In August 2026, brand discovery no longer comes down to a Google results page. It is spread across AI answer engines, algorithmic search inside social platforms, private communities, marketplaces and creators. Traditional search remains important, but it is no longer the single entry point to the journey.
The most important change is therefore not the “death of search”. It is the gradual shift from a search web to a recommendation web: users sometimes express an intent, but increasingly it is algorithms, contextual assistants and peers that select which brands to show them, and when.
The KAIS Discovery Mix 2026 is an attention allocation model, not an audience measurement. No single study covers these five environments with a consistent methodology, so the weightings below are a strategic reading published by KAIS, deliberately rounded so as not to imply a precision the underlying data cannot support.
One gap deserves to be stated openly. GWI measures that 22.1% of internet users aged 16 and over use AI platforms to research brands. KAIS nonetheless weights AI engines at 35%. The reason: we weight influence on the decision, not reported usage. A generative answer does not sit alongside the other channels — it filters consideration upstream, shortening the shortlist before the remaining channels come into play.
The KAIS Discovery Mix 2026
The KAIS Discovery Mix 2026 distributes influence over brand discovery across five environments: AI engines (35%), Social Search (30%), Dark Social and communities (15%), Retail Media (10%) and creators/UGC (10%). It is an attention allocation model published by KAIS, designed to arbitrate a visibility budget — not an audience measurement.
| Discovery channel | KAIS weight (model) | Primary mechanism |
|---|---|---|
| AI engines, AEO & GEO | 35% | Direct answers, comparisons and recommendations from ChatGPT, Gemini, Perplexity, Google AI Overviews and embedded assistants. |
| Social Search & short-form video | 30% | Interest graphs, TikTok, Instagram Reels, YouTube Shorts and native search inside social platforms. |
| Dark Social & communities | 15% | LINE, WhatsApp, Discord, Telegram, Reddit, private groups and peer recommendations that resist attribution. |
| Retail / Commerce Media | 10% | Marketplaces and retail networks using transactional and first-party data close to the point of purchase. |
| Creators, micro-influence & UGC | 10% | Demonstrations, human reviews, native content and social proof produced by creators or customers. |
GWI data reported by DataReportal in the Digital 2026 Mid-Year Global Update confirms the fragmentation. Asked which sources they use to research brands, internet users aged 16 and over cite search engines at 45.8%, social networks at 41.3%, review sites at 35.4%, brand websites at 32.2%, mobile apps at 30.7% and AI platforms at 22.1%.
Two readings follow. First, no channel reaches a majority of internet users on its own: the total far exceeds 100% because people combine several. Second, AI platforms already sit above Q&A forums, brand blogs and specialist reviews — in three years, with no media budget allocated to them.
1. The decisive shift from SEO to GEO
Keyword queries are giving ground to complex conversational questions. Instead of typing “men’s trail running shoes”, a user might ask: “Which ethical brand makes trail shoes with enough cushioning to run on urban trails three times a week?”
A classic search engine ranks links. An answer engine can synthesise several sources, retain two or three brands, explain the differences and produce a shortlist directly. A brand that is not included in that synthesis can lose consideration before any click happens at all.
In February 2024, Gartner predicted a 25% drop in traditional search volume by 2026 driven by chatbots and virtual agents. The deadline has arrived, and the forecast deserves to be tested against reality rather than recycled: Google still holds more than 90% of search market share and absorbed the shock by folding AI Overviews directly into its results. Volume did not collapse — the value of the click moved. A forecast partly wrong on the numbers, but right on the direction.
The behavioural data, by contrast, is unambiguous. HubSpot’s State of AEO 2026, based on more than 4,000 marketers, reports that 58% say their business already optimises content for answer engines and that 44% have made a business purchase from a brand discovered through an answer engine. On the consumer side, McKinsey’s Q1 2026 ConsumerWise survey finds that among Americans who use AI search, 44% name it as their most preferred information source, ahead of search engines (31%), retailer and brand sites (9%) and review sites (6%).
In Europe the shift is further along still: in a McKinsey survey across France, Germany and the United Kingdom, 84% of respondents use AI tools in everyday life, 63% use them to compare brands, models, prices and reviews, and 55% to learn about a product category. The battle is therefore no longer only about ranking. It is about eligibility for recommendation.
SEO → AEO → GEO
- SEO: being findable in search engines.
- AEO: structuring an answer clear enough to be lifted.
- GEO: building the entity, authority and corroboration signals needed to be cited or recommended.
2. Social Search: the interest graph replaces the social graph
Social platforms no longer primarily show what followed accounts post. They build a feed around what the algorithm predicts will be relevant to the user. HubSpot describes this shift from social graph to interest graph explicitly.
That evolution turns TikTok, Instagram and YouTube into discovery engines. Sprout Social’s Q2 2025 Pulse Survey measures that close to one consumer in three now begins a search on a social platform rather than Google, and more than one in two among Gen Z.
DataReportal confirms the hierarchy: social networks are the second global channel for brand research (41.3%) behind search engines (45.8%). On awareness specifically, social ads rank third behind search engines and television — but they become the leading driver of brand awareness among 16–34s.
The most powerful format in this environment is short-form video. In HubSpot’s State of Marketing 2026 (1,500+ marketers), 48.6% of respondents place short-form video in their top three for ROI — the highest score of any format measured.
Visibility is no longer won only on a results page. It is also won inside an algorithmic feed.
3. Dark Social: trust moves towards peers
As AI-generated content and targeted advertising proliferate, part of the trust shifts to less visible spaces: WhatsApp, Discord, Telegram, private messages, professional communities, forums and specialist subreddits.
SparkToro demonstrated how hard these journeys are to attribute. In its 2023 experiment, run with Steve Lamar across 11 networks and 16 referral types, 100% of visits from TikTok, Slack, Discord, Mastodon and WhatsApp were classified as “direct” traffic, with no source information at all. Facebook Messenger lost 75% of its referrals, Instagram DMs 30%. The protocol is three years old and in-app browsers have only become more common since: nothing suggests the situation has improved.
For a brand the stake is twofold: being genuinely recommendable inside these communities, then understanding that this off-site trust can itself become a reputation signal usable by search engines and AI systems when it is public or indexable.
4. Retail Media: discovery moves closer to the purchase moment
Amazon, Walmart, Instacart and large retail networks no longer serve only to close a decision. They participate in discovery through their proprietary transactional data.
Here too, projection and reality need separating. McKinsey estimated in 2022 that Retail Media Networks could represent as much as $100 billion in US ad spend by 2026. The measured figures are more modest: $53.7 billion in US RMN revenue in 2024 according to IAB and PwC, with a forecast of roughly $71 billion for 2026 according to eMarketer. Globally, eMarketer puts the market at around $140 billion in 2024 and $165 billion in 2026, with 277 networks counted as of November 2025.
Growth is therefore real but less spectacular than the initial projections — and considerably more concentrated: Amazon and Walmart alone are expected to capture close to 89% of incremental spend in the US market in 2026.
But AI is already displacing this model: McKinsey describes a journey in which the user asks an assistant for three recommendations, picks one and buys without ever browsing the retailer’s category page. The new question becomes: is the product present in the systems that decide what deserves consideration?
5. Creators and UGC: in an AI-saturated web, authenticity becomes scarce
The easier content production becomes, the more authenticity turns into an asset. HubSpot’s Social Media Marketing Report 2026, based on more than 1,100 marketers, reports that 77% consider authenticity more decisive than production value. The same survey identifies the difficulty of integrating AI without sacrificing quality or trust as one of the leading reported challenges.
A raw demonstration, a customer testimonial, an employee video or a specialist creator can therefore carry more weight than a perfectly produced advertisement. The parallel with GEO is obvious: in both cases, a brand can no longer simply assert its value. It has to be corroborated by external, human signals.
What this model becomes in Southeast Asia
The structure of the mix holds from one market to the next. The platforms that fill it do not. A GEO strategy calibrated on US data systematically underestimates two layers in the Thai market.
Dark Social runs through LINE, not WhatsApp
In Thailand the dominant messenger is LINE, with roughly 56 million monthly active users in a country of 71 million. Recommendations circulate in private groups, brands operate LINE Official Accounts, and after-sales service happens in conversation. This layer is structurally invisible to analytics tools — exactly the problem SparkToro documented, but at a scale where it constitutes the default trust channel rather than a residual blind spot.
Retail Media means Shopee and TikTok Shop
Amazon and Walmart have no local equivalent. Discovery near the point of purchase happens on Shopee, TikTok Shop and Lazada. TikTok Shop has overtaken Lazada for second place in the Thai market, and commerce driven by live and short-form video now accounts for a significant share of the country’s e-commerce sales. Thailand ranks among the highest social commerce adoption markets in the world: the boundary between social discovery and transaction has largely dissolved there.
The relative weight of Retail Media and Dark Social is higher there than in the global model, and the two layers overlap: a recommendation shared in a LINE group can lead directly to a purchase on Shopee, without any search engine or attribution tool ever seeing it.
The funnel becomes a recommendation ecosystem
The classic journey could be read as: Search → Website → Comparison → Conversion.
The 2026 journey looks more like: AI recommendation → Social validation → Reddit / reviews → Creator demonstration → Website / marketplace → Conversion.
But that order is not fixed. One prospect may start on TikTok, another on ChatGPT, a third in WhatsApp and a fourth on Amazon. Modern marketing is therefore no longer a perfectly linear funnel. It becomes a network of validation points.
What this changes for brands in 2026–2027
- Build a clear entity: name, activity, location, expertise, leadership, services and proof must be consistent everywhere.
- Map the real questions: prompts, comparisons, objections, alternatives and decision criteria replace the plain keyword list.
- Create authority territories: proprietary studies, guides, benchmarks, expertise and original data.
- Develop external proof: media, partners, reviews, specialist directories, forums and communities.
- Optimise for Social Search: titles, voice-over, captions and on-screen text must match how users actually search.
- Build a real community presence: not spam, but usefulness and credibility strong enough to generate natural mentions.
- Change the KPIs: AI citations, share of voice, sentiment, mentions, branded search, AI referrals, assisted conversions and reputation signals.
The most recent zero-click data is also the most brutal. SparkToro’s June 2026 study, based on Similarweb’s clickstream panel, measures that 68.01% of US Google searches produced no click at all in the first four months of 2026, up from 60.45% in 2024 — the fastest increase SparkToro has ever recorded on this metric. In other words, fewer than one search in three still sends a visitor to the open web.
One reading caution applies: SparkToro’s earlier studies relied on the Datos panel and reported 58.5% for the United States and 59.7% for the European Union in 2024. The change of data provider makes the series imperfectly comparable, and the 2026 figures published so far cover the US market only. The direction, however, is not in dispute. In this context, measuring traffic alone means ignoring a growing share of a brand’s real influence.
The winning strategy: Search Everywhere Optimization
SEO remains necessary. But it now has to sit inside a wider architecture: SEO to be findable, AEO to be extracted, GEO to be cited, E‑E‑A‑T to be credible, Social Search to be discovered, reputation management to be corroborated and commerce media to stay visible near the transaction.
Is your brand present in the environments your customers — and the algorithms assisting them — actually use to decide?
Sources and references
Every figure cited links to its primary source and publication date. Forecasts are flagged as such and kept distinct from measurements.
- Gartner — Search Engine Volume Will Drop 25% by 2026 (press release, 19 February 2024). Forecast, not realised in these terms.
- HubSpot — The State of AEO in 2026 (4,000+ marketers). Figures cited: 58%, 44%.
- HubSpot — State of Marketing 2026 (1,500+ marketers). Figure cited: 48.6% (short-form video, top-three ROI).
- HubSpot — Social Media Marketing Report 2026 (1,100+ marketers). Figure cited: 77% (authenticity over production value).
- DataReportal / We Are Social / GWI — Digital 2026 Mid-Year Global Update (April 2026). Brand research sources: 45.8% / 41.3% / 35.4% / 32.2% / 30.7% / 22.1%.
- DataReportal — Digital 2026 Global Overview Report. Social ads: third source of brand awareness, first among 16–34s.
- SparkToro / Similarweb — In 2026, Less than One Third of Google Searches Still Send a Click (June 2026). Figure cited: 68.01% (US, January–April 2026).
- SparkToro — Dark Social Falsely Attributes Significant Percentages of Web Traffic as “Direct” (April 2023). Figures cited: 100%, 75%, 30%.
- McKinsey — An Update on US Consumer Sentiment: Embracing AI-Supported Shopping (ConsumerWise, Q1 2026). Figures cited: 44% / 31% / 9% / 6%.
- McKinsey — Europe’s Agentic Commerce Moment (March 2026, n = 749, FR/DE/UK). Figures cited: 84%, 63%, 55%.
- McKinsey — Commerce Media: The New Force Transforming Advertising (2022). $100bn projection cited for historical context.
- McKinsey — State of the Consumer 2026 (June 2026). Fewer click-throughs to brand and retailer sites.
Verification note. The Retail Media spend figures ($53.7bn US 2024, ~$71bn US 2026, ~$140bn / ~$165bn global, 277 networks) come from IAB/PwC and eMarketer estimates relayed by the trade press. They have not been verified against the primary reports and should be before final publication. The figure of 56 million LINE users in Thailand comes from DataReportal’s Thailand data.