AI in marketing has stopped being a pilot and become a budget line. Gartner’s 2026 CMO Spend Survey found that CMOs now put an average of 15.3% of their marketing budgets into AI initiatives, yet only 30% describe their AI readiness as mature. Money is moving faster than capability. At the same time, AI is changing how people find brands, with Google’s AI Mode passing a billion monthly users, and British consumers are telling pollsters they trust brands less when ads feel machine made.
This report pulls together the most credible 2025 and 2026 data on AI in marketing: what brands are spending, how companies actually use it, what agentic AI really means, how AI search is rewriting discovery, and what the wins and backlashes teach. We finish with a practical 90 day plan for CMOs. Every number is sourced, vendor research is flagged, and where the evidence is thin we say so.
The short version
- Spend is ahead of readiness. CMOs give 15.3% of budget to AI but only 30% say they are ready to scale it, and marketing budgets themselves are flat at 7.8% of revenue.
- Use is broad but shallow. Generative AI now touches about 22% of marketing activities in US firms, up from 7% in early 2024, and the reported gains are real but modest: around 14% on productivity and overhead.
- Discovery is moving into AI answers. Clicks fall when Google shows an AI summary, ChatGPT has passed 900 million weekly users, and 41.5% of US marketers now work on generative engine optimisation.
- Trust is the constraint. UK consumers want disclosure and dislike AI actors. Polished AI work is tolerated; cheap looking AI work damages brands.
- The winners use AI for speed and keep humans on the idea. Unilever cut product imagery costs with digital twins while Dove pledged never to replace real people with AI. That split is the playbook.
What is AI in marketing in 2026?
AI in marketing now covers three distinct layers, and most confusion in boardrooms comes from lumping them together.
1. Generative AI for content
Tools that write, design, edit, dub and resize. This is where adoption is highest. In The CMO Survey 2026 (308 US marketing leaders, January 2026), content creation was the top use of AI at 73.9% of companies, followed by content personalisation at 65.4%.
2. Predictive and platform AI for media
The machine learning inside Meta Advantage+, Google Performance Max and AI Max, and TikTok Smart+. Brands often use far more of this than they realise. Meta says its end to end AI ad products now run at an annual revenue rate of more than $75 billion (Q2 2026 earnings call).
3. Agentic AI for workflows
Software agents that plan and carry out multi step tasks, such as building a campaign, briefing variants and reporting results, with limited human input. This is the newest layer and the most oversold. We cover it in its own chapter below.
How much are brands spending on AI in marketing?
Gartner surveyed 401 CMOs across North America, the UK and Europe between January and March 2026. The headline is that AI now commands 15.3% of the average marketing budget, rising to 21.3% at organisations Gartner classes as AI ready. Seventy percent of CMOs said becoming an AI leader is a critical goal for 2026.
The catch is the size of the pot. Overall marketing budgets rose only from 7.7% to 7.8% of company revenue, and 56% of CMOs said they lack the budget to deliver their 2026 strategy. In other words, AI money is mostly being found by cutting something else.
Marketing budgets as a share of company revenue
Gartner’s 2025 survey shows where that money gets taken from. Some 22% of CMOs said generative AI had reduced their reliance on external agencies for creativity and strategy, and the most common return reported was time efficiency (49%), ahead of cost efficiency (40%). The BCG picture is similar: 71% of CMOs in its 2025 study planned to invest more than $10 million a year in generative AI over the next three years, up from 57% the year before (BCG, June 2025).
The wider business context matters too. In the UK, the Office for National Statistics estimates that about 35% of businesses with ten or more employees used AI in June 2026, up from about 12% in late 2023. Your finance director has almost certainly seen AI savings elsewhere in the company, and will expect marketing to show its own.
How do companies use AI in marketing?
The CMO Survey offers the cleanest view over time because it asks the same marketing leaders the same question every year. The mean share of marketing activities using generative AI has roughly tripled in two years.
Share of marketing activities using generative AI (US, mean)
Usage is wide but not deep. HubSpot’s 2026 State of Marketing (vendor, 1,500+ marketers) found that 86.4% of teams use AI in at least a few areas, yet only 26.5% say it has significantly increased productivity. Salesforce’s tenth State of Marketing (vendor, 4,450 marketers) puts general AI use at 75% but says only 13% of marketers have moved on to agents.
Which AI uses are most common?
Top uses of AI in marketing, share of US companies using AI
Is AI marketing effective?
The honest answer is: moderately, and mostly on efficiency. The same CMO Survey asked leaders what AI had delivered. The mean reported improvements in 2026 were 14.1% on sales productivity, 14.6% on marketing overhead costs and 10.8% on customer satisfaction. Those figures have grown every year since 2024, but they are self reported estimates rather than measured results.
Other studies are more sceptical. Capgemini’s November 2025 research (1,500 executives at large firms) found only 7% of marketers strongly agree AI has boosted marketing effectiveness, even though about seven in ten large organisations use generative AI in marketing. McKinsey’s State of AI 2026 found that 37% of respondents attribute any EBIT impact to AI, roughly unchanged on the year, though marketing and sales is where revenue gains are most often reported. Efficiency is proven. Effectiveness, the bit that builds brands, is still being argued.
What is agentic AI in marketing?
Agentic AI describes systems that are given a goal and then take a series of actions to reach it: pulling data, drafting assets, launching tests, reading results and adjusting, with a human approving at checkpoints rather than doing every step. In marketing, the early use cases are campaign operations, audience research, reporting and customer service.
The momentum is real. McKinsey says the share of large organisations scaling agents in at least one function rose from 27% to 40% in a year, and that agents are most often used in marketing and sales. Meta has said publicly that it wants businesses to give it an objective and let its systems handle the rest, and Google claims advertisers using AI Max for Search typically see 14% more conversions at a similar cost per acquisition (Google, vendor data).
But the failure rate will be high. Gartner predicts that over 40% of agentic AI projects will be cancelled by the end of 2027 because of cost, unclear value or weak risk controls, and estimates that only about 130 of the thousands of vendors claiming to sell agentic AI are the real thing.
“Many use cases positioned as agentic today don’t require agentic implementations.” Anushree Verma, Senior Director Analyst, Gartner
Our view: treat agents as a way to remove repetitive work from the media and reporting cycle, not as a creative department. Start where errors are cheap and reversible, and keep a person accountable for anything that goes in front of customers.
AI search is rewriting how customers find brands
The most immediate commercial effect of AI in marketing may not be on what brands make but on where customers look. Three numbers capture the shift.
First, ChatGPT’s weekly audience has grown at a pace no social platform matched.
ChatGPT weekly active users, as stated by OpenAI
Second, Google is folding AI into search itself. Sundar Pichai told investors in July 2026 that AI Mode had passed one billion monthly active users (Alphabet Q2 2026). In the UK, Ofcom estimated in December 2025 that about 30% of searches show AI Overviews, and that ChatGPT had 1.8 billion UK visits in the first eight months of 2025, against 368 million a year earlier.
Third, those AI answers take clicks away from websites.
Share of Google visits ending in a click
What this means for brands is that being the source an AI answer draws on is becoming as important as ranking. Large language models summarise what the web says about you, and they lean heavily on video, forums, reviews and expert content. We covered the YouTube side of this in our guide to YouTube SEO and AI answers. The practical moves are clear: publish genuinely useful, specific content under your own name; earn coverage and creator content that third parties publish; keep product facts consistent everywhere; and track your share of AI answers alongside share of search.
What do consumers think of AI ads?
Here the data is unusually consistent, and it should worry anyone tempted to cut corners. Consumers are not against AI. They are against feeling fooled and against content that looks cheap.
UK adults on AI generated actors in ads
Kantar’s 2025 Media Reactions study (21,300 consumers in 30 markets) found more than half of consumers feel uneasy about AI generated ads and over 60% worry generative AI could lead to fake or misleading advertising, even as over 70% of marketers embrace it for creative work. DoubleVerify’s 2026 research (vendor, reported by Advanced Television) adds a vital nuance: 49% of UK consumers view a brand negatively if its AI ads look low quality or uncanny, but only 29% react negatively to polished, professional AI ads.
So the problem is not AI. It is slop. The creative evidence points the same way. Kantar’s analysis of 356 AI generated ads found their spread of performance matched traditionally made ads, though branding tended to be slightly weaker (Research Live, December 2025). System1’s testing of 2026 Super Bowl ads found those with AI messaging averaged 2.1 stars against 2.7 for all Big Game ads (System1, vendor). Meanwhile a small System1 and Jellyfish study of 18 AI assisted ads scored them well above average (Marketing Week). Read together: AI is a production method, and the old rules of emotion, distinctive branding and a clear idea still decide whether an ad works.
Wins and backlashes: what the case studies teach
The clearest lessons come from brands that went public with AI. Most efficiency figures below are self reported by the brand, so treat them as claims rather than audited results.
| Brand | What they did | What happened |
|---|---|---|
| Unilever | Product digital twins with NVIDIA for TRESemmé, Dove and Vaseline imagery (March 2025) | Claims product imagery 2x faster and 50% cheaper; TRESemmé Thailand content costs down 87% (self reported). Source |
| Dove | Pledged never to use AI in place of real people in its ads (April 2024) | A brand position built on trust, from the same parent company using AI for products. Source |
| Klarna | Generative AI for images and copy (Q1 2024) | Claimed $6M in image production savings and 80% of copywriting done with AI; later said human customer service was the way forward. Source, source |
| Coca Cola | AI made “Holidays Are Coming” films (2024 and 2025) | Most discussed festive ad of 2025 but only 61% positive social sentiment, against 87% for John Lewis. Source, source |
| McDonald’s Netherlands | AI generated Christmas ad (December 2025) | Pulled within days after being called “AI slop” online. Source |
| Duolingo | “AI first” memo on replacing contractors (April 2025) | Backlash led the brand to wipe its TikTok and Instagram posts for a period. Source |
| L’Oréal | CREAITECH generative content platform with NVIDIA (June 2025) | Positioned as augmenting creativity; no savings figures published. Source |
Figures are as reported by the brand or the cited publisher. None are independently audited.
Three patterns stand out. AI wins quietly behind the scenes, in product imagery, resizing, localisation and versioning, where consumers never see a seam. AI loses loudly when it replaces the human, emotional centre of a brand, especially at Christmas, the one moment when British audiences expect craft and warmth. And the companies that announce AI as a cost cutting story to the public tend to pay for it in reputation.
“You can’t rely on AI to come up with the idea.” Geoff Baillie, Creative Director, Rethink, on the Heinz AI Ketchup campaign
That matches our own experience. The work that has driven results for brands we have worked with, from Formula E’s 7M+ views across YouTube, Instagram and TikTok to the Sony Alpha 7 V launch’s 17M+ views across socials (see our work), came from a strong idea and real people on camera. AI made parts of the process faster. It did not supply the reason anyone watched.
Where AI belongs in a brand marketing team
Pulling the evidence together, this is how we advise clients to allocate AI effort. It is Advertain guidance rather than survey data.
| Area | Use AI? | Why |
|---|---|---|
| Research, briefs and audience insight | Yes, heavily | Fast synthesis; a human checks sources and draws conclusions |
| Versioning, resizing, captions and dubbing | Yes, heavily | Invisible to the audience and saves real hours |
| Media buying and bidding | Yes, with guardrails | Platform AI is already doing it; set clear goals, exclusions and incrementality tests |
| Product imagery and visualisation | Yes, with brand checks | Unilever style digital twins work when products are rendered accurately |
| Hero brand films and people on screen | Sparingly, and disclose | Consumers punish fake people and uncanny work; trust is the asset |
| The core creative idea | Assist only | Evidence shows emotion and distinctive branding still decide effectiveness |
Advertain guidance, based on the research cited in this report.
A 90 day AI in marketing plan for CMOs
If you are under pressure to show progress on AI in marketing this quarter, this is the sequence we recommend.
Days 1 to 30: audit and guardrails
- List every place AI already touches your marketing, including platform AI in Meta, Google and TikTok campaigns.
- Write a one page AI policy covering disclosure, likeness and voice consent, data handling and who signs off. Align it with ASA guidance, which we summarised in our ASA influencer guidelines explainer.
- Pick two measurable efficiency targets, such as hours per asset or cost per localised version.
Days 31 to 60: pilot where errors are cheap
- Run AI on versioning, captions, dubbing and research synthesis first.
- Test one agentic workflow in reporting or campaign operations, with a human approval step.
- Start tracking how AI assistants describe your brand and category, alongside share of search.
Days 61 to 90: reinvest the savings in what AI cannot do
- Move the hours saved into stronger ideas, real talent and creator partnerships. Our creator marketing report covers where that money works hardest.
- Measure brand outcomes, not just output volume, using the methods in our brand awareness measurement report.
- Report back to the board on time saved, cost saved and brand health, so AI is judged on business results rather than activity.
For YouTube specifically, including what you must disclose and what can cost you monetisation, read our guide to using AI for YouTube. If you want help building content people choose to watch, talk to us.
Frequently asked questions
How is AI used in marketing today?
Mostly for content creation, personalisation and media optimisation. In The CMO Survey 2026, 73.9% of US companies using AI applied it to content creation and 65.4% to personalisation, while 41.5% were working on generative engine optimisation. Generative AI now touches about 22% of marketing activities on average.
What is agentic AI in marketing?
Agentic AI refers to systems that pursue a goal across several steps, such as building, launching and adjusting a campaign, with humans approving at checkpoints. McKinsey reports that agents are most often deployed in marketing and sales, but Gartner predicts more than 40% of agentic AI projects will be cancelled by the end of 2027.
Is AI replacing marketing jobs?
It is changing them faster than it is removing them. McKinsey’s 2026 survey found two thirds of organisations saw little or no AI related change in headcount over the past year, though 39% expect declines in the year ahead. Gartner found 22% of CMOs say generative AI has reduced their reliance on external agencies, so the pressure is falling first on routine production work.
Should brands disclose AI in ads?
UK consumers clearly want it. YouGov found 94% of Britons who have seen AI actors in ads think companies should disclose them. There is no blanket UK law requiring an AI label, but ads must not mislead, so disclosure is the safer default whenever AI creates realistic people, voices or scenes.
Does AI make ads less effective?
Not automatically. Kantar found AI generated ads perform across the same range as traditional ones, with slightly weaker branding on average. What hurts is low quality: DoubleVerify found 49% of UK consumers view a brand negatively when its AI ads look uncanny, against 29% for polished AI work.
