Calculators Get Quoted. Articles Get Summarized

01Three kinds of question, three different fates

I pulled six months of search data from Bing for all 41 sites I have verified, about 695,000 impressions in total, and sorted the queries by what the person wanted. Some wanted something explained: what is, why, how does. Some wanted options compared: this versus that, the best, the alternatives. And some wanted a number computed for them: how much, what is it worth, a calculator, a checker. The three groups were seen a similar number of times. They were clicked very differently.

What the gap means

Explanation queries were clicked 0.66% of the time. Comparison queries, 0.49%. Queries asking for a computed answer, 3.47%, about five times the rate. The explanation and the comparison are increasingly delivered on the results page or inside an AI answer, so the person reads the summary and leaves. The computation is different. An answer can summarize your explanation. It cannot run your calculation on somebody else's numbers. To get their own number, the person has to come to the page that does the math.

A word on method, because these groupings are blunt. I matched simple word patterns, so a query asking for the best free converter lands in two groups, and the three groups do not add up to the whole. The direction is the point, not the decimal. Even if every boundary case fell the wrong way, the computed group would still be clicked several times as often as the others.

02The answer points at the tool

Quoted as the place to go

When an assistant is asked how much a claim is worth or whether a site is blocked, the honest answer is that it depends on inputs only the person has. So the answer explains the factors and names a place that can do the calculation. That is how a free tool earned ten thousand citations a month: not as a source of facts, but as the recommended next step. An explainer gets absorbed into the answer. A tool gets named in it.

The questions that prove it

The most seen long query across all 41 sites was a person asking how much their accident case was worth based on how badly they were hurt. That is not a request for an article. It is a request for a result. Many of the other long queries were variations on it: what a claim is worth in a particular state, for a particular injury, after a particular kind of crash. The same pattern made the checker tool everyone fears produce 36% of one site’s search impressions: people search for a verdict about themselves, and only something that computes can give one.

03A calculator that a machine can also read

Put the method in the page

The mistake most calculators make is being all widget. A form, a button, a number, and nothing a crawler or an assistant can read. The ones I build now put the method in plain text beside the form: what the inputs are, the formula, the typical ranges, and where those ranges come from. The tool serves the person. The text serves the machine, which can quote the method and send the person to the tool. A calculator with its method written out earns the click and the citation from the same page.

Where comparisons fit

Comparison pages are not wasted because they are rarely clicked. Their value is being the source an answer leans on when someone asks which option to choose, measured in citations rather than visits. The mistake is expecting them to bring traffic. Pair the two: the explainer from explainers and listicles to be quoted, and a tool on the same subject to be visited. Plan every topic as one page that explains and one page that computes.

One caution. A calculator in a sensitive area, like health, money or legal claims, has to be more careful than an article, because a number feels more authoritative than a paragraph. Show ranges rather than a single figure, say what the estimate cannot know, and name the sources of every multiplier. A confident wrong number from a tool does more harm than a vague paragraph ever could.

Tomorrow: the searches no person typed, and what they reveal about how assistants search on our behalf.