“Navigational. Informational. Transactional.” Those three words come from a 2002 paper by Andrei Broder, then a researcher at AltaVista, titled “A taxonomy of web search.” Broder surveyed AltaVista users and analyzed query logs, and found that a large share of searches were not about information at all. People were trying to reach a site or get something done. His categories have survived every redesign of the results page since.
What has not survived is the share each category holds, or where each one happens. The search intent statistics below trace how intent has moved from the early web to Google’s AI features and chat assistants, with sources so you can check each number and decide what it means for your own keyword plan.
The original numbers

Broder’s 2002 survey estimated that roughly 40% to 48% of queries were informational, about 20% to 25% were navigational and roughly 22% to 36% were transactional. The ranges were wide because he used two methods, a user survey and a log analysis, and they did not agree perfectly. The bigger point was that informational queries were not the whole story, which surprised a field built around document retrieval.
Six years later, a team at Penn State led by Bernard Jansen took a different approach. They built an automatic classifier and ran it over hundreds of thousands of queries from several search engines. Their 2008 results put about 80% of queries in the informational group, with about 10% navigational and about 10% transactional.
Why the gap between the two studies? Method matters. Broder counted many queries as transactional when users wanted to download, buy or use something online. Jansen’s classifier drew tighter lines. Both studies are useful, but neither number is “the” answer. Every modern intent statistic carries the same caveat: the share you get depends on how you define the categories.
The lasting lesson is the one most planning decks ignore. Most searches are people trying to learn something, but a meaningful minority are people trying to go somewhere or do something. A content plan built only for the first group misses the searches closest to a sale.
Google in 2026: more answers, fewer visits
Google does not publish an official intent breakdown, but it does share a few numbers that shape how intent plays out. The most quoted is that about 15% of the searches Google sees every day are new, meaning they have never been searched before. Google has repeated that figure since at least 2013. Its practical meaning is that no keyword tool can show you every query your buyers type.
The bigger shift is what happens after the search. SparkToro and Datos found in 2024 that 58.5% of US Google searches ended without a click to any website. Informational queries are the most likely to end that way, because the answer appears on the results page itself, in a featured snippet, a knowledge panel or an AI Overview.
That changes the value of each intent type. Informational searches still make up most of the volume, but they now deliver the smallest share of visits. Navigational and transactional searches send a larger share of their volume to websites, because the searcher still needs to go somewhere to finish the job.
Google has also said that queries in AI Mode tend to be much longer than traditional searches. Longer queries carry more intent detail. Someone typing “best running shoes” might be browsing. Someone asking AI Mode for running shoes for flat feet under $150 for a first half marathon is close to buying.
That detail is pushing marketers to treat commercial investigation as its own category, separate from pure information and pure transactions. Broder’s original taxonomy folded comparison shopping into other buckets, but today it is where much of the money sits. Queries with words like “best,” “vs,” “review,” “alternatives” and “pricing” sit between learning and buying. They trigger shopping features, review roundups, forum threads and AI comparisons, and they are the queries most likely to name specific brands in the answer.
Mobile adds another layer. Phone searches skew toward immediate needs: directions, hours, a quick fact, a nearby option. Desktop searches skew toward longer research sessions. The same keyword can carry different intent depending on the device, which is one more reason to check results on a phone before you decide what a query means.
Chat assistants changed the mix

The most useful recent data on intent outside Google comes from an OpenAI and Harvard economist research paper published in September 2025, “How People Use ChatGPT.” The authors classified a large sample of conversations and found that about 49% of messages were asking for information or advice, about 40% were doing something, such as writing, editing or planning, and about 11% were expressing, such as sharing feelings or chatting. They also found that the share of non-work messages grew from about 53% in mid-2024 to about 73% by mid-2025.
Those categories do not map perfectly onto Broder’s, but the overlap is clear. Asking is the chat version of informational and commercial intent, often with more context and more follow-up questions. Doing is a new category that classic search engines never served well: tasks people now hand to an assistant instead of searching for a tool.
This matters for brands because commercial questions are moving into chat. When someone asks an assistant to compare three CRM tools or recommend a lawyer for a small business in Denver, the assistant often answers with named brands. There is no list of ten blue links to compete in. You are either in the answer or you are not.
What the numbers mean for B2B and local brands
Search intent statistics are averages across every kind of query, from celebrity gossip to tax forms. Your market is narrower, and the mix inside it can look very different.
B2B categories tend to carry more commercial investigation than the averages suggest. A buyer choosing payroll software or a logistics partner researches for weeks, compares vendors, reads reviews and asks peers. Much of that research now happens in chat assistants, where the buyer can ask follow-up questions without opening twenty tabs. For a B2B brand, the most valuable queries are often low in volume and high in detail, the kind that never show up near the top of a keyword tool export.
Local businesses sit at the other end. Their searches are short, urgent and often on a phone: a plumber now, a dentist open on Saturday, a restaurant nearby. Intent is transactional almost by definition, and the battle happens in the map pack, reviews and Google Business Profile rather than in blog content. Informational content still helps, especially for trust, but it rarely drives the call.
Consumer brands fall in between, and their intent mix changes with the season. Gift searches swing commercial in November and December. Health and fitness searches swing informational in January. Watching that rhythm in your own Search Console data is more useful than any industry-wide percentage.
The statistics to stop quoting
Two numbers deserve a warning label. The first is the claim that 46% of Google searches have local intent. It circulates in hundreds of marketing articles, but its original source is thin, and Google has not published a methodology for it. Local intent is clearly common, and “near me” style searches are a large category, but the precise figure is not something to build a forecast on.
The second is any single intent percentage presented without a method. Different tools classify the same keyword differently. A query like “Salesforce pricing” might be labeled commercial by one tool and transactional by another. When you see a clean, round intent statistic, check who measured it, how and when.
The Intent Drift Index
Because intent depends on how Google interprets a query at a given moment, it changes over time. A query that returned blog posts two years ago may now return product pages, a local pack or an AI Overview. We track this with a simple measure we call the Intent Drift Index.
Take your 50 most important keywords. For each one, record what type of result dominates the first page today: guides, product or category pages, local results, videos, forums or an AI Overview. Repeat the check every quarter, signed out, on mobile. The Intent Drift Index is the share of those keywords whose dominant result type changed since the last check.
A drift of 5% a quarter is normal. A drift above 20% means Google has reinterpreted a big part of your market, and your content plan needs to catch up. This pattern showed up across many informational B2B queries during 2025, as AI Overviews absorbed simple definitions and pushed the remaining organic slots toward comparison pages and forums.
Three moves for your 2026 plan
First, sort your keyword list by intent, then by click potential. Informational queries with heavy AI Overviews are now citation targets, not traffic targets. Write clear, direct answers with sources so your brand gets named inside the summary. Commercial and transactional queries still deliver visits, so those pages deserve your best conversion work.
Second, research intent in the assistants too. Ask ChatGPT, Perplexity and Gemini the commercial questions your buyers ask and note which brands and sources come up. The search intent statistics from chat suggest those conversations will only grow, and the brands named in them gain a head start.
Third, run the Intent Drift Index every quarter. It takes an afternoon and catches shifts long before they show up as lost traffic. Pair it with the search intent statistics in your own analytics, such as the share of visits that land on commercial pages versus guides, so you can see whether the drift is costing you visits or only changing where they land.
Broder’s three categories are now 24 years old, and they still describe why people search. A user who typed a company name into AltaVista in 2002 and a user who asks ChatGPT to compare that company with two rivals in 2026 share the same goal. Only the place the journey ends has moved, and the brands that plan for that ending will keep the customers who never click at all.