The AI in Marketing Map

The AI in Marketing Map: October 2026

The AI developments that matter most to marketers, minus the hype. Always fresh, always practical, and always free.

The big shifts

The signals & stories

Announced At scale In use Not at scale yet Research/data

CloseThe big shifts · Early

Equity and legacy give brands a springboard

Searching for a brand by name now usually brings up Google's AI summary of it, and when Ahrefs tested 100 of the world's best-known brands, only 3.4% of the full summaries mentioned a competitor. For well-known brands, being searched for by name still keeps rivals out of the answer.

Elsewhere, synthetic (i.e. AI-generated) research respondents still over-index on established brands, and work made with AI tools continues to drift towards the average. This means that being well-known and being distinctive both matter more than ever.

Signals & stories

Synthetic respondents arrive, with a biasMar 2026

Why it matters for marketers

Synthetic panels are useful for early testing when they are built on your own data, and risky for decisions that depend on nuance, challenger brands, or minority views.

What happened

Qualtrics launched synthetic consumer panels in the US at its X4 conference in March, with other markets to follow. In July YouGov launched Parallax, which builds AI copies of members of its panel, known as digital twins, and checks their answers against live surveys. Bain reports that digital twins built on a client's own data reproduced about 90% of an earlier study's key findings.

A study published by NIM, the institute that founded GfK, compared 461 real US soft-drink buyers with digital twins built from their demographics. The twins overrated well-known brands most, and varied far less than real people.

Timeline

Announced: March 2026
At scale: Spreading
Jul 2026: YouGov launches AI twins checked against real panellists

Sources

AI-assisted work starts to sound alikeApr 2026

Why it matters for marketers

For brands the risk is a slow drift towards the category average.

What happened

A meta-analysis of 19 studies, posted in April and peer-reviewed in August, found that using AI tools makes people's ideas, designs, and writing more similar to one another. The effect is small, and the authors describe it as a gradual reorganisation of creative diversity.

In August, a study in Nature Human Behaviour of more than 880,000 texts found that rewriting them with language models cut the variation in how complex they were by between 21% and 50%. Earlier research found that Indian participants writing with AI suggestions drifted towards Western styles.

Timeline

Published: April 2026

Sources

AI agents become the main readersSep 2026

Why it matters for marketers

Shopping agents and AI assistants now read product data, pricing pages, and policies, and take every word literally. Track how often you're picked, as well as how often you're mentioned.

What happened

Mintlify hosts documentation for more than 20,000 companies. Its 2026 State of Knowledge report estimates that AI agents made 66% of the traffic to the documentation sites it hosts in August 2026, up from 20.8% in February. Mintlify counts agents and people in different ways and calls the comparison an estimate, and it sells documentation tooling.

A study of coding agents' tool choices by Armature found PayPal mentioned in 139 sessions and never picked, and LangChain mentioned 194 times and picked four times. Armature sells growth services to developer tool companies and the study covers developer tools, so it is an early signal.

AI agents' share of traffic to the documentation sites Mintlify hosts, February to August 2026

Chart: Mintlify, The 2026 State of Knowledge Report

Timeline

Published: September 2026

Sources

AI summaries are replacing the search clickMay 2025

Why it matters for marketers

Traffic built on answering generic questions is shrinking. Even a search for a brand's own name now usually starts with Google's summary of the brand.

What happened

Pew Research Center tracked almost 69,000 real Google searches in 2025. People clicked a traditional result 8% of the time when an AI Overview (Google's AI summary at the top of the results) appeared, against 15% when one did not. By May 2026, Similarweb's data showed AI Overviews on 43% of US Google searches, up from 15% a year earlier.

On 29 September 2026, Ahrefs found AI Overviews on 83% of US desktop Google results for brand names, up from a monthly average of 61% in July. When it searched for 100 of the world's best-known brands, only 3.4% of the summaries it could read in full (2 of 58) discussed a competitor. Ahrefs sells search analytics, and DemandSphere's tracking showed a similar jump.

Chart: Similarweb, via TechCrunch

Timeline

Announced: AI Overviews spread through 2025
At scale: January 2026Already at scale as 2026 began

Sources

What to do about it

Keep investing in brand equity and distinctive assets. Check what Google's AI summary says when someone searches for your brand, and check AI-assisted work against what makes the brand different as well as against the brief.

Try this week

Put ten recent pieces of AI-assisted copy next to two competitors' with the logos removed, and ask colleagues to pick out yours. If they struggle to pick yours out, your copy is drifting towards the category average.

What we're watching for

More studies testing whether synthetic research respondents favour established brandsWatching

Evidence on whether challenger brands lose ground in AI answersWatching

The AI in Marketing Map: October 2026

The big shifts

The signals & stories

Subscribe