Dialogue: Capturing New Consumer Touchpoints – How Brands Can Shift From Being Searched to Being Recommended

As consumers start asking AI assistants longer, more specific questions, the path to brand discovery is shifting. In the past, businesses fought for placement on search results pages and e-commerce shelves. Today, brands also need to earn a spot in AI-generated answers and build ongoing trust as consumers fact-check, compare options, and complete purchases. Faced with this new consumer entry point, how can brands move from being searched to being recommended, and how can they turn AI-driven exposure into actual business growth?

On September 17, 2026, at the first thematic forum of Accelerate 2026 China – a cross-border ecosystem acceleration event hosted by Pattern with Ebrun as the strategic partner – Steve He, Key New Account Manager at Google China, and Rex Yang (Rex), Vice President of China Operations at Pattern, sat down for a discussion around the topic of "capturing new consumer touchpoints and shifting from being searched to being recommended."

Below is the transcript of the conversation.

(Note: This is an initial edited transcript compiled by Ebrun based on selected highlights of the discussion, with changes made only to improve readability without altering the original meaning.)

01 Consumer Touchpoints Are Shifting to AI, and Users Are Comparing More Carefully Than Ever

Rex Yang: Twenty or thirty years ago, people discovered products mainly through outdoor ads, TV, newspapers, and print media. Then search engines came along – Google overseas, Baidu in China. After that, consumers started going directly to shopping platforms like Taobao, JD.com, and Amazon. Then Douyin in China and TikTok overseas brought social commerce into the mix. Starting around 2025, AI became the biggest topic of conversation. Even I now ask AI for recommendations before I buy things; I regularly use ChatGPT, Gemini, and Claude. Consumer behavior is shifting in subtle, natural ways – no one mandated the change, people just find it convenient to ask AI for advice.

From Google's perspective, what makes Gemini different from other products? And how is AI integrating into the shopping journey for both consumers and sellers?

Steve He: As Gemini continues to iterate on its core capabilities, Google's AI technology and commercial advertising solutions are already deeply deployed across a wide range of overseas expansion and marketing scenarios.

Take search ads in AI Mode, for example: when a consumer searches Google for a complex query like "how to make my home smell as relaxing as a high-end spa," the AI can provide detailed suggestions while naturally integrating an ad for a smart aromatherapy diffuser. The consumer might not have even thought about a diffuser initially, but the AI's answer and ad recommendation meet their needs and successfully generate purchase interest.

Then there are smarter shopping ads. AI can understand specific consumer descriptions of products and directly explain why a certain item fits their long-tail needs. For example, if a user is looking for "a coffee machine that doesn't take up space but can make both espresso and iced coffee," the ad won't just list dry specs – it will address their core need with a single, persuasive line.

Google is also continuously exploring more seamless shopping and conversion experiences, such as YouTube Shopping, more immersive Virtual Try-On, and price tracking and promotion alerts via Merchant Center, all of which help shorten the path from initial interest to purchase decision. For apparel and footwear sellers, Google AI's Virtual Try-On can reduce user decision time through immersive experiences. On the seller side, Google also offers AI Brief: sellers simply tell the AI their business goals, key messages, and target audiences in plain language, and it can set clearer directions and boundaries for Search and PMax campaigns, while also letting them preview AI-generated ad creatives and search examples. AI's impact is not limited to the consumer side – it is also starting to reshape how sellers run their businesses.

Rex Yang: In the past, consumers would search for products directly on Taobao, JD.com, Tmall, Amazon, or Walmart, or run searches on Google or Baidu. Now many people compare options in AI assistants first, forming a preliminary perception of a brand before they even land on Amazon or a brand's independent DTC site. Based on the global trends Google is observing, what key changes in consumer behavior are emerging amid the AI wave?

Steve He: Many cross-border brands are anxious: as the AI era arrives, do traditional performance ads still work? Will consumers even pay attention to them? Will they just let AI make all their shopping decisions for them? From the trends Google is seeing, there are changes, but there are also constants.

What has changed is that consumers, empowered by AI, have evolved into "super consumers" – their decision-making roles are more complex and less linear. The old consumer decision journey was very linear, with an average of just four touchpoints. For example, to buy noise-canceling headphones, someone might search Google, see an ad, click through to check the price and product details, and place an order if it looked good. Now, with AI, the shopping journey hasn't gotten shorter – instead, the number of decision touchpoints can jump to 12 or even more. Consumers will first give AI their personalized needs and ask it to generate a comparison list; they won't fully trust the recommendations they see, either, but will follow up with tricky, highly specific long-tail questions like "which headphones are more breathable to wear in summer" or "will they hurt if I wear them with glasses?"

After that, they might take the models AI recommended, exit the AI interface, and go to Google to verify specs, watch creator tests on YouTube, look for negative reviews on Reddit, check after-sales policies on the official website, and even only complete their purchase a few days later after being reminded by a retargeting ad. Consumers aren't just searching for two or three words anymore; industry trends show that as generative AI becomes mainstream, consumer search queries are evolving from short phrases into complex, multi-dimensional long natural language sentences.

What hasn't changed is trust. Third-party research shows most consumers still use AI as an initial information-gathering tool, and before making a final purchase, they tend to return to search engines and professional communities to cross-verify information. After all, consumers are the ones spending their own money – AI won't take full responsibility for its recommendations. AI improves the efficiency of information access, but it cannot replace transaction trust. Overseas consumers still rely on real reviews and feedback before making a purchase, and brand familiarity is also a key threshold for them to take a recommendation seriously.

So there are two key priorities for brands to break through: First, use AI-powered marketing tools (such as PMax) to precisely match consumers' long-tail intent. Faced with complex long-tail queries that average 8.5 words, manual keyword tagging is no longer fast enough; brands need to use semantic matching tools like AI Max to secure placement the moment AI gives an answer and consumers ask follow-up questions. Second, build trust barriers with authentic content. Brands need to lay out presence across creator ecosystems like YouTube for real, word-of-mouth reviews to capture high-value users who exit AI platforms to cross-verify information. You need to be mentioned by AI, but you also need to close the loop so users actually choose you.

Steve He: As consumers increasingly tend to ask AI first before making purchases on e-commerce platforms, what changes is Pattern seeing on the transaction side?

Rex Yang: To get straight to the point, we are definitely seeing clear changes. Users are taking longer and longer before they actually land on Amazon or a DTC site, because they are doing a huge amount of asking, comparing, and research upfront. They also visit professional review sites to check results and reviews, to further confirm whether the information AI gave them is reliable. After completing this round of comparison, when users do arrive on Amazon or an independent site, their purchase intent is basically already settled – they often already know exactly which brand they want to buy.

If a brand has laid the groundwork on AI assistants early on, three clear changes typically occur: First, conversion rates for brand terms and long-tail keywords go up. Second, the number of pages users browse before adding a product to their cart drops. Third, as AI exposure and positive reviews increase, there is a spillover effect on platforms like Amazon, leading to significantly higher sales.

02 GEO Is Not an Upgraded Version of SEO – Its Core Is Brand Consensus

Rex Yang: A lot of people used to do SEO, but starting last year, more and more people are talking about GEO, and quite a few friends have switched to focusing on GEO. Do you think GEO is an upgraded version of SEO, or is it something with a completely different underlying logic?

Steve He: GEO is absolutely not a simple upgrade to SEO – it represents a fundamental change in the underlying logic of how consumers see brands. Many sellers want AI to recommend them first, which is fair enough; but if they still treat GEO like a new version of ranking manipulation, buying backlinks, and trying to "hack" their way onto recommendation lists, they are taking too narrow an approach.

First, doing GEO doesn't automatically mean you'll get recommended – AI will only recommend you when your brand has already built broad consensus. Large language model recommendation mechanisms rely on authoritative, cross-web factual information and brand credibility. If there are no in-depth reviews, creator endorsements, or social buzz about your brand across the internet, the model won't have enough confidence to include your brand in its answers.

Second, consumers are increasingly using long, conversational sentences to search. If you rely solely on organic crawling for GEO, you might miss a huge volume of intent. User queries are becoming more like casual conversations, which manual keyword tagging and organic crawling can barely cover. When Google Search's AI Overviews generate conclusions, what actually gets recommended at the first touchpoint is keyword-free semantic matching and customized copy. For example, PMax can crawl your website listings and combine that information with consumer needs to deliver relevant answers.

Third, there are no more "see it and buy it" shortcuts. After getting AI recommendations, many consumers immediately exit to do in-depth research, and a large share of them still rely on real reviews as the basis for their purchase decision. If GEO puts your brand in front of consumers, but when they check on Google or YouTube they find nothing at all or only negative reviews, you have only stimulated demand – and the order might end up going to a competitor.

Brands absolutely should invest in GEO. On-site semantic tagging, Schema structured data, and clear product knowledge graphs are all foundational infrastructure that helps large models understand your brand. But real GEO is not an isolated project – it is a complete set of brand assets built across multiple channels: use tools like PMax at the front end to capture complex intent, use the YouTube creator ecosystem in the middle to provide authentic trust assets, and use structured data at the base layer so AI can smoothly crawl and understand your website. Only when these three pieces work together can you form a closed business loop from being recommended by AI to being chosen by the customer.

Steve He: From a brand operation perspective, how do you judge how well a brand is performing in AI search? What metrics deserve the most attention?

Rex Yang: I think there are two main dimensions to look at. The first is visibility: for example, if you search for your brand and related product keywords across Gemini, ChatGPT, or Claude, how many times does your brand appear out of 100 searches? We've used Pattern's GEO scorecard to demonstrate this: one brand had a visibility score of 64, placing fifth out of five brands. Even though appearing 64 times out of 100 searches doesn't sound too bad, compared to its competitors, it was still wasting a huge number of exposure opportunities.

The second dimension is user perception of the brand – the sentiment score. That same brand had a sentiment score of 82, which was better than its visibility score, reflecting the overall tone of reviews and discussions about the brand on popular social platforms.

Once you know these two scores, you need to analyze how to improve them. To boost visibility, you need to pay attention to content citations on mainstream authoritative websites, optimize the content format on your own site, use expressions that align closely with how users actually ask questions, and create content centered on the keywords you have a chance to win. To improve sentiment scores, you need to earn more positive reviews and reduce negative feedback on influential social channels, and collect comprehensive reviews from multiple channels instead of only looking at a single platform. Many of the brands we serve have already started using GEO scorecards for self-audits: how many times is their brand seen in AI assistants? And when it is seen, is the feedback positive or negative?

Rex Yang: Take projectors as an example: what signals affect how often a brand shows up in AI assistants, and how well it scores on sentiment?

Steve He: When faced with dozens of portable projectors that all meet basic parameter standards, AI might only pick two or three. It doesn't simply compare which one has higher lumens or who bought more backlinks – the underlying logic still comes down to trust-building and entity consensus. Large models primarily look at three dimensions:

First, whether there is a clear alignment between scenarios and entity: across the entire web corpus, does your brand repeatedly appear in contexts related to real camping use, outdoor nighttime projection, built-in battery life, and similar use cases?

Second, whether there is diverse, cross-verified information: self-promotion on your independent site carries limited weight – AI will also reference professional media, Reddit discussions, and real on-camera tests from YouTube creators.

Third, whether brand familiarity meets a minimum threshold: at the confidence threshold, systems naturally lean toward mature, low-risk options that are unlikely to be a bad recommendation.

Brands can't go into the back end of large language models and write code, but they can capture these signals at three levels. At the commercial level, you can use PMax's keyword-free semantic matching and dynamic text customization to embed your product as a sponsored recommendation in answers when consumers ask multi-dimensional long questions that trigger AI Overviews. At the awareness level, you need to lay out authentic review content: the visuals, subtitles, text, and comments on YouTube videos are not just for consumers – they also become key corpus for large models to judge whether a product actually delivers on its claims. At the technical level, you need to implement Schema structured markup, connect to the Google Merchant Center shopping graph, and clearly provide hard specs like lumens, battery capacity, and weight; your product pages can't just list cold parameters, either – they need to host scenario-based FAQs that directly respond to consumers' conversational queries.

03 Being Recommended by AI Is Only the Starting Point – Conversion Depends on Full-Funnel Support

Steve He: Platforms like Amazon and Walmart provide huge volumes of information like reviews and sales figures. Which brand signals deserve the most attention, and what metrics are common blind spots for brands?

Rex Yang: Platform data doesn't lie. Every transaction, click, review, and repeat purchase is a real signal of user behavior. But platform data signals need to be interpreted; what brands usually see is only the surface, without truly understanding what the data means underneath. When we talk to brand leaders, we've noticed a few common misconceptions.

First, many get excited when they see GMV growth, but deeper analysis might show that growth is primarily driven by ad spend. That approach might work for a short period, but it is unsustainable in the long run. Brands need to pay attention to Total Advertising Cost of Sale (TACoS) to judge whether their business growth is overly dependent on persistently high ad spending.

Second, many only care about the number of positive reviews and ignore negative ones. Negative reviews actually tell you where your product falls short, or which existing features you haven't clearly communicated in your marketing. Pattern compiles cross-web review data into reports that don't just summarize strengths – we also feed back the issues raised in negative reviews to brands, to help them improve products or identify new messaging opportunities.

The third misconception is thinking generative AI is too far removed from conversion. It doesn't give you the direct last-click results you see on Amazon or independent sites. Brands create a lot of content across AI assistants, forums, and professional media, and might not see immediate conversion, which leads them to think those upfront investments are useless. But we've observed that after brands secure early search, comparison, and exposure on AI assistants, conversions for brand terms and long-tail keywords rise, the number of pages users browse before adding items to cart drops, and overall sales get a boost.

Who wins and loses in the future for brands will depend largely on whether they can understand the language behind platform data. Only by connecting sales, reviews, content, and competitor signals can you clearly see why consumers buy, and truly understand your own position.

Steve He: Being recommended by AI is only the first step. Product discovery happens on AI interfaces, but transactions usually take place on e-commerce platforms and independent sites. How can brands actually connect the "AI discovery to platform transaction" pipeline?

Rex Yang: Some brands say they've done a lot of search optimization, have good visibility and sentiment scores, but don't see obvious conversion improvements. The problem usually lies in the second step: brands get traffic from AI assistants, but when consumers arrive on Amazon, Walmart, or TikTok with existing trust in the brand, they see listings that haven't been updated in ages, old images and titles, even counterfeit products or out-of-stock SKUs – the traffic isn't being properly captured.

To turn traffic into transactions, you have to shift your mindset from single-point optimization to closed-loop thinking. Your AI assistant presence, DTC website, and Amazon listings can't all operate in silos. Starting from the AI recommendation, all the way to how your brand is represented on DTC sites and different marketplaces, your listings, content, pricing, and inventory need to match the information presented in the AI interface. When users buy a good product and leave positive reviews, those reviews feed back into AI recommendations, forming a complete flywheel. If any link breaks, the flywheel stops working.

So after you get traffic from the AI funnel, you also need content tools to generate higher-converting images, titles, and selling point descriptions, while maintaining price competitiveness and stable inventory. Pattern's system is designed around this entire chain, connecting every link from AI visibility to final transaction as smoothly as possible.

Rex Yang: Beyond traditional rankings, how should brands ultimately measure the value of GEO?

Steve He: A lot of sellers ask this question, and the subtext is: "I spent money on GEO, how do I prove it wasn't a waste?" If you're still only looking at keyword rankings or ad click volumes, you're using an old map to navigate a new continent. The user journey in the AI era is a web-like path: AI sparks demand, users exit to verify information, then convert across all channels. You can't measure GEO by looking at single points – you need to look at the health of three dimensions.

The first report looks at the front end. Brands shouldn't only check if their product is mentioned, but also their first-recommendation rate tied to specific scenarios. For long-tail, complex queries like "high-brightness, drop-resistant projector good for camping," is your brand the AI's top recommended option? When AI makes recommendations, does it consistently highlight your core selling points, instead of just listing you as a filler alternative?

The second report looks at the middle layer: after users exit the AI interface, has your brand built cross-web verification buzz? You should check if there is growth in active searches like "[brand name] review" or "[brand name] real user experience" on Google, and whether views, likes, completion rates, and purchase-intent comments on YouTube review videos are rising. If AI mentions your brand but users don't buy or seek further verification, that recommendation is dead traffic.

The third report looks at the back-end pipeline: omnichannel conversion. You can't measure GEO's ROI in isolation – you need to look at whether it drives incremental growth across your entire business funnel and speeds up key decision points. For example, are conversions up? Can PMax capture users in the final stretch as they exit to verify information and get ready to buy? Is the consumer decision cycle shorter? Is brand premium higher? Is customer acquisition cost for branded terms lower?

04 Globalization Is Not About Stacking Channels – It's About Reusing Capabilities

Rex Yang: There are more and more channels, and brands are pushing forward with global expansion. At this stage, what foundational capabilities should they build?

Steve He: In the past, Chinese sellers easily fell into a trap: competing fiercely for top keywords in top categories on leading platforms, chasing hit products, undercutting on price, and buying traffic, which only led to rising customer acquisition costs and thinning margins. Now channels are more fragmented, markets are more globalized, and AI algorithms on every platform are upgrading – which actually opens up incremental markets in segmented scenarios.

In the past, someone buying a projector might only search for the word "projector"; now they ask conversational questions like "can it work for camping in minus-10-degree snow mountains?" or "how do I project in a small apartment without a screen?" The user demographics and needs behind independent sites, Amazon, Walmart, and regional platforms are all different: some users care about brand lifestyle, others prioritize cost-effectiveness and fulfillment efficiency. The emergence of platform AI shopping guides and smart tools also makes it easier to match these segmented needs.

What brands need to build most is not hiring more people for every separate channel, but three capabilities that can be reused across channels:

First, the ability to capture segmented intent. Can brands use AI to target long-tail scenarios, instead of only fighting for oversaturated, high-competition keywords? No matter what country a consumer is in, what language they speak, or what specific pain point they have, brands need to be able to put scenario-based solutions in front of them at the very first touchpoint.

Second, the ability to build global, reusable brand trust assets. Brands shouldn't create fragmented content for every channel – they need to build highly credible, multi-modal real-test assets across core ecosystems like YouTube, professional media, and real user communities. These assets can be reused long-term: they act as factual sources for AI engines to reference when recommending brands, and also serve as key endorsements when consumers move to Amazon, Walmart, or independent sites to make a purchase.

Third, a structured data foundation. Both AI conversation tools and e-commerce platforms need structured, standardized product data. Brands need to implement structured markup, knowledge graphs, clear product parameters, use case descriptions, and inventory signals so information can be efficiently extracted by AI across the web; at the same time, they can use first-party data from their independent sites to feed ad delivery algorithms, so smart bidding gets closer and closer to their target customers.

More channels and broader markets don't have to stretch brands thin – they can actually help companies break out of the trap of only competing for top keywords. The prerequisite is building these three core logic centers: intent capture, trust building, and data structuring.

Steve He: Chinese brands are often good at single-platform operations, but search, social media, marketplaces, and independent sites tend to operate as separate silos. Faced with the complexity of multi-platform management, how can they break through?

Rex Yang: More and more companies are building presence across multiple markets and platforms, which is an opportunity but also a challenge. The U.S. market alone has Amazon, Walmart, Target, Best Buy, TikTok, and more. Operating all these platforms at once while maintaining a unified brand image and telling a consistent brand story is no easy task.

We worked with one brand that used completely different creative assets across platforms – some had a years-old logo, others used the new version. Consumers saw one look on Amazon and a different one on Walmart, which made them doubt if it was even the same brand, eroding their trust. Every platform team worked hard, but without a unified core strategy, it was hard to get results.

The problem is not that there are too many platforms or too many regions – it's that brands are still using single-channel era thinking to run omnichannel, global operations. Brands can start with three steps:

First, build a centralized brand asset library, so all brand assets across channels come from the same master source. Take Pattern's PXM as an example: it can generate assets centrally and distribute them to different platforms, ensuring assets are consistent, up to date, and timely.

Second, build a data middle platform. In the past, data from Amazon, Walmart, and TikTok was siloed, and brands lacked a unified data source. By putting user behavior, clicks, conversions, procurement, and reviews into a central database, all decisions can be based on the same set of data, reducing fragmentation between channels.

Third, use AI tools as much as possible to boost productivity, handing repetitive work over to AI so people can focus on more creative, constructive, strategic work that requires thinking. Managing multiple platforms with a single system turns the complexity of multi-region, multi-platform operations into a scalable capability.

Take the mattress brand Leesa, which we work with, as an example: Pattern first used the GEO Scorecard to analyze how to improve the brand's credibility and recommendation rate in AI assistants. Once traffic came in, we used Content Brief to analyze its listings on sites like Amazon, breaking down images into categories like use case scenarios, size communication, feature communication, and brand messaging to identify missing high-conversion content and fill in those creative gaps. Then we unified and restructured images, titles, and bullet points, creating overall creatives that matched local consumer habits and aesthetics in terms of color, logo usage, and more, and rolled those assets out across 17 platforms. This let the brand connect AI visibility, content optimization, and multi-channel presence into a single system.

Steve He: GEO and platform ecosystems are constantly changing. What should brands plan ahead for right now? If a Chinese cross-border brand starts doing GEO today, what is the first step they should take?

Rex Yang: In the past, brands were used to letting consumers find them: people searched keywords and brand names on Google or Amazon. But now there are too many products and ad costs are very high, so it's getting harder and harder to be found. In the future, it may be AI that chooses for consumers. Consumers ask questions, compare, and evaluate options in AI, and before they even enter the purchase stage, they already have a fairly deep understanding of a brand. Brands need to work to be "AI-readable, AI-trusted, and AI-recommendable." First, you need visibility – if you can't be seen, you can't be trusted. Second, you need to build a multi-platform ecosystem, maintaining centralized management across more regions and platforms. Third, you need to build data-driven operational capabilities, not just looking at surface-level phenomena, but using data to find areas for improvement and growth.

Steve He: The most fundamental change in search over the next 1 to 3 years will be an acceleration away from the "people looking for goods" comparison model, into the era of intelligent agents where AI agents curate and filter options for users. In the past, when you searched a term on Google or Amazon, you'd see dozens or even hundreds of listings, and consumers would scroll and compare slowly. In the AI agent era, consumers only need to state their needs, and AI may curate just a tiny number of highly matched options.

That doesn't mean there will only be one winner in the market. As intelligent agents and multi-modal technology develop, consumer decision journeys are becoming more efficient, and platforms are setting higher and higher requirements for relevance and content quality. For brands, the earlier you build structured data and authentic brand buzz, the earlier you can win a first-mover advantage in the future intelligent search ecosystem.

For Chinese cross-border brands starting GEO today, the first step is to master two basic competencies, one "hard" and one "soft." The hard foundation is structuring your product data so AI can extract it at low cost. For independent sites, you can start by organizing your website and Google Merchant Center graphs, and improving Schema structured markup; at the same time, don't just stack parameter tables – build out conversational, scenario-based FAQs so your pages can directly answer the questions consumers ask AI.

The soft asset is building multi-modal reviews to give AI confidence in your brand. Brands need to accumulate real creator reviews and word-of-mouth corpus across core channels like YouTube, professional media, and real user communities, so that video footage, subtitles, and comments collectively prove your product's selling points.

To put it simply: structured data lets brands be understood, multi-modal real word-of-mouth lets brands be trusted, and when paired with commercial AI tools for precise placement, brands stand a chance of being selected for the scarce recommendation slots in AI results. GEO is not a trendy buzzword concept – in the end, it is rooted in understanding consumer needs, and the long-term building of brand assets and trust.

Rex Yang: Today's discussion covered why brands should invest in GEO, how to judge the real value of GEO, and actionable suggestions for making products "visible, trusted, and recommendable" in AI assistants. Next, the ten platforms present today will be holding one-on-one introductions in the adjacent hall. We welcome everyone to take the opportunity to learn more and build connections, to add new opportunities and ideas for your brands' future global expansion. Alright, that's all the time we have today. Thank you all for joining, thank you!

This article was first published on the official website of Ebrun.