955 of 1,000 economics searches ended without a click

1. The single most-searched thing was not a question. It was the back of a flashcard.

601 impressions, 0 clicks, average position 8.0

The top query of the entire window was economic profit equals total revenue minus explicit and implicit costs definition. The second was the same statement reordered, with 420 impressions. Two more rewordings appear further down the top 25. Students were not asking what economic profit is. They were typing an answer they already had and appending the word “definition” to see whether it matched.

That is term-to-definition matching behaviour, the shape of a flashcard deck or a matching worksheet, not of curiosity. For a teacher it is a useful tell: when a class searches like this, they are verifying memorised strings rather than testing whether they can apply the idea. All 17 queries in the economic profit cluster drew 1,519 impressions and zero clicks between them.

2. One search in ten is a homework assignment pasted whole into the box.

98 of 1,000 queries ran to 10 words or more. The longest was 117 words.

The median query is 4 words, so most searching still looks like searching. But 224 queries ran to six words or more and 149 to eight or more, and at the long end the text stops being a query at all. Several are assignment instructions copied verbatim, punctuation and all, including one that still carries the filename of the image the student was meant to be looking at.

Two things are worth separating here. The pasted worksheet items are what teachers already suspect. The other kind is newer: queries written as instructions to an assistant rather than as search terms, like the study-guide request above. Students are typing prompts, and the search box is catching them. If you write an assignment whose text can be pasted and answered, it will be.

3. Ranking on the first page bought almost nothing.

191 queries put us in the top 10 and returned zero clicks, across 5,197 impressions. 63 of those ranked in the top 5.

Strip out the two queries that are our own name and the picture is starker still: 998 non-brand queries produced 24,843 impressions and 48 clicks, a click-through rate of 0.19 percent. The two brand queries alone accounted for 64 of the 112 clicks in the sample.

We cannot see from this data why a student did not click, and we are not going to pretend otherwise. What we can say is that being visible and being read have come apart. A result that appears near the top of the page and is never opened is consistent with the answer already being legible on the results page itself. The practical consequence for a teacher is that a student can arrive at an answer having opened no source at all, which means there is no source to ask them about.

4. A seventh of all demand is students trying to tell two things apart.

143 queries were comparisons, drawing 3,452 impressions, 13.7 percent of the sample.

These are the most diagnostically useful queries in the whole set, because a comparison query is a student naming the exact boundary they cannot hold. The list below is a ranked map of which distinctions in the course are hardest.

Look at the mirror pairs. Students searched both elastic vs inelastic and inelastic vs elastic, and both sras vs lras and lras vs sras, at near-identical volume. Nobody searches a pair in both orders if they know which one is which. The order carries no information because the confusion is symmetric, and a symmetric confusion is usually a sign the two ideas were introduced together and never separated by a worked contrast.

5. Students search formulas they have already written down, to check them.

Five near-identical phrasings of one tax incidence formula drew 461 impressions and zero clicks.

Calculation queries were 145 of the 1,000, drawing 3,335 impressions and 6 clicks. The tax incidence cluster shows what is happening inside that number: the student has the expression in front of them and types it into the search bar, symbols included, to find out whether it is right.

Notice that two of the five drop the absolute-value bars around the demand elasticity term. A student who does not know whether the bars belong there does not understand why the formula uses a magnitude, and that is a conceptual gap wearing the costume of a notation gap. It is also the single cheapest thing on this page to act on in a lesson.

6. Almost nobody types “AP”.

25 of 1,000 queries mentioned AP, an exam, an FRQ, a unit number or a score. That is 2.5 percent.

This is a site built for AP Economics, and exam vocabulary is nearly absent from the way its visitors search. They type the concept: market structure, business cycle, total revenue test. Students in the moment of being stuck do not think of themselves as taking a course. They think of themselves as not understanding one thing.

The corollary matters for anyone writing materials: labelling a resource by unit and exam is labelling it for the teacher, not for the student who needs it at 11pm. Ranked by demand within this sample, the concepts students got stuck on were these.

Clusters were built by matching each of the 1,000 queries against a keyword set per concept, so a query can fall into more than one cluster and the columns do not sum to the sample total. Elasticity drew 2,054 impressions across 79 queries and not one click.