A SaaS team can publish a well-researched, well-linked page on the right keyword and still watch it sit on page three for months. When I trace those pages back, the cause is rarely the writing. The page answers a different question from the one the searcher asked. A buyer typing a comparison query gets a feature tour. A founder looking for a definition gets a pricing page. The keyword matched. The intent did not.
Search intent is the part of SEO that decides whether everything else on a page gets a chance to work. In 2026 it also decides whether ChatGPT, Perplexity, Gemini or Google's AI features pull your page into an answer. This explainer covers how intent is classified, how to read it from a query, where the standard models fall short for B2B SaaS, and how conversational AI search has changed what intent looks like.
Search intent (also called user intent, query intent or keyword intent) is the goal behind a search: what the person wants to know, do, compare or buy when they type or speak a query. It matters to anyone who publishes content to be found, because search engines and AI assistants rank and cite the pages that best fulfil that goal, not the pages that repeat the words most often.
The Four Classic Intent Types, and Why They Are Only a Starting Point
Most SEO teams sort queries into four buckets: informational, navigational, commercial investigation and transactional. The model is useful because it forces one question before writing anything: what does this person want to walk away with? It is a starting point only, because real SaaS queries often straddle two buckets and carry details the buckets ignore.
Informational intent
The searcher wants to understand something. Queries such as "what is product-led growth" or "how does SOC 2 audit work" are informational. The right page explains clearly and early, then adds depth. Selling hard on these pages usually backfires, because the reader is not yet evaluating vendors.
Navigational intent
The searcher wants a specific site or page, such as a login screen, a docs page or a brand's pricing page. You cannot win a navigational query for another brand, and you should not try. What you can do is make sure your own navigational queries (your brand plus "pricing", "API docs" or "status") land on the correct page.
Commercial investigation
The searcher is weighing options: "best CRM for startups", "Notion vs Confluence", "alternatives to a given tool". For B2B SaaS this is the most valuable bucket. The reader has a problem, knows the category exists and is choosing. Pages that win here are honest comparisons, decision criteria and use-case breakdowns.
Transactional intent
The searcher is ready to act: start a trial, book a demo, buy a plan. Queries like "[tool] free trial" or "[tool] pricing annual" belong here. These pages need to be short on explanation and fast to convert. This bucket is also smaller than most teams assume. A study of web search queries cites earlier research finding that approximately 5%-7% of distinct web search queries contain product intent, which is a reminder that most searching happens before anyone is ready to buy.
How a Single Query Reveals Its Intent
You read intent from the words in the query and from what already satisfies it. The words give modifiers. The existing results, and the answers AI assistants give, show what search engines have learned people actually want from that phrasing.
Modifiers are the fastest signal. Take one search intent example built on the same core term, "onboarding software":
- "what is onboarding software" asks for a definition and scope, so it is informational.
- "onboarding software for remote teams" adds a use case, which leans commercial and expects a filtered shortlist.
- "onboarding software vs HRIS" asks for a category comparison and expects a clear distinction.
- "Appcues pricing" names a brand and a commercial detail, so it is navigational with commercial weight.
- "onboarding software free trial" signals readiness to act, so it is transactional.
Five queries, one head term, five different pages. Treating them as one keyword with five variations is the most common way I see SaaS blogs end up with pages that compete with each other and satisfy none of the searchers.
The second signal is what already wins. If the top results for a query are listicles, Google has learned that searchers want options. If they are docs pages, searchers want implementation detail. If an AI Overview answers the question in two sentences and the organic results below are thin, the query may not support a long article at all. Reading that landscape is part of reading intent.
Intent Has Layers: Format, Angle and Depth
Knowing the intent type tells you what kind of page to build. It does not tell you how to build it. Three further layers decide whether a page fits.
Format. The shape of the content: a list, a step-by-step guide, a comparison table, a calculator, a template, a definition. A query like "SaaS churn rate formula" wants the formula near the top and a worked example, not a long essay on retention strategy. Format mismatch is common because teams default to one blog template for every query.
Angle. The specific frame the searcher expects: beginner or advanced, for startups or for enterprise, current year or evergreen. "Email deliverability for SaaS" read by a solo founder needs different examples from the same topic read by a lifecycle team at a 500-person company. The angle often shows up in modifiers such as "for small teams", "simple" or "enterprise".
Depth. How far the answer needs to go. Some queries are satisfied by a paragraph. Others need a full walkthrough with edge cases. Over-delivering on a shallow query buries the answer. Under-delivering on a deep one sends the reader back to search, and AI assistants learn to skip pages that never get to the point.
These layers are why two articles with the same intent type can perform very differently. Both may be informational. Only one matches the format, angle and depth the searcher expected.
Where B2B SaaS Intent Gets Complicated
B2B buying adds pressures that consumer-focused intent models do not account for. The standard four buckets assume one searcher making one decision. A SaaS purchase usually has several people searching at different stages with different goals.
- Multiple roles search for the same product category. An engineer searches for API limits and integration effort, a finance lead searches for pricing models, and a marketing lead searches for outcomes. Each role is a separate intent.
- Problem-aware queries come long before solution-aware ones. A buyer searches "why are trial users not activating" months before searching for an onboarding tool. That query is informational in form but commercial in value.
- Jargon shifts intent. "PLG metrics" and "product-led growth metrics" may be the same topic for different audiences, one already fluent in the space and one learning it.
- Comparison queries carry high stakes. A reader searching for alternatives to a named tool is often unhappy with a current vendor, and expects frank trade-offs, including where your product is weaker.
- Integration and compatibility queries ("does X integrate with Salesforce") look navigational but are often late-stage purchase checks.
Mapping these properly usually means one page per distinct intent, connected through deliberate internal linking so a reader who starts on a problem-aware explainer has an obvious path to the comparison and then the product page. The links do work for search engines and AI crawlers too, because they make the relationship between the pages explicit.
How AI Assistants Read Intent Differently From Ten Blue Links
AI assistants infer intent from the whole conversation, not from one short query. That changes what intent looks like and which pages get used to answer it.
On Google, a query is often two to four words, and the engine has to guess the rest. In ChatGPT, Claude, Perplexity or Google AI Mode, the user writes the context out: their company size, their stack, their budget, what they already tried. The intent is stated, not implied. The assistant then pulls passages from pages that answer that specific, stated need.
The scale of this shift is significant. According to Search Engine Journal's review of two 2026 studies, Google's AI Mode has surpassed 1 billion monthly active users, follow-up questions are rising 40% month-over-month, and more than one in six searches in AI Mode now incorporate voice, image or video. The same report notes that the intent behind longer AI Mode queries is interrogative and heavily weighted toward evergreen content, at 75.2%.
Three practical consequences follow for how intent shapes content in 2026:
- Follow-up questions are part of the intent. A page that answers the first question and the obvious next two is more useful to an assistant building a multi-turn answer.
- Passages need to stand alone. An assistant lifts a paragraph, not a page, so each section should answer its own question without relying on a heading or an earlier pronoun.
- Evergreen explanation carries weight. Assistants favour pages that explain how something works, which rewards clear definitions, mechanisms and trade-offs over thin news posts.
Clicks are also less reliable as the reward for matching intent. SparkToro's 2026 clickstream data, reported by Search Engine Journal, shows 68% of Google searches in the U.S. now end without a click to the open web, up from 49% in 2019. When the answer is delivered in place, the win is being the source that answer names. Some teams add machine-readable signposts for AI crawlers as well, which is where questions like what is llms-txt come in, but no file substitutes for a page that matches the stated intent.
Mixed and Shifting Intent
Some queries serve more than one intent at once, and the intent behind a query can change over time. Both cases explain why a page that ranked well last year can slip without any change to the page itself.
Mixed intent shows up when the results for a query blend explainers and vendor lists. "Customer data platform" might return a definition, a vendor comparison and a product homepage in the same set. Google is hedging because searchers using that phrase want different things. A page can sometimes serve both, for example a definition followed by a short comparison of approaches, but trying to be a definition, a listicle and a sales page at once usually serves none of them well.
Shifting intent happens when a category matures. When a product category is new, searches for its name are mostly informational. As it becomes established, the same searches turn commercial, because people already know what it is and want to choose. AI categories moved through this cycle fast. A page written for the learning phase can drift out of alignment as the audience moves on, and the drop shows up gradually in traffic data before anyone notices the cause.
What Intent Mismatch Looks Like in Your Data
Intent mismatch leaves fingerprints in Search Console and analytics long before a page drops out of view. Learning to recognise them is how intent turns from a planning concept into something you can act on.
- High impressions with a low click-through rate on a query where the page ranks on page one, which suggests the title promises something the searcher does not want.
- A page ranking for queries you did not target, often a sign Google sees a different intent in your content than you intended.
- Positions stuck between 8 and 20 for months despite strong links, which often points to a format or depth mismatch.
- Short engagement time on informational pages, meaning readers did not find the answer early.
- Two of your own URLs alternating for the same query, which signals both pages target one intent.
- AI assistants naming competitors for buyer questions your pages address, which usually means your answer is buried or framed for a different reader.
Google data covers the first five. If you have not yet, Connect Google Search Console to whatever system you use to review pages, so query-level data drives decisions. The sixth needs AI visibility tracking, because Search Console shows nothing about ChatGPT or Perplexity. That gap is why many teams pair classic seo monitoring tools with a separate view of what assistants say.
Fixing mismatch is rarely a full rewrite. Often the answer moves to the top, the format changes from essay to table, or one bloated page splits into two with distinct intents. What matters is diagnosing the intent first, then editing toward it.
Gofylo shows which businesses six AI assistants (ChatGPT, Google AI Overviews, Google AI Mode, Gemini, Perplexity and Claude) name for your buyers' searches, then researches, writes and publishes intent-matched articles to your CMS and refreshes them from your Search Console data. Plans start at $149 a month for 10 articles. Start a 7-day free trial or run the free AI Visibility report to see the buyer questions where competitors are named instead of you.
Related: SEO Content Creation in 2026: What Ranks and Gets Cited