TL;DR
- A study of 5.32 million organic result rows published on 24 August 2026 by Kevin Indig in his Growth Memo newsletter finds Google trimmed listicles from its highest positions without removing them from the page.
- The dataset covers 60,000 United States English-language desktop queries across 15 verticals, collected with SE Ranking data over a 47-hour window in August 2026.
- Query phrasing is the strongest predictor: listicles reached the top three for 54.5% of explicit ideas, examples, options or best queries, against 10.2% for implicit category or comparison queries.
- Position-one listicles fell from a 18.9% to 20.2% range in January to 15.1% to 15.6% in August, and the top three from 35.9% to 39.0% down to 32.0% to 33.0%.
- Indig calls it almost 4% fewer listicles, but the top-three decline runs from 3.9 to 6.0 percentage points depending on the bounds compared, and the memo does not reconcile the two.
A study of 5.32 million organic result rows, published on 24 August 2026, finds that Google trimmed listicles out of its highest positions without removing them from the results page. Growth advisor Kevin Indig published the analysis in his Growth Memo newsletter that Monday, describing it as the first of two instalments, and PPC Land covered the findings.
The headline that matters for a content team is not the decline. It is the predictor. On this data, the strongest signal of whether Google shows a listicle at all is how the searcher phrased the query, not whether Google prefers the format.
What the dataset covers
The study drew on 60,000 United States English-language desktop Google queries across 15 verticals, collected with SE Ranking data during a 47-hour window in August 2026. That produced the 5.32 million organic result rows the analysis rests on. SE Ranking is a software vendor supplying the rank data, and Indig is an independent growth advisor writing in his own newsletter.
Every result was classified automatically as a listicle or not, using its title, snippet, URL and website type. The classification separated numbered list articles from unnumbered editorial roundups, and it excluded product grids, directories, forums, video pages, single-item reviews and step-by-step guides. Indig's working definition is that a listicle is an article in which the list is the content: discrete items of the same type, comparable with one another, each getting its own self-contained treatment, under a title that promises a bounded set. A sequential tutorial does not qualify, because step three depends on step two. Listicle items can be reordered without breaking the logic.
Two limits on the method are worth stating up front, because they change how firmly the numbers should be held. The classifier was automatic rather than hand-labelled, though the initial pilot reached 100% precision and recall on 30 queries and a blind quality review covered 30 pages. And the memo states that an untouched production holdout, the final validation step, has not yet been run.
Listicles still fill half the top 10
The raw counts do not support a universal penalty. At least one listicle appeared in the organic top 10 for 55.1% of queries in the sample, and in the top three for 32.3%.
A looser signal ran much higher. On 85.9% of results pages, at least one title carried a broad list cue, meaning a one-to-three-digit number or a word such as best, top, ways, ideas, examples, alternatives, tips, tools, apps or options. The stricter classification, which requires the page structure to match the definition rather than the title alone, brought that back to 55.1%.
The 30-point gap between those two measures is itself a finding. Roughly a third of pages that read like listicles in the results snippet turn out to be something else once opened. Anyone auditing a competitor set by scanning titles is over-counting the format by a wide margin.
Query wording is the real predictor
Google returned 4.5 times more listicles when users explicitly asked for a set of options. The split by query type is the most actionable number in the study.
| Query type | Listicle in the top three |
|---|---|
| Explicitly asks for ideas, examples, options or the best items | 54.5% |
| Implicit category or comparison queries, where the demand for options is only inferred | 10.2% |
| Informational comparison queries that did not ask for a list | 12.1% |
Read that as a response to phrasing rather than a format preference. A page structured as a numbered list is well matched to a query that asks for a set and poorly matched to a query that asks a question. The gap between 54.5% and 10.2% is far larger than the decline the same study measures at the top of the page.
The January to August comparison
To test the demotion claim directly, Indig held a fixed subset of 2,400 queries and 24,000 results constant and compared January against August. The first thing that subset shows is churn unrelated to format. Only 52.5% of queries returned the same exact URL at position one in both periods, and only 58.3% returned the same domain. Half the top slots changed hands regardless of what kind of page occupied them.
Within that churn, listicles did lose ground at the very top.
| Position band | January 2026 | August 2026 |
|---|---|---|
| Position one | 18.9% to 20.2% | 15.1% to 15.6% |
| Top three | 35.9% to 39.0% | 32.0% to 33.0% |
Indig characterises this as almost 4% fewer listicles at position one and in the top three, and attributes the effect to an unannounced mid-January update along with quality and velocity signals in recent spam work. He writes that Google has systematically and carefully weeded out certain listicles.
The word almost is doing real work there. That description sits at the conservative end of what the ranges allow. Depending on which bounds are compared, the top-three decline runs from 3.9 to 6.0 percentage points, and PPC Land notes that the memo does not reconcile the two figures. Anyone repeating a single number from this study should say which end of the range they are quoting.
Vendor listicles were not the ones that fell
The prevailing account through the first half of 2026 was that the January movement targeted self-promoting listicles, the format in which a software vendor publishes a ranking of its own category and installs itself at number one. The dataset does not confirm that. Of vendor-authored listicles, 62% gained estimated organic traffic since January. Indig states plainly that he cannot trace the observed decline back to self-promoting listicles specifically.
That sits alongside earlier research the memo cites rather than replicates. Analysis covered by PPC Land in March 2026 found that sites built on self-serving listicles lost up to 49% of visibility after the January ranking fluctuations, with all identified cases drawn from the SaaS sector and sharing traits including scaled output, unedited AI text, artificial freshness updates and schema misuse. Both statements can hold if the January action was narrow enough to hit a specific cluster of publishers while leaving the format alone, which is the reading the memo offers.
A second open contradiction sits in the generative layer. Peec data referenced in the memo shows no demotion of self-promoting listicles in citations across ChatGPT, Google AI, Perplexity, Copilot, Gemini and AI Overviews. Whatever filter the organic results applied has not been applied to AI citations on that evidence.
Who is supplying the listicles
Almost half the listicles in the sample came from publishers rather than from known brands or vendors. In 4,026 direct matchups where a publisher listicle and a brand or vendor listicle appeared on the same results page, the publisher won 54.0%. Across the top five results in those matchups, the publisher averaged position 4.24 against 4.71 for the brand or vendor. Conditional rank, measured when a domain type placed a listicle in the top 10 at all, followed the same order: publishing and content platforms averaged 3.76, publishers 3.99, government, academic and nonprofit sites 4.48, and brands and vendors 4.65.
The supplier base is wide. Some 5,765 domains supplied top-10 listicle leaders, with the top 10 domains accounting for 18.3% of those placements and the top 50 for 36.5%, leaving 63.5% spread across the remaining 5,715 domains. On this measure concentration is low, which is a useful corrective to the assumption that a handful of large sites own the format.
The vertical spread is where B2B stands out
Listicle occurrence in the organic top 10 ranged from 42.0% in entertainment and gaming to 67.2% in beauty and fashion. B2B software reached 62.5% and travel reached 49.3%. Brand and vendor listicles varied far more sharply than the overall figure: they appeared in the top 10 on 46.2% of B2B software results pages, against 3.3% in food and 6.2% in travel.
Indig reads that concentration as evidence that self-published category rankings are a business-to-business tactic almost exclusively, deployed where the buyer researches through search and the vendor has the content operation to sustain it, and that the lever has limited application in consumer categories.
What this means for Thai marketers
The study covers United States English-language desktop queries. It says nothing about Thai-language results, nothing about mobile, and nothing about any market outside the United States. Nobody should present these percentages as a measurement of Thai SERPs, and the sensible use of the study here is as a hypothesis to test against your own data rather than a finding to import.
What does travel across markets is the mechanism, and the mechanism is about phrasing. If Thai searchers ask for a set of options with words that map to best, top, ideas or examples, a list page is a sensible answer format. If they ask a comparison or category question without asking for a set, a list page is answering a question nobody asked. That logic does not depend on the language of the query, and it is testable in Search Console this week.
Two practical consequences follow for a Thai content team. First, the Thai market skews heavily to mobile, and this dataset is desktop only, so the numbers do not transfer even in direction without checking. Second, the B2B concentration in the data is a reminder that self-published category rankings are a narrow tactic, and a Thai consumer brand copying a SaaS vendor's format is copying something the data shows barely appears in food or travel results at all. A content programme built on format imitation is the weak version of this.
What a content team should actually do
This part is reasoning, not reporting. The study measures what Google shows; it does not prescribe what you publish. But the query-phrasing finding points at a straightforward audit.
- Pull your existing list pages and split the queries they rank for into two buckets: those that explicitly ask for a set, and those that ask a category or comparison question.
- For pages ranking mostly on explicit list phrasing, the format is doing its job. Leave it.
- For pages ranking mostly on implicit category or comparison phrasing, the list structure is a mismatch with the demand, and the study says listicles reach the top three for those queries at a rate of 10.2%.
- Stop defaulting to a numbered headline for every commercial page. Match the format to how the query is actually phrased.
- Check position-one turnover before blaming a format. Only 52.5% of queries in the fixed subset kept the same URL at position one between January and August, so a page that slipped may have slipped for reasons unrelated to being a list.
None of this argues for deleting list content. Listicles held the organic top 10 for 55.1% of queries in August, which is not the profile of a format Google has removed. The honest summary is that the top of the page got harder for lists while the page as a whole did not. Teams doing serious organic search work should treat that as a targeting question.
Frequently asked questions
Did Google penalise listicles?
The study does not support that framing. It measures a decline at position one and in the top three between January and August 2026, while at least one listicle still appeared in the organic top 10 for 55.1% of queries, so the format was trimmed at the top rather than removed from the results page.
Does this apply to Thai-language search results?
The study did not measure them. The dataset covers 60,000 United States English-language desktop queries, so there is no basis in it for any statement about Thai results, Thai-language pages or mobile search.
Should we stop publishing list posts?
Not on this evidence. Listicles reached the top three for 54.5% of queries that explicitly asked for ideas, examples, options or the best items, and for 10.2% of implicit category or comparison queries, so the useful decision is which queries get a list rather than whether to publish lists at all.
How reliable is the study?
It is one dataset with stated limits. The listicle classifier was automatic, the memo is the first of two instalments, and the memo states that an untouched production holdout, the final validation step, has not yet been run.
Why do people quote different decline figures?
Because the study reports ranges rather than single values. Indig describes almost 4% fewer listicles at position one and in the top three, while the top-three decline runs from 3.9 to 6.0 percentage points depending on which bounds are compared, and PPC Land notes that the memo does not reconcile the two figures. The full write-up is at PPC Land.
Where to take this next
The audit above takes an afternoon and needs nothing beyond Search Console and your own page list. If it turns up a set of list pages ranking on question phrasing rather than list phrasing, that is a rewrite queue with evidence behind it. If you would rather have that mapped against how your pages are performing in AI search results as well as the classic ten blue links, that is a reasonable place to start a conversation.
