Reading intent from question phrasing

When teams classify website questions, they usually start with the page URL. It is the obvious feature: pricing page means pricing intent. That works, but it is weaker than it looks, because a lot of visitors ask a question on whatever page they happen to be on when the thought occurs to them.

Phrasing is the stronger predictor, and it costs nothing to use because it is already in the text.

Modal verbs carry stage

The auxiliary verb in a question is a surprisingly reliable stage marker.

• “Can I” and “do you” indicate capability checking. The person is establishing whether the product qualifies. Early stage.
• “How do I” indicates procedure seeking. The person has accepted the product does it and wants to know the mechanism. Middle or post purchase.
• “Why does” or “why is” indicates troubleshooting. Something did not behave as expected. Post purchase, usually.
• “What happens if” indicates risk assessment. Late evaluation stage, often the last question before a decision.
• “Which” or “should I” indicates comparison. The person has multiple options in play, either your plans or your competitors.

The last two are commercially the most interesting and the least likely to be prioritised, because they are lower volume than “can I” questions and higher value.

Specificity is a proxy for depth

A question that names a specific feature, plan, integration or limit came from someone who has read something. A question phrased in general terms came from someone earlier in the process.

A rough measure that works: count the proper nouns and product specific terms in the question. Zero means general browsing. Two or more means the person has done homework, and their question is worth reading individually rather than only in aggregate.

Negation is a strong signal

Questions containing a negative, such as “does it not support”, “is there no way to”, or “can I avoid”, usually indicate the visitor has already found something they consider a problem and is checking whether it is a dealbreaker.

These are low volume and high stakes. If you only read one category by hand, read this one.

A practical classification approach

You do not need a model to start. A small set of pattern rules applied to the question text will sort the majority correctly, and the residue can be read manually. Something like:

stage = 'evaluate' if question starts with can/do/does/is there
stage = 'procedure' if question starts with how
stage = 'trouble' if question starts with why
stage = 'risk' if question contains 'what happens' or 'what if'
stage = 'compare' if question contains 'vs', 'versus', 'or', 'better than'
stage = 'unknown' otherwise

This is deliberately crude. The point is not accuracy in the abstract, it is that a crude classifier applied consistently reveals distribution shifts, and a distribution shift is the thing you actually want to notice. If risk stage questions double after a pricing change, that is worth knowing even if the classifier misfiles a third of them.

Where phrasing analysis breaks

Two failure modes are worth stating. Non native speakers phrase questions differently and a rule set tuned on native phrasing will misclassify them systematically. And visitors who paste a long block of context do not fit any of these patterns.

Both argue for the same thing: use the classification for aggregate trends, and read the individual conversations that matter rather than trusting a label. The classifier is a sorting aid, not a substitute for looking.

The signals that are actually reliable

Phrasing carries stage more reliably than the page someone is on, because a visitor can land on pricing from a search result without being anywhere near a decision.

Comparative language is the strongest late stage signal. Anything naming a competitor, asking about migration, or asking how something differs. Someone comparing has already decided they want the category.

Constraint questions are next. Limits, edge cases, what happens when you exceed something. Nobody interrogates a constraint they will not hit, so these come from people modelling their own usage.

Definitional questions are early and should usually not be qualified at all. Answering them well and letting the person leave is the correct outcome.

Signals that look useful and are not

Message length. Long questions feel more serious and are not. They correlate with unfamiliarity more than with intent, because people who know the domain ask shorter questions.

Politeness. Some visitors write full sentences with greetings and some write three words. This is personality and regional convention, not buying stage.

Time on site before asking. It is genuinely ambiguous. A long dwell before a question can mean careful evaluation or it can mean the person could not find anything and finally gave up looking.

Urgency words. Asking for something quickly reads as high intent and frequently means the opposite, because people in a hurry are often price checking rather than buying.

Using intent without being creepy

The visitor should never feel profiled. Nothing about the intent read should surface in the conversation as an observation about them.

In practice that means intent changes what the assistant does next, not what it says about the person. A high intent question gets the answer plus one relevant follow up. A low intent question gets the answer and nothing else.

It also means not escalating on intent alone. Someone asking a comparison question does not want a salesperson to appear. They want the comparison answered.

Where this breaks down

Short factual questions carry almost no signal, and they are a large share of traffic. Any system that assigns confident intent to a four word question is overfitting.

Multilingual sites are harder, because phrasing conventions differ and the patterns above come from English. Treat them as a hypothesis to check rather than a rule to apply.

And a single question is a weak signal in general. Intent becomes legible across two or three exchanges, which is another reason to answer first: the second question tells you far more than the first.

Intent changes what to log, not just what to say

The usual framing treats intent as something that shapes the reply. It is at least as useful for deciding what to record.

High intent questions that were refused are the most valuable line in any content gap report. Someone close to a decision asked something your site could not answer, and that is a different priority from a definitional question going unanswered.

Low intent questions that were answered well are worth logging as coverage rather than as opportunity. They confirm the content is doing its job and they should not be counted in a pipeline number.

Sorting the gap list by intent rather than by frequency changes what gets fixed first, and usually for the better, because frequency favours easy general questions and intent favours the ones attached to revenue.

Terms in this articleIntent