What 1,000 Student Searches Reveal About How Students Actually Learn Economics

In one week this July, EconLearn showed up in Google results for roughly a thousand different searches. That cap is not a rounding choice on our end; Google stops reporting distinct queries at 1,000, so the real number is at least that and probably higher. Those searches came from students trying to get economics unstuck at the exact moment they were stuck, and because every one of them is logged in our own search console, we can read the shape of that confusion instead of guessing at it.

This post is the first time we have pulled that data apart and written down what it says. It is not a survey and not a lab study. It is one platform, one subject, mostly one week, looking at the queries real students typed and where our pages met them. Four patterns came out of it clearly enough to be useful to anyone who teaches economics, and each one changed how we build. The full method, sample sizes, and limits are at the end, and teachers are welcome to reuse the findings with a link back.

For scale: the site is 1,120 free pages. In the week of July 8 to 14, 2026, it earned 28,650 impressions across 774 different pages, up from 9,860 impressions across 311 pages the week before. So the window we are reading is a moment of fast growth, which is part of why the patterns are legible. Growth exposes what students reach for first.

1. Students search their homework, word for word

The single most surprising thing in the data is how literally students search. They do not type tidy keywords like an SEO textbook imagines. They paste the sentence off the assignment, symbols and all.

Here is a real query that drew 96 impressions in one week: "economic profit equals total revenue minus explicit costs minus implicit costs definition." That is not how anyone talks. That is the exact wording of a definition a student was told to know, dropped into the search bar in the hope that something on the other side would confirm it. We ranked around position 10 for it, which means we were on page one but near the bottom, close enough to see the intent and far enough to know we had not fully earned it.

The tax-incidence formula is the clearest case. Students searched five separate phrasings of the same homework formula, things like "tax incidence formula share borne by consumers es/(es+|ed|) per unit tax," and together those phrasings pulled roughly 115 impressions at positions 7 to 10. Read that query again. It contains the actual algebra, the absolute-value bars, the ratio of elasticities. A student copied a formula they did not yet trust and asked the internet to walk them through it. Another one, "linear demand elasticity equals 1 at midpoint total revenue maximized," drew 22 impressions at about position 9.5. Same behavior: the whole result of a graph, typed out as a question.

When students search like this, the page that wins is not the page with the prettiest introduction. It is the page that states the formula and the answer at the top and then shows the work. We watched this happen in our own data. Our step-by-step calculator pages went from 71 impressions across 3 pages to 4,770 impressions across 32 pages in a single week, almost entirely on these formula-shaped queries. The single tax-incidence calculator page alone earned 1,596 impressions in the week at an average position of 9.9. We did not market that page. Students found it because it answered the sentence they had pasted.

The lesson is quiet but firm. If a student is typing the formula, the page has to lead with the formula. Answer first, explanation second. Meeting students where they are means meeting them at the literal string they searched.

2. Where students get stuck is a pair of things that look alike

The second pattern is that a large share of economics confusion is comparison confusion. Students are rarely lost on a single idea in isolation. They are lost on the difference between two ideas that sit next to each other and blur together.

We build pages specifically for these "X vs Y" moments, and they perform. In the same week, 147 comparison pages earned 4,690 impressions. That is a lot of demand for one narrow format, and it tells you that a meaningful slice of studying is really the work of telling two similar things apart.

The strongest single signal we have for this lives on the AI side. On Bing Copilot, the biggest single query our content gets cited for is "microeconomics vs. macroeconomics," with 565 citations, which is 4.82 percent of all the sources the assistant grounded its answers on for that query. Think about what that query is. It is the most foundational distinction in the entire subject, the first fork in the road, and it is also one of the most searched. Students do not master micro and macro as separate silos and then compare them. They meet them as a confusing pair on day one and spend real energy untangling which is which. You can see how we handle exactly that split on the microeconomics vs macroeconomics page.

The practical read for teaching: when you notice a student struggling, it is worth asking not just "what concept is hard" but "what is this concept getting confused with." Substitutes and complements, movement along a curve versus a shift of the curve, nominal versus real, a change in demand versus a change in quantity demanded. The confusion pairs are where the learning actually happens, and they are searched as pairs.

3. AI search is already a discovery channel, misspellings and all

The third pattern would have sounded speculative a year ago and is now just what the data shows: AI assistants are pulling study content directly, citing specific instructional pages, and doing it at volume.

On Bing Copilot, our content holds a 23.03 percent citation share for "how to draw graphs in economics ?", which works out to 120 citations. That is close to a quarter of everything the assistant leaned on to answer that question. And it does not require the student to spell the query correctly. For the misspelled variant "how to daw graphs in economics," our share is actually higher, 28.57 percent, on 72 citations. The assistant reads through the typo, finds the instructional page, and cites it anyway. That is the graph walkthroughs library doing its job in a place we never optimized for.

The pattern holds across the mechanics of the subject. Our citation share is 42.28 percent for "when supply shifts right and demand shift..." (63 citations), 11.46 percent for "supply and demand graph" (102 citations), 15.09 percent for "market structures in economics" (83 citations), and 3.19 percent for "monetary and fiscal policy" (164 citations). The consistent thread is that these are how-to and outcome questions, not definition lookups, and instructional depth is what gets grounded.

There is an honest tension worth naming plainly, because it cuts against our own interest. When an AI assistant cites a page to build its answer, a large share of students may read the answer inside the assistant and never click through to the page at all. Good discovery through AI does not automatically become a site visit. We can see we are being used as a source far more than we can see students arriving. For a teacher, the takeaway is that the study content your students encounter increasingly reaches them pre-digested, assembled by an assistant from sources they never see. Knowing which sources are solid matters more, not less, in that world.

4. What this means for teachers: mechanics beat definitions, depth beats breadth

Put the three patterns together and a fourth, more useful, conclusion falls out. Students are searching for mechanics far more than for definitions. They want to know how to run the formula, how to draw the graph, and what happens when the curve shifts. They are much less interested in being told, in the abstract, what a word means.

Our own data makes this uncomfortably concrete, because we tried both approaches and one of them lost. We have 267 short definition-stub glossary pages. In the week, all 267 of them together earned 3,287 impressions, and they sat at average positions down in the 40s and 50s, which in search terms means page four or five, effectively invisible. Meanwhile 97 long-form explainer guides earned 10,253 impressions at an average position of 15.5 and took the majority of all the clicks the whole site received. Ninety-seven deep pages beat 267 shallow ones, on every measure that matters, by a wide margin.

That is the depth-beats-breadth finding stated as plainly as we can state it. A wall of short definitions does not rank and does not get read. A smaller number of pages that actually work a problem through does both. And the depth pays off fast: new guides entered Google's first page within two days of publishing. One law-of-demand guide was sitting at position 9.1 within 48 hours of going live, which you can see in the law of demand explainer. Students reward the page that does the work, and the search engine follows the students.

So if you are pointing students toward study resources, or building your own, the data argues for fewer, deeper things. Show the mechanism. Work the example. Draw the graph and then shift it. State the formula at the top where a student who pasted their homework will actually see it. Definitions have their place, but the demand, and the learning, is in the mechanics.

A short, honest methodology note

Everything above comes from EconLearn's own Search Console and Bing Webmaster data, plus Bing Copilot AI performance reporting, for July 2026. It is one platform, in one subject, and the reader should weigh it as exactly that.

The Google figures cover a single week, July 8 to 14, 2026. The distinct-query count is reported by Google as roughly 1,000 because that is where Google caps the report, so the true count is a floor, not a ceiling. The Bing and Copilot figures cover a rolling three-month window, which is why the click-through comparison spans longer than a week. On that comparison, one engine asymmetry is worth flagging: Bing's click-through rate on the site ran about 0.99 percent over three months, from 104 clicks on 10,500 impressions, roughly five times Google's 0.2 percent for the same young domain, so behavior clearly differs by engine and none of these numbers should be read as universal.

Sample sizes are stated inline throughout rather than hidden, and some of them are small: 22 impressions on one query, 63 citations on another. This is a description of what one growing study site saw, not a peer-reviewed study, and it makes no claim to represent all students everywhere. It is a real signal from a real corpus of student searches, offered because the raw shape of how students look for economics help is not something most people get to see.

Teachers, writers, and researchers are welcome to cite and reuse these findings. If they are useful in your own writing or your classroom, a link back to this post is all we ask. More of the underlying study material lives in the guide library, the glossary, and the micro and macro course hubs.