How AI adoption in ad production split across teams in 2026

AI adoption in ad production is universal, execution is not

AI adoption in ad production crossed the line from experiment to default in 2026. XR Extreme Reach's State of Ad Ops study, fielded with more than 400 advertising professionals across the US and UK, finds 88 percent of marketers on both the agency and brand sides are now using or piloting AI inside creative production. The technology is no longer a pilot parked in an innovation lab; it is on the floor of working campaigns.

The headline number hides the more useful story, which is variance. Nearly half of those teams use AI every day, while a meaningful slice still treats it as an experiment they have not wired into the actual workflow. That gap between intent and output is the real {{link}} most teams still haven't closed. For production leaders the question has moved on from whether to use AI to where it earns a durable place in the pipeline.

XR's own read is that skepticism has not slowed adoption. Graham McKenna, the company's CMO, notes that teams are drawing clear lines between output and originality, letting AI tackle the busy work while keeping the big ideas in human hands. That framing explains the pattern: AI is hired for the parts of production nobody wants to do by hand, not for the parts that define the brand.

That gap between intent and output is the real video adoption execution gap most teams still haven't closed.

Where AI Actually Does the Work

The task-level data shows AI is concentrated in the finishing and testing layers of production, not in the original idea. Across all respondents, visual effects and compositing led at 45 percent, followed by performance analysis and creative testing at 44 percent and image generation at 44 percent. Video and motion generation landed at 42 percent, script and copywriting at 41 percent, and ad versioning and adaptation at 38 percent.

Read together, those ranks describe a post-production and iteration profile. Teams are handing AI the busy work of refining, resizing, and stress-testing creative rather than asking it to conjure the big concept from a blank page. The versioning and adaptation layer is where {{link}} make the marginal cost of another cut effectively zero. For high-volume performance campaigns that run dozens of platform-specific edits, that is precisely where the technology pays for itself.

The implication for creative teams is that AI is now a finishing tool before it is a generative one. A motion designer can produce forty resized cuts in the time it once took to hand off five, and a compositor can test three looks before the client meeting instead of one after it. The creative concept still has to come from somewhere; what changed is how much iteration a team can afford before the idea goes live.

The versioning and adaptation layer is where AI video variant economics make the marginal cost of another cut effectively zero.

Why Brands Lag Production Companies

Adoption splits sharply by who is actually doing the work. Production companies are furthest along, with 79 percent already using AI in workflows or piloting it on specific projects. Full-service agencies have made it a daily habit at 62 percent, leaning on AI to scale VFX, run ad testing, and accelerate storyboarding. National and global brands sit earlier in the process at 42 percent.

Brands mostly use AI for brief writing, image creation, scripting, and voice-over, the upstream steps that stay close to strategy. Media agencies trail the field at 36 percent, treating AI more as an experiment than an established tool. The caution tracks with a wider {{link}} that separates strategy from the production floor. Brands carry more legal exposure and a larger reputation at stake, so they pilot where production companies simply ship.

The pattern also reflects risk tolerance more than enthusiasm. A production company ships AI-assisted work because its deliverable is the asset itself, and a rejected cut is just another revision. A brand hands the same tool to a legal and brand-safety review because its deliverable is reputation, and a misstep travels further than the campaign does. Adoption follows the cost of being wrong.

The caution tracks with a wider CMO AI adoption gap that separates strategy from the production floor.

The US and UK Are Optimizing for Different Things

The same technology is being pulled in two different directions across the Atlantic. US marketing teams prioritize quality and personalization, using AI to make each cut feel more crafted and more relevant to a specific audience segment. UK teams lean toward speed to market and creative volume, treating AI as a way to push more versions through the pipeline faster than before.

Both markets now treat AI as core infrastructure, but they are configuring it around different incentives. The split shows up clearly in how teams approach {{link}} when they hand AI a brief. A US team asks for relevance and craft; a UK team asks for throughput and reach. The tooling bends to match the question, which is why a single global AI policy rarely fits every office.

Neither priority is wrong, but they change what good means for the same model. A US team will trade speed for a frame that feels hand-made; a UK team will trade polish for the extra reach a higher version count buys. The platforms are learning to optimize for whichever signal the team weights first, which makes the initial brief the single most important input of the whole run.

The split shows up clearly in how teams approach AI video personalization at scale when they hand AI a brief.

What This Means for Production Teams

For teams building an AI practice, the adoption map is a planning tool rather than a trophy. The task-level data says to start where AI already clears the quality bar: compositing, versioning, and creative testing, all areas where a human reviewer stays in the loop and the cost of a bad output is low. Those are the fastest places to show a win without reorganizing the whole department.

The seat-level data says to expect friction the moment AI moves from the production floor up into brand and media planning, where governance and legal review slow everything down. The market data says to name the objective before the tool: are you optimizing for craft or for volume, because the two demand different defaults and different review gates. Teams that skip that step end up with expensive assistants and no measurable change in output.

The practical move is to instrument adoption instead of celebrating it. Pick one task where AI already clears the quality bar, set a before-and-after measure for time and cost, and report the delta to the people who own the budget. Teams that can show a number rarely lose the argument about the next tool, while teams that report they are using AI now stall at the same approval every quarter.

The Risk of Treating Adoption as the Goal

The danger in a year when 88 percent of teams use AI is confusing adoption with capability. Epsilon's 2026 benchmark study of more than 250 marketing decision makers finds 100 percent of respondents are now using AI, yet only 9 percent say it drives revenue and 46 percent still measure AI by revenue outcomes rather than activity.

Production companies can ship daily because they tied AI to a concrete job: finish the cut, run the test, resize the asset. Brands that lag are often the ones still measuring AI by the number of assets produced instead of the campaign result it changed. Adoption is the easy part; wiring it to a business outcome is the work that separates the teams that lead from the ones that merely keep up.

The lesson from the two data sets is the same from opposite ends. XR shows adoption is nearly universal; Epsilon shows proof of impact is rare. The teams that close that distance treat AI as a job description, not a capability badge, and they review the output against the campaign result rather than the activity log.