Two properties with identical traffic can be worth wildly different amounts, and the entire difference is usually in the queries. Intent is the highest-value thing you can read in a search report, and it is free.
The standard buckets ... informational, navigational, commercial, transactional ... are a reasonable starting point and they are also where most analysis stops. The interesting work starts when you look at what the query is trying to decide.
Reading intent by the decision behind it
Instead of asking what stage a query belongs to, ask what the person is about to do and what they need in order to do it:
- "What is X" ... they cannot act yet. Enormous volume, real value for building association with a subject, almost no immediate commercial value.
- "X vs Y" ... they have narrowed to two options and are choosing. High commercial value, and notably hard to serve well, which is why so much comparison content is thin.
- "X pricing" or "X cost" ... budget stage. Extremely valuable and mostly unanswered, because vendors hide pricing and content sites will not guess.
- "X alternatives" ... they have an option and a problem with it. This is the highest-intent query most categories have, and it is a competitor's customer in motion.
- "How to fix X" ... they have a live problem. Value depends entirely on whether your product or client sells the fix.
- "Is X worth it" ... they are looking for permission or a reason not to. Underrated, and usually served by a review nobody trusts.
The shape matters more than the volume. A category with fifty searches a month for "alternatives to" is a better commercial surface than ten thousand for "what is."
Intent profile as a property-level measure
At the property level the useful number is not any single query. It is the distribution: what share of your impressions, and your clicks, sit on queries attached to a decision.
Two properties, same 50,000 monthly impressions. The first has 95 percent definitional traffic and the intent profile of an encyclopedia. The second has 40 percent comparison and pricing queries. The second is not slightly more valuable. It is a different asset class, and a buyer reading the query report will see it in the first five minutes.
The same measure works as a diagnostic on your own property. Intent profile drifting toward the definitional end over time is what content expansion usually does if nobody is watching, because informational topics are easier to write and rank for. The traffic line goes up. The asset gets weaker. That is the specific failure the Measure stage exists to catch.
Where the model breaks
Three honest limits.
Intent is inferred, not observed. We are reading a few words and assigning a mental state to a stranger. A query like "IBM Power pricing" might be a buyer with a budget or a competitor checking positioning. Aggregate it and the inference holds up reasonably. Do not use it to make confident claims about individuals.
Some of the highest-value queries are invisible. Search Console does not disclose every query, low-volume terms are withheld, and the queries a buyer types right before converting are often the rarest ones. Absence in the report is not absence in reality.
Machine-shaped queries complicate the picture. Some query rows now have a shape no person would type, long and specific and fully formed. Whether a person or a system originated them, they represent a real expressed information need, and they are worth investigating rather than filtering out. What they are not is proof of a human buyer.
What to do with it
The practical use of intent analysis is deciding what to build next, and it usually points somewhere uncomfortable: the queries with real commercial intent are harder to write for, lower volume, and less fun than the topic you had in mind.
The check I use is whether a page can honestly help someone decide. If the answer is no, it is fine to publish it, but it belongs in the discovery column and not the commercial one, and it should not be counted twice.