RankBrain is the code name behind Google’s Artificial Intelligence (AI) program.
In Oct 2015, Google rolled out an algorithm update based on the Rankbrain program. This update was particularly earth-shattering at the time it was released, raising a lot of speculations/misconceptions that are worth being clarified. A year later, we can look back in retrospective and see if indeed this was something to be feared.
What was RankBrain all about?
Rankbrain was designed to improve search results for ambiguous or poorly worded long-tail queries.
Greg Corrado, a senior research scientist at Google, was the first to reveal existence of the RankBrain program in an interview to Bloomberg in October 2015.
From the story:
“RankBrain uses artificial intelligence to embed vast amounts of written language into mathematical entities — called vectors — that the computer can understand. If RankBrain sees a word or phrase it isn’t familiar with, the machine can make a guess as to what words or phrases might have a similar meaning and filter the result accordingly, making it more effective at handling never-before-seen search queries.”
Did RankBrain impact Search Results at all?
Here comes the main confusion that was initiated by Gary Illes, Google Webmaster Search Analyst, via the following tweet:
Lemme try one last time: Rankbrain lets us understand queries better. No affect on crawling nor indexing or replace anything in ranking— Gary Illyes (@methode) 18 mars 2016
RankBrain had an impact on ranking for a certain portion of queries reshaping the search pages; in that sense Gary’s tweet was misleading.
What volume of queries were affected?
RankBrain wasn’t supposed to tackle all searched terms, rather only those misunderstood by Google’s core algorithm. Those are mainly the “never seen” queries (accounting for 15% total queries).
Many SEOs misinterpreted this figure, claiming 15% of ALL queries were affected by Rankbrain, which doesn’t seem to be the reality. Lacking official figures from Google, let’s dig into research available at this stage:
Earlier this year, Stonetemple, a Boston based Marketing Agency, ran an in depth study to assess RankBrain’s impact on Search results. The study sampled 1.4 million queries. The data was recorded pre-Rankbrain in order to detect particular queries that would be treated differently pre VS post-Rankbrain.
At the end of the screening process, 163 queries (out of 1.4 Million initial dataset ~0.0001%) matched the following criterias:
- The search results indicated that Google didn’t understand the initial query.
- A reasonable set of results from Google index was found for the query.
The last step was checking how Google Algorithm post-rankbrain would handle these 163 queries. Stonetemple noticed a clear improvement with fully relevant results returned for 89 queries, i.e 54.6% success rate, which is quite impressive.
An interesting example from the study was with ‘why are pdfs so weak’ phrase:
Image credit: stonetemple.com
Looking into January’s SERPs (right screenshot), it seems Rankbrain succeeded to interpret the ‘weak‘ adjective in the context of security encryption and returned results accordingly.
What conclusions can be drawn from Stonetemple study?
- The volume of queries requiring Rankbrain’s intervention is likely to be very tiny (0.00001% in Stonetemple study)
- When Rankbrain steps in for out-of-ordinary query its success is likely to be high (54.6%)
Of course, those conclusions should be taken cautiously. We can’t eliminate the possibility of the impact of additional algorithm updates (rolled out in conjunction to Rankbrain’s update) that might have mislead Stonetemple’s conclusions.
Can we Optimize for RankBrain?
Unfortunately, from an SEO perspective, optimizing for Rankbrain remains a vain wish.
RankBrain acts as a kind of translator that pre-processes certain queries prior to letting the regular search algorithm take the reins. Your website’s on-page factors aren’t screened by Rankbrain.
The following infographic illustrates where Rankbrain is supposed to come into play.
Image credit: Damian Fanish
Covering a maximum of longtails variations in your content is still part of the fundamentals to achieve good organic performance regardless to Rankbrain mechanisms.
Rankbrain is likely to bring solid improvements in several fields:
- Improved tolerance to complex long-tail keywords. Users searching for unusual and complex queries are likely to find satisfactory search results quicker (without need to run consecutive queries)
- A better ability to handle Voice Searches – Voice search technologies like Siri, Google Now, Cortana should return more accurate results thanks to Rankbrain.
A look forward to AI future
AI and machine learning will keep playing a key role in enhancing Google products across the board.
Sundar Pichai (recently appointed Google’s new CEO) affirmed in several occasions that Google engineers are working on different solutions involving AI, as he said at one of the recent Google financial earnings: “Machine learning is a core transformative way by which we are rethinking everything we are doing”.
Let’s stay tuned and see what the future holds for us!
If you have a question or a business inquires, feel free to contact us through our contact form.back to blog