Before your team writes the 2027 plan, cut five habits. Stop counting engagement as proof the brand is growing. Stop approving AI creative because it looks fine. Stop paying for AI tools that cannot show a return, stop assuming a good reputation carries into AI answers, and stop judging launches only after they end. Research Kantar and Forrester published in the last few weeks, together with an Ipsos creative study from May, puts a number on each habit, and most of those numbers point the wrong way for companies that keep them.
The list matters more than any new tactic. A marketing plan is mostly inherited. Last year’s programs roll forward because nobody asked whether they still work, and the budget for anything new comes out of what you are willing to drop.
Stop counting engagement as proof the brand is growing
Creator and social budgets are rising fast. In its [Media Reactions 2026 study]( https://www.kantar.com/press-center/net-75-percent-of-marketers-plan-to-increase-investment-in-ai-visibility), Kantar found that two thirds of the more than 800 senior marketers it surveyed plan to increase investment in creator content, and 66% plan to invest in micro-creators.
Here is the problem. Kantar’s [review of its Kantar Forward 2026 broadcast]( https://www.kantar.com/inspiration/brands/kantar-forward-shaping-brand-growth), published on November 3, reports that only 6% of creator content works for both the platform and the brand. Kantar’s Creator Game Plan research, cited in the Media Reactions release, found that platform engagement and brand-building potential line up only a third of the time. Likes and saves tell you the algorithm approved. They do not tell you a buyer will remember who paid for the post.
Kraft Heinz is one company doing something about it. According to Kantar, Will Cooke’s team used Kantar’s LINK AI to analyse 18 months of the company’s social content and build a baseline for what contributed to brand equity over the long run, and that baseline is already shaping its briefs and 2027 plans.
Kantar’s lesson is to give creators room for their own voice while tying the work to a clear brand truth. We would put it more bluntly. If a creator could swap your logo for a competitor’s and the post would still work, you paid for the creator’s audience and got nothing for your brand.
What to do this quarter: pull last year’s ten best performing posts and ask one question of each. Did it say something only your company could say? Keep funding the ones that did.
Stop approving AI creative because it looks good enough
This is the habit with the least visible cost. In a study [published by Ipsos]( https://www.ipsos.com/en-us/ai-ads-are-good-enough-and-thats-problem), researchers from Syracuse University’s Newhouse School and Ipsos took 20 ads from major brands. Ten were the original human-made spots. For the other ten, Gemini turned each original into a brief and shot list, and OpenAI’s Sora 2 produced a new version with no human creative input. Three thousand US consumers then watched them in a cluttered setting meant to feel like real life.
Only 25% of people who saw an AI ad were at least somewhat confident it was AI-made. Yet the human ads averaged 14% stronger on short-term sales impact and 17% stronger on long-term brand health, by Ipsos’s measures. The researchers found one reason in the structure of the work: Ipsos’s wider research shows 49% of ads use storytelling, but the AI chose to tell a story only 30% of the time.
Adam Peruta, a Syracuse professor and one of the study’s authors, summed up the risk in [Harvard Business Review]( https://hbr.org/2026/09/research-ai-generated-ads-perform-worse-than-human-made-ones-even-when-customers-cant-tell-them-apart): the danger “is not obviously bad creative but work that looks good enough to approve.” He argues leaders should treat AI creative as a judgment and governance issue.
The study has limits, and they are worth knowing. The AI ads were made without any human direction, which few serious teams would do. And the strongest AI ads were plain, product-led spots in formats viewers already knew, such as those for Febreze and Herbal Essences. AI can do that kind of work well.
Where it fell short was story. That is exactly the part of a brand that buyers remember. Ipsos put it in one line: “Credible is not the same as compelling.”
What to do: add one question to every creative review. Does this piece tell a story, with a person, a tension and a change? If the answer is no, it can still be fine for a product page. It should not be the film that carries your brand for a year.
Stop paying for AI that cannot show its return
The first wave of AI tools was bought on enthusiasm. Forrester expects 2027 to change that. In its [predictions for B2C marketing]( https://www.forrester.com/blogs/predictions-2027-cmos-reckon-with-an-ai-reality-check/), published October 21, the firm predicts marketers will scale back a third of their AI initiatives because usage-based pricing now puts a cost on every prompt and every agent task.
Its Q3 2026 CMO Pulse Survey shows the pressure is already here. Eighty-two percent of US B2C marketing executives expect to give stronger justification for future AI spending, and 78% say leadership demands stronger evidence of value than it did a year ago. The same survey found 64% believe their own AI use is likely to cause a brand-damaging incident within twelve months.
Forrester also expects AI agents to replace legacy marketing and insights tools at one in ten companies. Its point is that the winners will redesign how work gets done rather than swap one tool for another.
Few teams are set up to answer that pressure. [Marketing Dive reported]( https://marketingdive.com/news/the-changing-cmo-role-in-3-numbers/873219) on October 30 that only 11% of CMOs in the Lippincott and Bloomberg Media CMO Outlook 2026 rate their company’s ability to adopt new marketing technology as excellent. The same roundup cites Findem and CMO Huddles data showing 26% of newly hired CMOs have AI-related skills, against 9% of CMOs who have been in the job four years or more. Companies are already hiring for it.
AI tools will not disappear from marketing teams. The ones that survive will be tied to a decision someone makes faster or better. Write down every AI subscription your team pays for, its monthly cost and the last decision it improved. Anything with an empty third column goes.
Stop assuming a strong brand shows up in AI answers
This one surprises leaders of well-known companies. Kantar reports that 82% of people it surveyed across 12 markets used an AI assistant in the past six months, and it says plainly that a strong brand does not automatically become visible in AI answers. Its test has three parts: is your brand present, is it described accurately, and can a reader see what makes it different?
There is a second half to the evidence, and it argues for patience with your media budget. Kantar’s Media Reactions study found only 32% of consumers use AI assistants to research brands and products, and only 23% trust their recommendations. Meanwhile a net 75% of marketers plan to increase investment in AI assistants in 2027. “There’s a lot of money moving towards AI on the assumption people will use it the way the industry expects,” said Gonca Bubani, Kantar’s global director of media.
So stop treating AI visibility as a media line, and start treating it as a description problem. AI assistants repeat what the open web says about you. If your website is vague about what you do and what you have delivered, the answer will be vague too. Fixing that costs far less than buying placements, and it helps the human buyers who visit as well.
Stop judging a launch after it is already over
Most companies measure a launch with a recap deck, weeks after the money is spent. Kantar’s ConceptEvaluate data shows two in three innovations fail to build brand meaning. At Kantar Forward, the company unveiled a tool that reads live digital signals in the first weeks of a launch and recommends changes while there is still time to make them.
You do not need Kantar’s product to adopt the principle, and it is fair to note that the tools in its review are Kantar’s own products. The idea underneath is sound. Kantar frames it as two questions: what will you know in the first few weeks, and who can act on it?
Before your next launch, pick two early signals you will check in week two, such as branded search volume or the share of sales calls that mention the campaign. Name the one person allowed to change the creative based on them. Then give that person a budget held back for the change.
What to keep, and what to start, in 2027
None of this argues for doing less marketing. It argues for spending on the parts machines and algorithms cannot fake.
Keep the story. The Ipsos study shows story is where human work still beats generated work, and Kantar’s creator data shows content without a clear brand truth rarely builds anything. Keep documenting the real work your company does, because a true account of how you work is the one asset a competitor’s AI cannot copy, and keep paying for the craft that makes a growing company look as serious as it is.
Start a stop list. Our prediction, and we label it as one: by 2028 the strongest marketing teams will review what to cut as often as what to add, and the habits above will look as dated as buying fans on social media does now. The companies that get there first will have more money for the work that builds a brand.
- Book a 90-minute session with your marketing lead before the 2027 budget is final.
- Go through the five habits above and mark each one as keep or stop.
- Move the savings into one story-led project, such as a brand film or a documentary series about your real work, with a clear owner and a launch signal you will check in week two.
- Repeat the review every quarter.
