Yes, with a caveat worth keeping. Google's own documentation still lists RankBrain as one of its ranking systems, reviewed as recently as October 2026. What Google does not support is the familiar claim that RankBrain is still the third-most-important ranking factor, or that you can optimize for a separate RankBrain score. Evidence grade: confirmed for RankBrain ranking factor status in current public documentation, unknown for its exact weight or implementation today.
That distinction matters if you still write briefs, audits, or client reports that lean on the old "top three" framing. A lot of teams searching "is RankBrain still used" or typing "RankBrain 2026" into a search bar are really asking whether a decade-old talking point is safe to keep repeating. The short answer is you can keep the system in your explanations. You just need to retire the precision that was never really confirmed in the first place.
What did Google originally say about RankBrain?
Google introduced RankBrain publicly on October 26, 2015. Bloomberg broke the story that day, with comments attributed to Google's Greg Corrado, and Danny Sullivan covered it the same day for Search Engine Land in a piece called "Meet RankBrain." The early reporting described a machine learning system built to handle search terms Google had not seen before, by connecting them to words and phrases with a similar meaning.
The claim that stuck was that RankBrain had become the third-most-important signal in a search result, after only a few months of use. That single line did more to shape how marketers talk about RankBrain than almost anything written since. It gave the industry a tidy spot in a hierarchy, which is a much easier thing to repeat than "a component that helps interpret unfamiliar queries."
Five months later, in March 2016, the story got its familiar shape. Barry Schwartz reported on a Q&A with Google's Andrey Lipattsev, in which content and links were named alongside RankBrain as the other leading ranking factors. The industry turned that into its own shorthand: top two spots, RankBrain in third. The transcript, published by Search Engine Watch, tells a messier story than the shorthand suggests. Lipattsev said there was "no order" between content and links. He called third place "a hotly contested issue." He told the room to take the whole framing "with a grain of salt," and he specifically rejected the idea that links, keywords, and RankBrain could each be assigned a value and multiplied together into a ranking score.
A few months after that, at SMX Advanced in June 2016, Google's Gary Illyes called RankBrain a ranking factor, part of machine learning, and a new signal, which at least confirms the term wasn't purely an industry invention. His advice at the same event was restrained: write in natural language and don't stuff your content with keywords. That's a sentence you can still act on today, which says something about how little has actually changed in the underlying guidance even as the terminology around it has shifted.
What evidence says RankBrain is still used?
The strongest evidence sits in two places, and it's worth using them together rather than leaning on just one.
The first is Google's current ranking systems guide, last reviewed for this piece on October 26, 2026 and displaying an update date of December 10, 2025. RankBrain has its own entry in the main list of systems, and it is not filed under the separate retired systems section. Google describes it as an AI system that helps the company understand how words relate to concepts, so it can surface relevant content even when that content doesn't contain the exact words someone searched.
The second is a more direct statement from Pandu Nayak, a Google Fellow and Vice President of Search, in a February 2022 piece called "How AI powers great search results." Nayak calls RankBrain Google's first deployed deep learning system in Search, launched in 2015, and says plainly that it helps decide the order of top search results. He also says it continued to be one of the major AI systems powering Search at the time he wrote that. Taken together, you have an explicit functional explanation from 2022 and a current public listing from late 2025 into 2026. Neither one alone tells the whole story, but the two together make a reasonable case that this is a named system Google still runs, not a historical footnote.
Nayak's piece includes an example that's genuinely useful here. Google uses the query asking for the name of the consumer at the highest level of a food chain. RankBrain connects that wording to the concept of an apex predator, rather than reading "consumer" as a shopping term. It's a clean illustration of word-to-concept matching, and it's a good one to keep in mind the next time someone asks what RankBrain actually does.
What evidence challenges the standalone-factor framing?
Here's where the caution comes in. Google's current guide describes a system built around understanding meaning, not a page-level number you can check in an audit tool. Nowhere in the reviewed documentation is there a current numerical weight for RankBrain, a present-day third-place ranking, or anything resembling a published optimization score.
Even the 2016 transcript that gave us the "top three" framing already warned against treating it that simply. And a year later, Danny Sullivan reported that RankBrain was involved in every query but affected the actual ranking of many queries rather than necessarily all of them. That's a distinction worth sitting with: being part of how Google processes a query is not the same as changing the final order of results for that query every single time.
In 2019, Gary Illyes gave a Reddit AMA that added more texture. He described RankBrain as using historical search data, sometimes months old, to predict what a person would likely want to click for a query Google hadn't seen before. He also drew a line between data about what happens on the results page and data about what happens after someone lands on your site. That line matters because it rules out two opposite oversimplifications: RankBrain isn't just word matching, and it also isn't a live dwell-time or click meter watching your page in real time.
None of this amounts to evidence that RankBrain stopped mattering. It's evidence that the version of the story most people repeat was always more precise than the facts supported.
It's also worth naming what doesn't count as evidence either way, since people reach for shortcuts in both directions. A newer AI system launching isn't proof that RankBrain got retired. A contemporary SEO article skipping the word "RankBrain" isn't proof of anything either, since plenty of writers just assume it's old news and move on. And Google using the word "system" in one document while an older interview used "signal" isn't a sign that the underlying thing changed, since both words have been used by Google representatives to describe the same component at different points. The listing in current documentation is a more relevant piece of evidence than any of those absences.
What do Google's primary sources actually establish?
It helps to separate three claims that tend to get collapsed into one:
- RankBrain is a named Google system involved in ranking. This is confirmed, by the current guide and by Nayak's explicit description.
- RankBrain remains the third-most-important Google ranking factor. This is a historical claim from 2015 and 2016 reporting, and nothing in the current documentation reestablishes it for 2026.
- You can isolate and improve a separate RankBrain score for a page. This isn't established anywhere in the record, and no credible source has ever published a method for doing it.
Also worth clearing up: in November 2022, Google introduced a centralized guide to its ranking systems, announced by Danny Sullivan on the Google Search Central blog. That announcement is sometimes read as Google formally reclassifying things from "factors" into "systems." It isn't that. It's a documentation project aimed at being clearer about the difference between an ongoing system and an update to that system. It says nothing about RankBrain's current importance, and it isn't a retirement notice for anything.
So when you're deciding what to write in a brief or an audit, the accurate version is closer to "RankBrain is one of Google's AI ranking systems" than "RankBrain is a standalone signal with a known weight." Both can technically use the word "factor" in a broad sense, which is why calling it a RankBrain ranking factor is defensible as long as you mean something involved in ranking rather than a tunable score. Only one framing survives contact with what Google has actually published, and it's worth being specific about which one you're using when you write it down.
It's also worth being precise about dates when you cite any of this. The 2015 launch, the 2016 "top three" reporting, the 2019 AMA, and the 2022 Google explanation each answer a different question at a different point in time. Treating them as one continuous, current technical specification is how the overreach happens in the first place. If a brief or deck cites RankBrain without a date attached to the claim, that's usually a sign the claim needs a second look before it goes out the door.
Should marketers still call RankBrain a ranking factor?
You can, as long as you mean it in the broad sense: something involved in deciding how results are ordered, not a separately measurable property of a page. Here's the decision rule in practice.
Before: "RankBrain is Google's third-biggest ranking factor, so we need to optimize for it directly."
After: "Google still lists RankBrain as an AI ranking system that helps it connect words with concepts, so write for the concept and the intent behind a search, not just the exact phrase."
That second version keeps everything the evidence actually supports and drops the part that was never confirmed. It also happens to point at something genuinely useful: because RankBrain's job is connecting related wording to the same underlying idea, a page that answers the real question behind a search has more going for it than one that just repeats the exact keyword. That's not a loophole or a trick. It's permission to write clearly about what someone actually wants to know, instead of writing around a phrase.
What you shouldn't do is commission a separate "RankBrain optimization" workstream. There's no published score to chase and no confirmed formula to reverse engineer. Illyes's 2016 advice to write naturally and skip the keyword stuffing still holds up, but it's general writing advice, not a special RankBrain lever.
This verdict also has a boundary worth stating plainly. RankBrain is a Google ranking system, and Google's own guidance on AI Overviews and AI Mode says foundational SEO practices carry over to those features without special optimizations. Neither of those sources ties RankBrain to citation selection in ChatGPT, Perplexity, or any other company's answer engine. Keep the two questions separate: what Google's ranking system does, and what gets a page cited somewhere else.
What would change this verdict?
A few things would genuinely move the needle here, and it's worth keeping an eye out for them rather than assuming the current status is permanent. A dated Google statement that RankBrain has been retired or folded into something else would do it. So would a documentation change that moves RankBrain into the retired systems section with an explanation attached. A new primary source describing a different current role, or a fresh statement about relative importance or query coverage, would also warrant an update.
What wouldn't be enough on its own: a newer AI system launching, a blog post that doesn't happen to mention RankBrain, or an SEO checklist that's dropped it from the list. Silence isn't a retirement announcement, and a page disappearing from documentation without explanation is a reason to look closer, not a reason to declare anything dead.
If you're reviewing old briefs or training materials, this is the moment to go find every place "RankBrain" shows up next to an unqualified "third most important" and replace it with the documented description instead. It's a small edit, but it's the difference between a sentence you can defend and one you're just repeating.
This is also where it's worth separating what you observed from what you can prove caused it. If a page's rankings move after a rewrite aimed at intent rather than exact wording, that's a real, observable change. Labeling it a "RankBrain effect" is a step further than the evidence supports, since nothing in the public record gives you a way to attribute a specific ranking change to that one system rather than to the dozens of others working alongside it. It's fine to say the rewrite helped. It's a stretch to say exactly which part of Google's machinery is responsible.
None of this should turn into an excuse to stop writing clearly, either. The practical lesson from RankBrain's word-to-concept role isn't "keywords don't matter," it's "answer the real question behind the search, in language a person would actually use." That's a useful standard whether or not you ever mention RankBrain by name again.
