Insights / Search & AI Visibility

Article

Where do AI engines get their answers? The real key to AEO.

Shamik Talukder, Co-founder & CEO·4 November 2026·6 min read

Short answer: Answer engine optimization (AEO) starts with one question: where does the AI engine get its answer? ChatGPT, Google AI Mode and Perplexity build replies from a small set of sources, often retailers, forums, creators and even competitors. Brands win by mapping those sources for every buying question, then earning accurate, credible mentions in them.

What is AEO, and how is it different from SEO?

AEO is about being the answer, not one of ten blue links. When someone asks an AI assistant which running shoe, bank or school to choose, the engine writes a reply. It names a few brands and cites a few sources. AEO is the discipline of making sure your brand is in that reply.

SEO still matters, because AI engines often retrieve pages from a search index before they write. But the unit of competition has changed. SEO competes for a ranking position. AEO competes for a sentence, a recommendation or a citation card.

That shift changes who you are really competing with. Your rival is no longer only the brand ranked above you. It is every review, forum thread and creator video the engine might quote instead of you.

Why does AEO matter now?

Because the answer is now the main event. At I/O in May 2026, Google said AI Mode had passed one billion monthly users in its first year. Queries in AI Mode have more than doubled every quarter since launch.

Clicks are thinning out. SparkToro data reported by TechWyse found that 68% of US Google searches ended without a click in early 2026, up from about 60% in 2024. When an AI Overview appears, click-through drops by nearly 60%.

The last four weeks added three signals:

  • Citations moved up the page. On 15 October, Google redesigned AI Overviews and AI Mode so source citations sit higher, as larger thumbnail cards, according to Similarweb’s October roundup.
  • Being recommended beats being footnoted. The same roundup cites a study where ChatGPT’s top recommended brand drew a 10.3% to 14.6% click-through rate. Standard footnote citations drew about 2.3%.
  • Each engine picks its own winner. On 27 October, Bluefish reported that Google and ChatGPT name a different top brand in 35% of 5,810 buying topics, per PPC Land.

Adoption is moving fast in our other markets too. A Visa study released in June found 85% of UAE consumers have used AI tools to help them shop. Across six Southeast Asian countries, an impact.com, Cube and Dentsu report found 24% already shop with generative AI, rising to 31% in Indonesia.

In India, Google extended AI Mode beyond English and Hindi to seven more languages, including Bengali, Tamil, Telugu and Marathi. Google calls India its largest user base for voice and visual search. Answers in those languages draw on a much thinner pool of content, so brands that publish in them have more room to be cited.

Where do answer engines get their answers?

Mostly from people talking, not from brand websites. In Ahrefs data on ChatGPT citations for US queries, Reddit holds a 16.8% share of top-source citations. Wikipedia follows at 7.0%, and YouTube ranks eighth at 2.5%.

The trend favours creators. AirOps research across about 3.5 billion citations found owned brand and product pages lost 10% of their citation share between August 2025 and June 2026. Creator and social citations grew 140% over the same period.

This is why we say AI recommends the brands people talk about. Engines look for consensus. If reviewers, forums and creators describe you the same way your website does, the engine has confidence to name you.

Shoppers lean the same way. In the Southeast Asia report above, 36% of consumers said they lean on creators over AI, against 25% who lean on AI. Creator content shapes the purchase and feeds the engines that answer the next shopper.

What does a source map look like? A Complan example

At BeTagged, we track which sources AI engines cite for every question in a brand’s category. The view below is from our Brand Chatter Intelligence platform. It shows the top sources ChatGPT, Gemini and other AI models cited for neutral, unbranded questions in Complan’s category in India, from 28 October to 3 November 2026. We have hidden most source names.

Top AI citation sources · Complan’s category, IndiaNeutral (unbranded) queries · All models · 28 Oct – 3 Nov 2026 · Source: BeTagged Brand Chatter Intelligence
#SourceModelsImpact scoreMention %SOA %SentimentCitations %
1horlicks.in CompetitorChatGPTGeminiOther40.616.2%7.5%515.6%
2apollopharmacy.in CorporateChatGPTGeminiOther39.111.4%5.9%574.9%
3amazon.in UGCChatGPTOther38.911.9%9.2%534.6%
4Other36.38.1%7.0%494.8%
5ChatGPTGeminiOther35.511.4%8.8%553.7%
6ChatGPTGeminiOther35.510.5%4.5%553.6%
7GeminiOther34.16.2%15.4%622.9%
8Gemini33.23.3%11.3%662.9%
9ChatGPTGeminiOther32.48.1%4.7%582.8%
10Other31.99.0%7.8%463.7%
Rows 4–10 hidden for client confidentiality. Mention % = share of AI answers that cite the source. SOA % = share of attention. Citations % = share of all citations.

Four patterns stand out:

  • A competitor’s own site leads. horlicks.in is the top source. It appears in 16.2% of answers and earns 5.6% of all citations.
  • Retailers act as reference books. Apollo Pharmacy and Amazon rank second and third, and more retailers follow further down the list. Engines read their product listings, ingredient details and reviews.
  • Each engine has its own sources. Several sources in the top ten are cited by only one of the tracked models, so winning one engine does not win them all.
  • The brand’s own site is not in the top ten. complan.com does not appear in the top ten, even for questions about its own category.

Once you can see the map, the to-do list writes itself. Fix the product pages on the pharmacy and commerce sites engines already trust. Brief nutritionists and parent creators on the social platforms one engine favours. Give the brand’s own site question-led pages worth quoting. Then rerun the map every week, because it moves.

The same exercise works in any market. A snack brand in Jakarta, a bank in Dubai or a skincare label in New York will see a different list. The method of mapping first and acting second stays the same.

How do you build an AEO programme, step by step?

We run AEO as six steps. The first three fix your own house; the last three shape what others say about you.

  1. List the questions that matter. Write down 50 to 100 buying questions your customers ask, in every market and language you sell in. Include comparisons like “best X for Y” and “X vs Y”.
  2. Map the sources behind each answer. Ask each question in ChatGPT, Google AI Mode, AI Overviews and Perplexity. Record who gets named first, how you are described and, above all, which sites the engine cites. Test each engine separately, since their winners and sources differ.
  3. Make your pages easy to quote. Put a direct two-sentence answer under each question-style heading. AirOps found 68.7% of ChatGPT-cited pages use a sequential heading structure, versus 23.9% of Google’s top results. Add FAQ and product schema, and refresh key pages at least yearly.
  4. Earn third-party mentions. Brief creators, reviewers and community voices on the claims you want repeated, in their own words. Prioritise the sources your map shows engines actually use, whether retailer listings, Reddit threads, YouTube reviews or regional news.
  5. Go local and vernacular. Publish answers in the languages your buyers search in, such as Bahasa Indonesia, Tagalog, Arabic, Hindi or Tamil. In our experience, fewer brands publish there, so each citation is less contested.
  6. Re-test monthly and close gaps. Rerun your question set every month. Where a rival wins, study what the engine cites for them, then fill that gap with content or creator coverage.

How do you measure AEO?

Measure presence in answers first, traffic second. Rankings and sessions alone will undercount AEO. Similarweb notes that ChatGPT Search sometimes strips UTM tags, so some AI visits land in analytics without a source.

We track four numbers for each market:

  • Answer share: the percentage of your question set where the engine names your brand.
  • First-mention rate: how often you are the top recommendation, not just one of several.
  • Citation share: how often your pages, or pages about you, appear as sources.
  • Sentiment and accuracy: whether the engine describes you correctly and favourably.

Then connect those numbers to revenue. A Semrush study reported by MarTech found the average visitor from AI search was worth 4.4 times a traditional organic visitor on conversion. Fewer clicks can still mean more sales, if they come from the answer.

Frequently asked questions

What is the difference between AEO and GEO?

AEO focuses on being the direct answer to a question, whether in a voice assistant, a featured snippet or an AI chat. GEO, generative engine optimization, focuses on how large language models gather and combine sources into a written response. In practice the work overlaps heavily, and we run both as one programme covering content, structure and third-party mentions.

Does AEO replace SEO?

No. Many AI engines retrieve pages from a search index before writing an answer, so crawlable, fast, well-ranked pages still help. AEO adds new work on top: question-led content, schema, freshness and earned mentions on forums, video and media. Treat SEO as the foundation and AEO as the layer that decides whether you are named.

How long does AEO take to show results?

Page fixes such as clearer answers and schema can be picked up once engines recrawl the page, often within weeks. Changing how engines describe your brand takes longer, because it depends on reviews, forum threads and creator content building up. We plan for a first read at one month and meaningful movement over three to six months.

Can influencer marketing improve AI visibility?

Yes. Creator and social content is a growing share of what AI engines cite; AirOps measured 140% growth in that share in under a year. When creators explain your product in their own words, on YouTube, Reddit or regional forums, they add the independent mentions that engines use to judge whether a brand is worth recommending.

Shamik Talukder
Shamik TalukderCo-founder & CEO, BeTagged

← More from Insights

Want to know what AI says about your brand?

We can run an answer-share benchmark across ChatGPT, Google AI Mode and Perplexity in the markets you sell in, from the US and UAE to Southeast Asia and India. You will see where you are named, where rivals win, and what to fix first.