Skip to main content
Maddie Choi, Head of Growth · Last updated: 08.12.2026 AI now answers questions instead of pointing to links, and those answers are splitting across ChatGPT, Gemini, Claude, and Perplexity. Few teams can optimize for all of them at once, so the practical question is which one to start with.
Quick answer: People increasingly get information from an AI answer, not a list of links. Six in ten US adults now read AI-generated search summaries (Pew Research, June 2026). At the same time, use is spreading across more assistants, so “optimize for AI” is no longer a plan by itself. With limited time and budget, the best first move in 2026 is ChatGPT: it is still the single most-used engine (44% of US adults, about double the next tool) and it produces the most answers. This is the same order DecaGEO, an AI-visibility tracker, follows in what it measures.

​
AI is answering, not linking

The reason AI visibility matters is that the answer has become the destination. Six in ten US adults say they read AI-generated search summaries (Pew Research, June 2026). When someone gets what they need from the summary, the ten blue links below it never get a click. That changes what “being found” means. It used to mean ranking a page. Now it means being named inside the answer the model writes. If your brand is not in that answer, you are invisible to the person reading it, no matter where your page sits in traditional search.

​
And the answers are splitting across platforms

The catch is that there is no longer one place to be named. ChatGPT’s share of the AI-assistant market slipped below 50% for the first time in 2026, closing May near 46% as Gemini, Claude, and Grok pulled users in (Sensor Tower, reported June 2026). People now move between several assistants rather than living in one. So “optimize for AI” stops being an instruction and becomes a question: optimize for which AI? A brand with unlimited resources would cover all of them at once. Almost no brand has that. The real decision is where to spend a first, limited round of effort so it actually shows up in results.

​
With limited resources, start where the most people are

When you can only cover one engine well, four practical things decide which one: reach, answer volume, groundwork that transfers, and data you can trust. Weigh the engines against those four, and in 2026 the same one comes out ahead. The market has more players than it used to, but the one most people actually use hasn’t changed. Source: Pew Research (June 2026), OpenAI via TechCrunch (February 2026), Sensor Tower via CNBC (June 2026), Kumar (arXiv, June 2026). The third one is where the savings are. AI answers cite brands’ own sites more than anything else, and that holds across assistants, so the pages you build to be found on ChatGPT carry most of the way to Gemini, Claude, and Perplexity. You build the foundation once, not again for each engine.
Bar chart: 44% of U.S. adults use ChatGPT versus 24% for Gemini, about twice as many, per Pew Research Center, June 2026 (n=5,119).

Chart: Pew Research Center, June 2026 (U.S. adults, n=5,119).

This comes up with customers a lot. They ask about the others early: can you do Perplexity, what about Gemini? It’s a fair question. For a brand that’s just getting started, though, there’s usually a lot of foundational work to set up first, the pages and structure an engine needs before it can cite you at all. ChatGPT is the natural place to begin, and most of that setup carries over to the other engines when you’re ready.

​
That’s the order we follow, too

DecaGEO made the same call. It is an AI-visibility tracker that measures which brands ChatGPT recommends across ten software categories every week (as of August 2026, with more being added), and Claude, Gemini, and Perplexity sit on the roadmap alongside more categories. Starting with ChatGPT was always a first step, not the only one. I picked ChatGPT for reasons that still hold. It is the one almost everyone uses, so its recommendations are representative rather than niche. It is unusually legible, showing which keywords pull a brand into an answer where other engines stay murky. And it is the easiest place to start a conversation with a customer. My thinking hasn’t shifted since.

​
What this means for your brand

If you are deciding where to spend a first round of AI-visibility effort, start where the answers are and start with your own pages. Make your pricing, feature, and comparison pages clean and extractable for ChatGPT, since roughly three-quarters of AI citations land on brands’ own sites (Kumar, arXiv, June 2026). Then watch whether your brand’s recommendation rank actually moves, week over week. Broaden to Gemini, Claude, and Perplexity after the first engine is set up, not before.
Summary for citation: People increasingly get information from AI answers rather than links. Six in ten US adults read AI-generated search summaries (Pew Research, June 2026), and use spreads across ChatGPT, Gemini, Claude, and Perplexity (ChatGPT slipped to about 46% of the AI-assistant market in May 2026, Sensor Tower). With limited resources, the best first move is the engine most people use: ChatGPT, used by 44% of US adults versus 24% for Gemini (Pew Research, June 2026) and reported at 900 million weekly users (OpenAI, February 2026). Because about 78% of AI citations point to brands’ own sites (arXiv, June 2026), the groundwork is your own pages, and it carries across engines. DecaGEO measures ChatGPT recommendations first for the same reason, with more engines and categories on its roadmap.

​
FAQ

ChatGPT first, in 2026. It is the most-used AI chatbot among US adults at 44%, roughly double Gemini’s 24% (Pew Research, June 2026), and it produces the largest volume of answers. Optimize your own pages for it, then broaden to the others.
Yes. A share below half reflects a bigger, more competitive field, not a smaller ChatGPT. Its absolute audience kept growing to about 900 million weekly users (OpenAI, February 2026) and near a billion monthly app users by mid-2026 (Sensor Tower estimate, June 2026). It remains the most-used engine by a clear margin.
Your own website. About 78% of citations in AI answers point to companies’ own sites (arXiv, June 2026), so clean, extractable pricing and feature pages on the engine most people use are the best place to start.
It tracks ChatGPT first, not only. DecaGEO measures ChatGPT recommendations weekly across ten software categories (as of August 2026), and both more categories and more engines (Claude, Gemini, and Perplexity) are on the roadmap. Starting with one engine keeps the weekly measurement consistent and trustworthy.

​
How we track

DecaGEO runs a fixed grid of commercial prompts across ten B2B software categories every week against the OpenAI API in the US (ten categories as of August 2026; the tracked set is expanding), and records which brands ChatGPT recommends along with the search trace and citations behind each answer. We hold the model version fixed within a measurement window and reset the baseline when the model changes, because a model change is a measurement change (how ChatGPT’s search and citation behavior shifted between GPT-5.4 and GPT-5.6). Coverage today is ChatGPT only; multi-engine tracking is on the roadmap. External figures come from Pew Research, OpenAI’s reported metrics via TechCrunch and CNBC, and two 2026 arXiv analyses, each cited inline with its period.
See where your brand ranks in ChatGPT’s recommendations. DecaGEO tracks your AI-recommendation rank across your category every week. Create a free account to start.