering more personalized ads to viewers."

Engineering at Meta

The important change is how that shortlist is built. The old system leaned on separate models and rule-based shortcuts. Andromeda uses a far more powerful neural network that personalises much earlier, and Meta reports a 6% improvement in recall and an 8% improvement in ads quality on selected segments since rolling it out.

In practice, this means the system is matching individual people to individual ads, based on what is actually in the creative. Meta has said as much in its Creative Advantage guidance for advertisers: the main lever for finding the right audience has moved away from niche targeting and towards creative diversification.

What "diversification" means matters. It is not 10 versions of the same video with a different headline. It is genuinely different ideas: different hooks, formats, settings, people, tones and angles, each built to speak to a different type of customer. Give Andromeda 3 near-identical ads and it has 1 idea to work with. Give it 10 distinct concepts and it can find the right person for each.

The brands winning on Meta right now are the ones feeding the machine a steady supply of varied, fresh creative. The ones struggling are usually running a handful of tired assets and wondering why costs keep creeping up.

Google Ads: automation runs on assets

Google has been making the same point for a while. Its Creative Performance Best Practices guide cites NCS research showing creative is responsible for 49% of the total sales impact of advertising. Nearly half the result, decided before a single bid is set.

"All marketers want more and better creative."

Vidhya Srinivasan, VP and GM, Ads and Commerce, Google (Google Marketing Live 2026)

That matters because Google's growth campaign types are built on assets. Performance Max and Demand Gen take the headlines, images and videos you supply and assemble them across Search, YouTube, Shorts, Discover, Gmail and Display. Google's guidance is to give the system both quality and quantity, so its AI can test combinations and find what works for each person.

Video is where many accounts fall short. Google's own data shows Performance Max advertisers who included at least 1 horizontal, 1 vertical and 1 square video delivered 20% more conversions on YouTube than those using horizontal video alone. Plenty of brands are still running PMax with no video at all, which leaves Google to auto-generate something generic on their behalf.

Google is so convinced creative is the bottleneck that it now builds generation tools into the platform. Its creative hub announces Veo in Google Ads, which Google describes as its "most capable generative video model", now available in Asset Studio. When the platform itself starts handing advertisers an AI video model, the message is clear: more, better video is what moves performance.

The challenge: keeping up with the volume

Put those 2 platforms side by side and the demand is obvious. Meta wants genuinely distinct concepts. Google wants video in every orientation, refreshed regularly. Both reward brands that keep feeding them new ideas and penalise those that don't.

A professionally shot film or photoshoot is still the gold standard. Real people, real places and a skilled crew create work that nothing else quite matches, and for a flagship campaign it is often exactly the right call. But a shoot means a crew, a location, talent, kit and weeks of editing, and that budget puts it out of reach for many brands, especially at the volume Meta and Google now reward.

It is not always needed, either. A new hook to test, a seasonal variation or a vertical cut for Reels rarely justifies a full production day. So many brands end up with 1 shoot a year and a handful of assets doing all the work, or with no video at all, leaving YouTube, Reels and Shorts largely untouched.

That is the gap: between what the platforms reward and what most brands can realistically produce. The answer is not to replace great photography and film, but to fill in around it.

AI video: a whole new world of creative opportunity

The latest generation of AI video models closes that gap. They produce realistic motion, lighting and sound, and they can create footage that would be impossible or impractical to film. A product on a mountain summit at sunrise. A kitchen in 5 different interior styles. The same advert set in summer, winter and on a rainy high street. No crew, no location fees, no weather risk.

For smaller brands, it changes the rules

Until now, high-end video was a big-budget game. Smaller brands simply could not afford the volume of polished creative that Meta and Google reward. AI video changes that equation completely:

• Volume: produce 10 or more genuinely distinct concepts for the cost of a single traditional shoot
• Variety: test different hooks, settings, audiences and moods, exactly the diversification Meta asks for
• Formats: deliver vertical, square and horizontal versions from the outset, as Google recommends
• Speed: turn a new idea or seasonal offer into live video in days, not months
• Freshness: refresh creative before it fatigues, rather than once a year

It means a growing UK business can now compete on creative with brands many times its size.

And the biggest brands are already doing it

This is not a shortcut reserved for brands on a budget. Some of the world's best-known names have already released AI-generated campaigns, from festive TV spots to big-game adverts, and global agencies are building AI production into their creative workflows.

Google's own ads leadership is pointing the same way. In her 2026 outlook, Vidhya Srinivasan highlights performance apparel brand Rhone using Google's Nano Banana and Veo models to turn existing images into video. She also reports a 3x increase in Gemini-generated assets created by advertisers during 2025.

"Helping advertisers produce studio-quality work in minutes."

Vidhya Srinivasan, VP and GM, Ads and Commerce, Google, on Veo 3 in Asset Studio

They are not doing it to save money. They are doing it because AI lets them explore more ideas, personalise for more audiences and move faster, alongside the photography and film they still invest in. When the largest advertisers in the world are adopting these tools, standing still is the real risk.

Creative direction still matters

The models are only as good as the thinking behind them. Without strategy, AI video produces content that looks impressive but says nothing. The brands that win combine the new tools with proper creative direction: a clear brand look, a sharp message and a structured approach to testing what works on each platform.

Introducing Clicky AI Video

This is exactly why we have launched Clicky AI Video. We are not tied to a single tool. Our AI production system works across the leading video models, choosing the right one for each brief, so we can create footage that would be impractical or impossible to film. And if you already have great photography or video, we build on it, turning an existing shoot into more concepts, formats and seasonal versions.

The film at the top of this page is one of ours: a sportswear campaign for Stride, an invented brand, made without a single day on set. Here are a few more, each a different idea for a different kind of business.

That is 5 completely different concepts. To Meta's Andromeda, that is 5 ideas to match to 5 different types of customer, not 1 idea in 5 colours.

Made for every platform

Because it sits alongside our paid media team, every piece of creative is built with the platforms in mind from the first frame, not cropped to fit at the end. Here is how we make each one.

Meta: Facebook and Instagram

• Reels and Stories: full-screen vertical 9:16, with the hook in the first 2 to 3 seconds and captions on screen, because most people scroll with the sound off.
• Feed: 4:5 or square 1:1, which fills far more of a phone screen than a landscape video does.
• Andromeda: we plan a set of genuinely distinct concepts rather than variations of one, each written for a different audience, motivation or moment, so the system has real ideas to match to real people.

Google: Performance Max and Demand Gen

• All 3 orientations: every film is delivered as horizontal 16:9, vertical 9:16 and square 1:1, the mix Google's own data links to 20% more conversions on YouTube.
• Several lengths: a short version and a longer one, so the system can place the right cut in Search, Discover, Gmail and Display as well as on YouTube.
• Your own video, not Google's: no more letting the platform auto-generate a slideshow from your product images.

YouTube

• Skippable in-stream: landscape 16:9, with the brand and the hook inside the first 5 seconds, before the skip button appears.
• Bumpers: 6-second cut-downs from the same footage, ideal for reach and for reminding people who have already seen the longer film.
• Shorts: vertical 9:16 versions built for the Shorts feed rather than a squeezed landscape advert.

TikTok

• Native, not polished: vertical 9:16, fast cuts, on-screen text and a pace that feels like the feed it sits in. See our TikTok service for how we run it.

The trick is that we compose every scene for every shape from the start. When a film is reframed to vertical, the frame follows the subject scene by scene instead of sitting dead centre, so nothing important is cut off.

Then comes the testing plan. We launch several concepts side by side, read what each platform tells us, and use that to decide which ideas to scale and which to retire, and what to make next.

Whether you are a growing brand that has never been able to justify a video shoot, or an established business looking to multiply the output of your existing creative, AI video opens up possibilities that simply did not exist a year ago.

The algorithms have made creative the biggest lever in paid media. Now there is a way for every brand to pull it.

Want to see what AI video could do for your campaigns? Explore Clicky AI Video or get in touch with our team.

Clicky™ / ClickyGroup™