Google's 2026 holiday guide leads with 61% AI adoption, which means two in five shoppers used none

Google distributed its 2026 retail holiday guide to advertisers on 21 August 2026 under the email subject line "Lead the AI-driven holiday shift", and the document opens by welcoming readers to what it calls the first AI-powered holiday season. The headline adoption figure inside it is 61%, the share of shoppers who used AI platforms or features during last year's holiday season according to a Google-commissioned Ipsos study. That figure is framed as proof of a completed shift. Read the other way round, it says that roughly two in five shoppers used no AI at all. PPC Land, which obtained the guide and worked through its reference list, reported on 21 August 2026 that the footnote trail behind several of the headline numbers does not always match the claims those numbers support. The full analysis is at PPC Land.

The document is titled "Supercharge your business this season: The retail marketers' guide to the AI-powered holiday season 2026". It is co-branded with YouTube and carries an August 2026 date on Think with Google. It runs across three chapters: consumer behaviour, a product checklist Google calls the Commerce 4, and three customer case studies. The framing claim about firsts sits awkwardly against Google's own trading history, because the company deployed agentic checkout and conversational shopping across Search and the Gemini app for the 2025 season and published a separate Holiday Essentials guide in November 2025.

The 61% figure and what it actually counts

The opening statistic of the guide is 61% of shoppers using AI platforms or features last holiday season. The footnote points to a Google-commissioned Ipsos Holiday Shopping study fielded between October 2025 and January 2026 among 8,237 United States consumers aged 18 and over who had conducted holiday shopping activities in the previous two days.

Every part of that base is load-bearing. The country is the United States, so the figure carries no reading for any other market. The screening condition is recent holiday shopping activity within a two-day window, so the sample is people who were actively shopping at the moment they were asked, rather than a general adult population. The field window spans four months across the peak season and its immediate aftermath. Any of those three conditions would move the number if it were changed, and none of them can be removed by rounding the figure or by quoting it on its own.

A second adoption-adjacent figure comes from somewhere else entirely. The guide records that 64% of surveyed holiday shoppers wished they had used AI more, drawn from an Ipsos Consumer Continuous survey run in January 2026 across sixteen markets with 2,241 respondents. The source note states that this sample was not weighted to reflect population size. A sixteen-market pooled sample that has not been weighted to population cannot be read as a global average, and PPC Land reported the caveat as it appears in the document rather than as an inference.

The number Google put at the top of its email is different again. According to the email, 94% of United States holiday shoppers who used AI view AI-powered tools, resources or technologies as valuable for their shopping. The base for that figure is adopters. It measures satisfaction among people who had already used the tools and says nothing about how many people that is. Placed next to 61%, the pair describes a market in which the users are largely content and about two in five shoppers are not users at all. Reporting 94% without its base is the single easiest way to turn this guide into a claim it does not make.

The journey figures, and the sentence underneath them

Two behavioural findings anchor the first chapter. Three in four users say AI Mode or AI Overviews help them make faster and more confident decisions, drawn from an Ipsos Global Consumer Journeys survey of 13,189 online shoppers across 25 countries in December 2025. Separately, shoppers who use AI platforms in their purchase journey interact with 2.8 times more touchpoints than those who do not, a comparison between 14,305 AI users and 21,881 non-users in the same December 2025 wave.

The chapter heading reads "AI doesn't cut the journey short. It expands the horizon." The guide is explicit that the friction being removed is not the number of touchpoints but what it calls the frustration of finding good answers, and that by the time a shopper reaches a retailer's site the decision has already been made. That last clause is the operative one for anyone running an ecommerce marketing programme, because it describes a shopper arriving already decided, having done the comparison work somewhere the retailer does not own.

Analysis, not a sourced claim: a 2.8 times touchpoint multiple is compatible with fewer visits to any individual retailer, if the additional touchpoints happen inside Google surfaces. The guide's own description of shoppers vetting and validating choices before they arrive points in that direction. PPC Land made the same observation, noting that documented click-through reductions on the publisher side of these surfaces and Google's touchpoint expansion claim can both hold at once.

Where the footnotes stop matching the claims

The guide states that AI Overviews now has over 2.5 billion monthly users and that AI Mode on Search has surpassed 1 billion monthly users. According to PPC Land, the 2.5 billion claim carries a footnote pointing to an Ipsos post-holiday survey of 1,275 United States consumers aged 18 to 64 who shopped online during the 2025 season. A consumer panel of 1,275 people cannot produce a platform-wide user count, and PPC Land reported that mismatch as a finding about the document rather than about the underlying number.

The adjacent AI Mode figure is footnoted to Google I/O 2026, which is where that number was disclosed. The same slippage appears one page later, where the 94% valuable-tools statistic is also footnoted to Google I/O 2026 rather than to a survey. PPC Land judged the most probable explanation to be an off-by-one error in the reference list rather than a substantive claim about sourcing, while noting that the published document is what advertisers receive.

Both scale figures are independently documented elsewhere. Google confirmed the one billion AI Mode monthly active user milestone on 19 May 2026, and the 2.5 billion AI Overviews figure appears in Google's own Search Console documentation for the control that lets site owners exit both surfaces. The numbers are not in dispute. The reference list is.

Etsy, and the word "partially"

Etsy is the headline case study. According to the guide, the marketplace used AI Max for Search campaigns to interpret shopper intent and automatically tailor ad headlines, pairing that with shoppable YouTube videos. During the 2025 holiday shopping season, Etsy drove a 36% year-over-year increase in gross merchandise sales, partially attributed to AI Max, using unbranded Search ads while maintaining strict return targets. The guide also states that the combination of Etsy's data setup and AI Max supported a 21% rise in new buyer acquisition.

The hedge in that sentence is Google's own. The guide says partially attributed, and PPC Land reported that none of the three case studies in the document carries a measurement window, a control condition or a statement of incrementality method. What is on the record, then, is a year-over-year comparison with a named contributing factor. What is not on the record is how much of the 36% the tool produced, what the counterfactual was, or over what period the comparison ran. Anyone repeating the figure as an AI Max result is adding a causal claim the source does not make.

The other two case studies follow the same pattern. The guide credits Wall Blush with a 98% year-over-year rise in conversions, a 13% increase in clickthrough rate and a 111% increase in revenue after using Demand Gen video and Performance Max product feeds. Nex Playground is described as tripling year-over-year sales through YouTube Connected TV and Demand Gen. Neither carries a measurement window or a control either.

The performance record on AI Max is more contested than the guide suggests

The guide's own performance claim is that advertisers adopting AI Max in Search or Performance Max see an average of 15% more conversions or conversion value on Search at a similar cost per acquisition or return on ad spend, provided they have enough budget to capture the additional value. PPC Land reported the base: the underlying data covers 27 February to 12 March 2026 and is restricted to non-retail advertisers with at least 30% of their Search text ad conversions coming through the full AI Max feature suite. A retail guide is therefore citing a measurement restricted to non-retail accounts.

Independent measurement has landed elsewhere. PPC Land cited a Smarter Ecommerce analysis of more than 250 retail campaigns in November 2025 that found conversions delivered at roughly 35% lower return on ad spend than traditional match types, and a Lunio report finding that retail AI Max search campaigns carry 72% more invalid traffic than the platform average, accounting for 68% of all invalid clicks in its dataset. Google's own headline claim for AI Max has also moved over time, from 14% at the May 2025 launch, to 7% for the full feature suite at the April 2026 Dynamic Search Ads retirement announcement, to 15% in this guide under a different definition.

The 37% figure Google led with in its email deserves the same treatment. According to the guide, marketing mix models run by TransUnion in the United States found that running Google Search and YouTube campaigns together drove 37% higher return on ad spend than all other media in aggregate. The source note describes a meta-analysis and synergy simulation across all verticals, based on 40 studies and more than 800 models covering the first quarter of 2024 through the fourth quarter of 2025. It is a simulation rather than a controlled experiment, and the comparison is against all other media pooled together rather than against a named alternative.

The timing matters for anyone managing Google Ads accounts. Campaigns running automatically created assets or the campaign-level broad match setting convert to AI Max for Search on 1 September 2026. The bidding target optimisation change began rolling out on 17 August 2026, moving budget-limited Target CPA and Target ROAS campaigns toward their stated targets across Search, Shopping, Performance Max, Demand Gen and Travel. The guide's instruction to use flexible budgets arrives in a quarter where the mechanics of budget-limited bidding have just changed.

Every headline number against its base

The table below sets each figure the guide leads with against what it measures and the sample behind it, exactly as PPC Land recorded the footnotes. Reading any row without its third column produces a different claim.

What this means for Thai marketers

This section is analysis rather than sourced fact, and the first point is a limitation. The holiday-specific research in the guide covers the United States. The journey and trust questions cover a global average of up to 25 markets. Thailand is not broken out anywhere in the document, and PPC Land noted that none of the consumer research was weighted to reflect population size. A Thai retailer reading 61% is reading a United States adoption rate collected from United States shoppers, and there is no defensible way to convert it into a local one.

That does not make the guide useless here. It relocates the useful part. The transferable question is not whether to believe a global adoption number but which feed and content work actually moves visibility on AI surfaces, because that work is testable in a Thai account this quarter regardless of what the survey says. Product data now determines placement across Shopping ads, AI Mode listings, AI Overviews and the Gemini app at the same time, which concentrates the return on getting it right. PPC Land cited Productrise research published in August 2026 finding that AI Mode surfaces 95% fewer Shopping listings than conventional results, which raises the value of appearing at all.

Three concrete checks follow from the calendar rather than from the survey. Confirm whether any live campaign uses automatically created assets or the campaign-level broad match setting, because those convert to AI Max for Search on 1 September 2026. Look at whether budget-limited Target CPA and Target ROAS campaigns changed spend behaviour after 17 August 2026. And audit feed attribute coverage against the conversational attributes Google introduced at Google Marketing Live on 20 May 2026, which pass customer questions and answers, variant options and use cases to Google's models. Teams building for answer engines can treat that same feed and content work as the practical core of generative engine optimisation.

Frequently asked questions

Does 61% mean most shoppers now use AI to shop?

It means 61% of a specific United States sample said they used AI platforms or features last holiday season, which leaves about two in five who did not. The sample was 8,237 consumers aged 18 and over, screened on having shopped for the holidays in the previous two days, fielded between October 2025 and January 2026. It is a majority, and it is a long way from universal.

Is any of this data broken out for Thailand?

No, and the source did not say anything about Thailand. The holiday questions were asked in the United States only, the journey and trust questions pooled up to 25 markets into a global average, and the guide's own source notes state the samples were not weighted to reflect population size. Any Thai read on these numbers is inference, not data.

Did AI Max cause Etsy's 36% increase in gross merchandise sales?

The guide does not say that, and this post will not say it either. It says the 36% year-over-year rise during the 2025 holiday shopping season was partially attributed to AI Max. PPC Land reported that the case study carries no measurement window, no control condition and no statement of incrementality method, so the size of the contribution is unknown.

Do I have to change anything in my account this week?

Only if the September deadline touches you. Campaigns using automatically created assets or the campaign-level broad match setting convert to AI Max for Search on 1 September 2026, so that is worth confirming now rather than in October. The bidding target optimisation change that began on 17 August 2026 applies to budget-limited Target CPA and Target ROAS campaigns and is already in motion.

Why does the guide's own reference list get flagged?

Because PPC Land found figures whose footnotes point somewhere the figure cannot have come from, including a 2.5 billion monthly user claim cited to a 1,275-person consumer panel. PPC Land's reading is that this is most probably an off-by-one error in the reference list rather than a claim about sourcing. The underlying scale numbers are documented elsewhere by Google.

If you sell online in Thailand and want a clear read on which of these products and feed changes are worth acting on before the fourth quarter, rather than a summary of a vendor guide, that is a conversation worth having now while the September deadline is still ahead of you.