Are Listicles Good for SEO? We Measured 2,000 Queries

By Ugur Saritepe · September 10, 2026

Are listicles good for SEO? That is the wrong question, and I have the data to show why. I split 2,000 of my own site's search queries by what their answer demands, at the same average rank. Queries whose answer is a ranked set clicked at 3.41 percent. Yes or no questions clicked at 0.13 percent. The answer's shape decides; the format just follows it.

The numbers that made me ask

My test bed is game-scout.app, a games directory I run and write about openly. Over the 90 days ending 6 September 2026, its articles earned 13.7 percent of the site's impressions but only 6.6 percent of its clicks: article click-through ran 0.94 percent against 2.10 percent for the ranked list pages. I had already measured that supporting articles were not carrying traffic, but two articles broke the pattern, and I wanted to know why before writing any more. Published advice would not settle it: the closest thing to a study on this question, Grow and Convert's client data (updated May 2026), measures rankings, not clicks.

The obvious explanations did not survive contact with the query table. It was not topic, not rank, not age. The split ran straight through what the searcher's question demands as an answer.

One site, 2,000 queries, four answer shapes

Every query Google reported for the site got classed by what its answer looks like, and I held rank constant by only comparing queries at average position 5 to 15, so ranking power is not doing the work. Here is the session where the split landed, run over the site's Search Console data from the assistant:

90-day query and page tables, ~2,000 queries classed by answer shape, position held to the 5-15 band

ranked set (best X, tier list): 693 queries, 16,587 impressions, 3.41% CTR
comparison (2 options):          22 queries,    809 impressions, 1.48% CTR
single number:                    8 queries,    100 impressions, 0.00% CTR
yes/no question:                 39 queries,    795 impressions, 0.13% CTR

A query asking for a ranked set of contested options clicked 26 times better than a yes or no question at the same rank. The comparison and single-number classes are small samples, 22 and 8 queries, so I hold those two loosely. The set class and the yes/no class are not small, and the gap between them is the finding.

Why a yes or no question loses at the same rank

A yes or no question is answered by the results page itself. The searcher reads a title, a snippet, maybe a highlighted answer, and leaves satisfied without clicking anything. Your page appearing at position 8 was part of that answer; it just never got the visit. A ranked set cannot be consumed that way: nobody settles "which cars should I pick" from a snippet, so the click happens.

One hypothesis of mine failed here, and it is worth keeping because it is the advice you will read elsewhere. I guessed that good article topics would show a richer query vocabulary, more distinct ways of asking. The opposite held: the site's It Takes Two article generates the most diverse query set of any page, 59 unique content words across 212 distinct queries, and has the worst click-through, because all 212 queries are restatements of one yes or no question. Volume of distinct queries predicted nothing. How many contested options the answer holds predicted everything.

The two articles that work are secretly lists

The site's two winning articles over that 90-day window are both rankings wearing an article's clothes: a best-cars guide for one racing game earned 170 clicks from 3,800 impressions at 4.47 percent, position 7.2, and a platinum-trophy list earned 23 clicks from 411 impressions at 5.60 percent. The racing-game page is still doing it: in the 28 days ending 10 September 2026 it took 93 clicks from 1,863 impressions at 4.99 percent. Both pages answer an in-game question whose answer is a contested set: which cars, which trophies, in what order.

Here is the part I find funny, written from the inside. Those two pages exist only because I wrote them by hand, against my own automation's rules. When the assistant asked why they worked, this is what I typed:

I think asphalt articles are working because I spend time on it. I decided what to write and I thought about what I need.

The instinct was right but incomplete. The effort mattered because it went into picking a question with a set-shaped answer. And my own article pipeline explicitly forbids ranking-shaped articles, a rule written to stop articles from duplicating the site's catalogue lists. The rule overshot: it banned in-game rankings, the one article format the data says earns clicks, along with the catalogue duplicates it meant to stop. We have not yet changed that rule in our own pipeline; the ban is still in place as I write this.

Run the same split on your own queries

The test transfers to any site with Search Console data. Pull your query table, then class each query by what its answer demands before you look at any other column:

• Ranked set("best", "top", "which X should I"): the searcher needs to weigh options. This is the class worth a page of its own.
• Yes or no("is", "does", "can"): the results page answers it. Fold your answer into an existing page; do not spend a post on it.
• Single number or fact: same story as yes or no, only stronger.
• Comparison of two: middling in our data, 1.48 percent, but from 22 queries; treat it as untested rather than settled.

One honest warning about what your query table cannot do: it only shows demand for pages you already built. Of the 328 impressions on in-game ranking queries in my data, 294 belong to the one such page that exists. Before it existed, that demand was invisible. Absence in Search Console means not visible, never zero, so use the table to test pages you have, not to rule out pages you do not. If a page is collecting impressions without clicks and you want the query-level diagnosis, I walked through that failure case in a separate post on impressions without clicks.

The limits of this measurement, stated plainly: one site, one 90-day window, and the query classes were labeled by the assistant in that session, not by a second rater. The two winning articles are also the two I wrote by hand, so effort and format are tangled in a way the data cannot separate. What the position control does rule out is the easy story that the lists simply rank better. They do not; they click better at the same rank.

This post is part of our AI-assistant SEO series, written from the SEO work we do on our own sites. For a five-minute check of the basics on any site, the free triage report runs eight checks from a URL alone, no account needed.

FAQ

Are listicles good for SEO?

Listicles earn clicks when the query's answer really is a list, and that is the only time they do. Across 2,000 queries on one site, measured at the same average rank, queries asking for a ranked set of options clicked at 3.41 percent while yes or no questions clicked at 0.13 percent. The format is not the lever; the shape of the answer the searcher needs is.

Why does my article rank but get no clicks?

Check what the query demands before blaming the title. If the query is a yes or no question or asks for a single number, the results page usually answers it before anyone clicks: in our data such queries averaged position 8.7 and converted at 0.13 percent, which no title rewrite will fix. Pages targeting queries with many contested options clicked 26 times better at the same rank.

Should I turn my articles into listicles?

Only if the underlying query wants a set of options. Reformatting a one-line answer into ten numbered headings changes nothing, because the searcher still gets the answer on the results page. The useful move is upstream: before writing anything, ask whether the answer is a ranked set, a comparison, a number, or a yes or no, and only the first of those reliably earns the click.

In your AI assistant

Ask your AI about your own site

TrafficTriage plugs into Claude, Cursor, and ChatGPT. Your assistant runs the 8 checks, reads your Search Console data, and can even check your site every morning. Free account, no API key, nothing to install.

Sign up freeSee how it works