Google's talk 'Mastering the messy middle', given by a Google marketing research and insights manager, presented Google's research on how AI is changing the way consumers make purchase decisions.
Speaker Pablo PérezEvidence slide photo, transcript
Day 3 · Friday 2 October 2026 · 15:30
Mastering the messy middle
Named on the title slide. How AI changes consumer decision making between trigger and purchase; the recording starts mid-talk.
Shown on screen 6
Consistent with docsD3-C577
Google's slide defined the brand heuristic as a cognitive shortcut in which consumers use a brand's name and reputation as a proxy for product quality, reliability and value.
“A cognitive shortcut where consumers use a brand's name and reputation as a proxy for product quality, reliability, and value”
Wording checked against the slide or recording
Speaker Pablo PérezEvidence slide photo
Used byglossary term Brand heuristic
Not in docsD3-C581
A Google slide showed nine Google Trends charts of trust-checking searches in several markets and languages (source line: Google Trends, January 2016 to March 2026, web search), all rising toward the end of the range: 'is [brand] legit?' (two panels), '[brand] è affidabile?', '[brand] jest bezpieczne?', 'Is [brand] betrouwbaar?', 'Ist [brand] seriös?', '[brand] est fiable?', '¿es [brand] falso?' and 'trustpilot'.
Speaker Pablo PérezEvidence slide photo
Things
Consistent with docsD3-C586
Google's marketing research talk reached the same advice as Search: optimise for people to win in generative AI search.
“Optimise for people to win in generative AI search”
Wording checked against the slide or recording
Speaker Pablo PérezEvidence slide photo, transcript
- Optimizing your website for generative AI features on Google Search
- Creating helpful, reliable, people-first content
- Repeats D1-C058 Day 1: Optimising for people is optimising for generative AI Search.
Confirmed by docsD3-C588
Google's slide advised producing unique, helpful, human-centric content.
Speaker Pablo PérezEvidence slide photo
- Optimizing your website for generative AI features on Google Search
- Creating helpful, reliable, people-first content
- Repeats D1-C055 Day 1: Myth: build content for every possible consumer need. Google's answer: prioritise unique perspectives…
Not in docsD3-C590
Google's slide advised partnering with category opinion formers, such as YouTube creators, experts and institutions, who can give consumers extra reassurance.
Speaker Pablo PérezEvidence slide photo, transcript
Things
Said on stage 32
Google has used the 'messy middle' decision-making model since 2019; the messy middle is what happens between the trigger that starts a purchase journey and the purchase itself.
Speaker Pablo PérezEvidence transcript
Used byglossary term Messy middle
In the messy middle people are in one of two mindsets: exploring, which looks across the category, its price points, its brands and what is available, or evaluating, which works toward a position where a decision can be made.
Speaker Pablo PérezEvidence transcript
Used byglossary term Messy middle
Shoppers loop back and forth between exploring and evaluating until they feel confident enough to commit to a purchase.
Speaker Pablo PérezEvidence transcript
Google's researcher explained the looping in the messy middle as a tension between people's urge to collect more information and the mental effort of processing it, a subject studied by behavioural science.
Speaker Pablo PérezEvidence transcript
StageD3-C552
How AI is changing decision making is the number one question Google's consumer insights team receives, according to the speaker in October 2026.
Speaker Pablo PérezEvidence transcript
Google's research on how AI changes consumer purchase decisions, presented in October 2026, was based on 23,000 conversations with consumers (the talk did not say in which countries).
Speaker Pablo PérezEvidence transcript
Because people do not always do what they say they will do, Google's research on AI and purchase decisions also analysed panels with 40,000 participants, together with research partners, besides the consumer conversations (the talk did not make clear whether 40,000 is one panel or the total).
Speaker Pablo PérezEvidence transcript
Google's headline finding was that people who use AI features such as AI Mode or AI Overviews feel much more confident about their purchase decisions.
Speaker Pablo PérezEvidence transcript
Things
Behavioural science distinguishes a fast, emotional way of deciding, used for frequent low-stakes decisions, from a slow, effortful way used for high-stakes decisions, which is tiring.
Speaker Pablo PérezEvidence transcript
Google's research found that consumers delegate the mental effort of a purchase decision to AI without surrendering the choice itself, which Google calls being 'boosted'.
“that doesn't mean they are surrendering their choices. What's happening is that consumers are feeling boosted.”
Speaker Pablo PérezEvidence transcript
Used byglossary term Boosted consumers
According to Google's research, AI turns consumers who were confused and overwhelmed by information and options into consumers who feel empowered, like category experts, and able to make the best purchase for their needs.
Speaker Pablo PérezEvidence transcript
The first benefit consumers get from AI, according to Google's research, is ease: AI synthesises information quickly into an easy-to-digest form and makes choice overload manageable, so consumers can offload cognitive effort.
Speaker Pablo PérezEvidence transcript
Used byglossary term Boosted consumers
Google's research also counted easy access as part of AI's ease benefit: multimodal input, such as taking pictures, and low friction lower the barrier to entry.
Speaker Pablo PérezEvidence transcript
- Extends D1-C018 Day 1: One in six AI Mode searches is multimodal, using voice or images.
The second benefit, assistance, is that AI reduces a complex category, such as smartphones or trip planning, to the few features that really matter (for a phone, the battery and the picture quality), so consumers feel like experts.
Speaker Pablo PérezEvidence transcript
Used byglossary term Boosted consumers
Assistance also includes showing consumers what people in similar situations chose, because people are social and want that reassurance (the wording of this passage is partly uncertain in the recording).
Speaker Pablo PérezEvidence transcript
Google's research found that consumers who do not use AI pinball between exploration and evaluation, while AI users start a purchase journey mostly exploring and then move on to evaluating, a path that looks more like a funnel.
Speaker Pablo PérezEvidence transcript
StageD3-C567
The speaker called the marketing funnel a useful common language for strategy but not the reality of how consumers decide.
Speaker Pablo PérezEvidence transcript
The third benefit, suggestion, is that AI often proposes ideas, brands and products a consumer had not thought of, which lowers the cost of curiosity.
Speaker Pablo PérezEvidence transcript
Used byglossary term Boosted consumers
Google's research found that people who use AI run more searches than people who do not.
Speaker Pablo PérezEvidence transcript
- Extends D1-C004 Day 1: Google says users like AI Overviews and search more because of them, with early feedback overwhelmingly…
Google's research found that people who use AI also visit more brand websites, because it is simply easier.
“the people that are using AI, they search more, and they also visit more websites of brands.”
Speaker Pablo PérezEvidence transcript
- Extends D1-C005 Day 1: Google says AI Overviews help with new types of questions and lead users to visit a greater diversity of…
AI users sometimes take even longer purchase journeys, but they perceive their journeys as shorter, a perception the data does not always support (part of this passage is uncertain in the recording).
Speaker Pablo PérezEvidence transcript
For high-risk or high-involvement purchases, ease, assistance and suggestion are often not enough, and consumers add a reassurance step to reduce the risk of the decision.
Speaker Pablo PérezEvidence transcript
After exploring with AI, people often go back to Google Search and to the websites of the product or service providers to double-check before buying, for example trust in the company and cancellation policies.
Speaker Pablo PérezEvidence transcript
In the reassurance step the brand heuristic is especially important: a brand familiar to the consumer conveys a lot of information and confidence.
Speaker Pablo PérezEvidence transcript, slide photo
Used byglossary term Brand heuristic
A new or less familiar brand should expect people to double-check it and should prepare assets that give them the extra reassurance they need at that last step.
Speaker Pablo PérezEvidence transcript
Google Trends shows a rise since about 2020 in searches of the form 'is [brand] legit', in the UK and in most other countries Google looked at.
Speaker Pablo PérezEvidence transcript, slide photo
Things
StageD3-C583
The speaker read the rise in 'is [brand] legit' searches as a signal that people go back to the brand's website and to Google Search to take an extra reassurance step.
Speaker Pablo PérezEvidence transcript
Consumers who use AI no longer feel frustrated by a lack of progress in a purchase decision; they feel they are moving through the funnel and feel like category experts.
Speaker Pablo PérezEvidence transcript
Consumers boosted by AI in their purchase decisions were still a small group in October 2026, but the group is growing as more people start using AI (the word 'boosted' is an uncertain reading at this point of the recording).
Speaker Pablo PérezEvidence transcript
StageD3-C587
The reason given for optimising for people was that a human still makes the final purchase decision, so content has to be useful to that person.
Speaker Pablo PérezEvidence transcript
Because purchase journeys can be long and competitive, brands should be present where decisions are made: on Google Search and on their own brand website.
Speaker Pablo PérezEvidence transcript
StageD3-C592
A brand's own website matters in AI-assisted journeys because it is the one source where the brand speaks in its own voice and controls the message.
Speaker Pablo PérezEvidence transcript
What Google's documentation says 8
Google's Think with Google report 'Marketing in the messy middle' places the messy middle between trigger and purchase and models it as two looping mindsets: exploration, an expansive mindset that adds brands, products and category information, and evaluation, a reductive mindset that narrows the options down to a decision.
Publisher Think with Google (Google and The Behavioural Architects; Decoding Decisions, part 2)
Used byglossary term Messy middle
Google's messy-middle report cites its original Decoding Decisions study, run with The Behavioural Architects in the UK in 2019 with 31,000 in-market online shoppers and published in 2020; the report says the findings were validated in more than 20 countries.
Publisher Think with Google (Google and The Behavioural Architects; Decoding Decisions, part 2)
Google's messy-middle report names six behavioural principles at work in the messy middle: category heuristics, power of now, social proof, scarcity bias, authority bias and power of free; category heuristics are rules of thumb for a quick, satisfactory decision within a category, and social proof is the tendency to follow others' opinions and behaviour.
Publisher Think with Google (Google and The Behavioural Architects; Decoding Decisions, part 2)
Google's AI Mode launch post (March 2025) said that with Google's new AI features people are using Google Search more than ever, as they get help with new and more complex questions.
Publisher Google blog (5 March 2025)
Things
Google's AI features guide says that with AI Overviews people have been visiting a greater diversity of websites for help.
Publisher Google Search Central
Things
Google's helpful-content guidance asks whether content presents information in a way that makes you want to trust it, such as clear sourcing, evidence of expertise and background about the author or the site, for example through an author page or the site's About page.
Publisher Google Search Central
Google's messy-middle report says leading brands keep a meaningful share of consumer preference even against competitors with much better offers, and calls brand one of the strongest heuristics in consumer choice.
Publisher Think with Google (Google and The Behavioural Architects; Decoding Decisions, part 2)
Google's guide for generative AI features says creating content that people find unique, compelling and useful will likely influence a website's presence in generative AI search.
Publisher Google Search Central
What the press reported 1
The DMEXCO 2026 programme described a Google session, 'Beyond the Messy Middle: Mastering the consumer journey with AI' (24 September 2026), as based on tens of thousands of AI-moderated interviews and observed journeys across Europe, finding that consumers are being 'boosted' by AI.
Reported by DMEXCO 2026 event programme
Analysis by the author 6
AnalysisD3-C556
Keep the research numbers apart when quoting them: the 23,000 consumer conversations and 40,000-participant panels behind the 2026 AI findings are a newer research wave than the 2019 messy-middle study of 31,000 UK online shoppers.
Author Ibrahim Anjro
AnalysisD3-C593
For high-involvement products and services, make the double-check easy on your own site: company details, reviews, cancellation and return policies, guarantees and contact options should be on crawlable, indexable pages linked from product and service pages.
Author Ibrahim Anjro
AnalysisD3-C594
Track trust queries about your brand, such as '[brand] legit', '[brand] reviews' or '[brand] seriös', in Search Console and Google Trends for each market, and answer them on the site, for example with a trust or about page, instead of leaving the answer to third-party review sites.
Author Ibrahim Anjro
AnalysisD3-C595
The messy-middle findings come from Google's marketing research on consumer behaviour, not from Search documentation; they describe how people decide, not how Google ranks pages, and none of them is a ranking signal.
Author Ibrahim Anjro
AnalysisD3-C596
Google's finding that AI users visit more brand websites is Google's own research; check it against your own data, such as branded and direct traffic in analytics and the Generative AI performance report in Search Console, before planning on it.
Author Ibrahim Anjro
AnalysisD3-C597
Content that helps AI-assisted shoppers names the few features that decide a category (for a phone, battery and camera), compares options on them and shows what similar buyers chose, which matches the ease, assistance and suggestion benefits Google described.
Author Ibrahim Anjro
Links to other sessions 5
- Stage D3-C562 Day 3 · Mastering the messy middle
Google's research also counted easy access as part of AI's ease benefit: multimodal input, such as taking pictures, and low friction lower the barrier to entry.
extendsD1-C018 Day 1 · What's new in the world of SearchOne in six AI Mode searches is multimodal, using voice or images.
- Stage D3-C569 Day 3 · Mastering the messy middle
Google's research found that people who use AI run more searches than people who do not.
extendsD1-C004 Day 1 · Welcome and opening keynotesGoogle says users like AI Overviews and search more because of them, with early feedback overwhelmingly positive and satisfaction highest among 18–24 year olds.
- Stage D3-C571 Day 3 · Mastering the messy middle
Google's research found that people who use AI also visit more brand websites, because it is simply easier.
extendsD1-C005 Day 1 · Welcome and opening keynotesGoogle says AI Overviews help with new types of questions and lead users to visit a greater diversity of sites, with more total sites shown on the results page.
- D3-C586 Day 3 · Mastering the messy middle
Google's marketing research talk reached the same advice as Search: optimise for people to win in generative AI search.
repeatsD1-C058 Day 1 · How Search works and where's AI?Optimising for people is optimising for generative AI Search.
- D3-C588 Day 3 · Mastering the messy middle
Google's slide advised producing unique, helpful, human-centric content.
repeatsD1-C055 Day 1 · How Search works and where's AI?Myth: build content for every possible consumer need. Google's answer: prioritise unique perspectives, expertise and in-depth experience that go beyond common knowledge.