Passage Indexing: A New Ranking Algorithm from Google
By: Dileep Thekkethil
August 30, 2022
Table of Contents
Your Google search experience will reach new vistas starting 2021 as the search engine giant has rolled out a ranking factor – Passage Indexing.
The new ranking technology was officially rolled out on February 11, 2021. The official confirmation of the rollout came from Google SearchLiaison Twitter account:
Update: passage ranking launched yesterday afternoon Pacific Time for queries in the US in English. It will come for more countries in English in the near future, then to other countries and languages after that. We'll update this thread as those further launches happen.
— Google SearchLiaison (@searchliaison) February 11, 2021
What is the Passage Indexing Algorithm?
Passage Indexing is a new technology used by Google’s Algorithm that can rank individual passages within a page on search results based on the search query of the users.
Remember the announcement that Google made in September 2019 about BERT? They said BERT will impact 10% of English Language search queries. Now, believe it or not, it actually impacts more than 99% of queries.
Passage Indexing will have a similar ripple effect on the search results page as Google Crawlers will start to understand the relevancy of specific passages within your content.
Though it’s called Passage Indexing, there is no major change that happens in the current indexing process of Google. However, the new technology has to do more with ranking.
So, what this literally means is if you have long-form content written about Off-Page SEO and you have individual passages about each of the off-page SEO strategies, Google will now rank your individual sections of the content that matches the search query.
Since its inception, Google has been moving in the direction of better relevancy and context in the search results. You can now see that the two core ranking factors – Content and Links- are valuable only when supplemented with relevance.
The official Google announcement about the Passage Index says the search algorithms will now understand the relevance of the content within a page at the most granular level.
This means that even though the answer related to the search query is deeply buried within a page, Google can pick that particular passage up for you. The search giant says it’s now capable of finding “that needle-in-a-haystack information you’re looking for.”
Passage Index will impact 7% of search queries worldwide. Since we have BERT’s example right in front of us, the percentage is sure to reach an exponential level within a year’s time.
Passage Indexing Focuses on Ranking Not Display
Let me make it easy for you to understand the concept of Passage Indexing.
First of all, Passage Indexing is not about where Google is going to display the results. It’s an additional ranking factor that enables Google to understand individual passages within the page.
Passage ranking isn't about display. It doesn't cause snippets to somehow get longer. It's about better understanding what a page is about by understanding the context of passages of text, where they can be identified, in *addition* to other ranking factors.
— Danny Sullivan (@dannysullivan) December 29, 2020
So, what this means is that Passage Indexing will be just like the BERT or Link Analysis algorithm. It works in tandem with other ranking factors to bring better results to the users without making much of a change to the overall look and feel of the SERP.
The confusion about Passage Indexing was caused because of a bad example from Google’s end and Danny Sullivan confirmed this in one of his tweet.
That's a bad illustration because it compares a regular snippet with a featured snippet. Any regular listing that becomes a featured snippet right now — without passage ranking — looks like that.
— Danny Sullivan (@dannysullivan) December 29, 2020
Here is an interesting example to understand how Passage Indexing Algorithm works:
Consider the page you want to rank as a book with multiple chapters. Until now Google used to rank the book based on the main topic you have covered.
But with Passage Indexing, Google understands individual chapters within your book. So what this means is that your individual chapters (sections of the page) will show up on results when a highly relevant query is entered on search.
This means you don’t have to do anything as of now to make your pages Passage Indexing-friendly because it’s more of an internal ranking change.
However, structuring your content may make it easier for Google’s Passage Indexing algorithm to better understand the meaning of the text.
So, if you are someone who writes long-form content with multiple sub-headings and if you are not seeing organic traction to these articles Passage Indexing is a boon.
Earlier, pages with granular content failed to rank because the main topic may be extensive.
But with Passage Indexing, such long-form pages can now rank for relevant queries that are contextually related.
So, if you are running an eCommerce website, your product pages may not get the benefit of Passage Indexing because the content is usually to the point.
How the Passage Indexing Algorithm Works?
Passage Indexing will not alter the crawling and indexing process but it will help Google understand the meaning of the passages within the page.
This means Google will not be indexing individual passages independently. Google will show the most appropriate passages in the results based on the relevance and meaning whenever a query is entered.
If you do a quick Google search for a long-tailed question-based search query, you may end up seeing results with a list of websites. But what you want is a specific answer to your question, and Google was not able to deliver before.
However, with the Passage Indexing algorithm, Google search will fetch you the most relevant answer for your query.
Interestingly, the page giving you the answer may have long-form content, but Google just saves your time by showing you the most contextually relevant answer.
Here is the official word from Google about Passage Indexing:
So, for example, let’s say you search for something pretty niche like ‘how can I determine if my house windows are UV glass.’ This is a pretty tricky query, and we get lots of webpages that talk about UV glass and how you need a special film, but none of this really helps the layperson take action. Our new algorithm can zoom right into this one passage on a DIY forum that answers the question. Apparently, you can use the reflection of a flame to tell and ignores the rest of the posts on the page that aren’t quite as helpful.
Is Google Just Going to Index Parts of pages?
Google officials have confirmed that Passage Index is not going to replace the normal page indexing. This means that Google’s crawlers will continue to index pages entirely but, as it does this, it will try understanding the content and the meaning of each passage within the content.
This is not pointing at a shift in the way Google indexes pages. Still, it’s adding another layer to the existing process, which of course, will alter the rankings of pages.
As far as SEOs are concerned, this is big because, rather than an indexing change, the Passage Indexing algorithm will bring in a paradigm shift in rankings and the way results are displayed.
Difference Between Featured Snippet and Passage Indexing
According to Google, a featured snippet result is a passage from a page that has overall topic relevance.
However, Passage Indexing doesn’t consider the page’s overall relevance but just the relevance of passages to the search query.
Google’s Danny Sullivan also tweeted regarding the same saying, “Featured snippets are used for voice search. They are already identified using different systems than passages.”
Featured snippet are used for voice search. They are already identified using different systems than passages.
— Danny Sullivan (@dannysullivan) October 21, 2020
Is Google Using SMITH Language Processing for Passage Indexing?
We have heard about BERT that it’s able to understand long queries both in search and within individual pages. But now, Google seems to have unleashed a Megalodon, which has been codenamed SMITH.
SMITH stands for Siamese Multi-depth Transformer-based Hierarchical (SMITH) Encoder, which is a new natural language processing patent that has filed for.
This new language processing model is aimed at making the Google Algorithms understand passages within a page.
So, now we know Google’s roadmap to how Passage Indexing will work in the near future.
In the document share by Google about SMITH it says, the language processing system will help in recommending news articles, related articles, and importantly clustering documents.
The third point about document clustering is an important aspect as it’s directly related to Passage Indexing.
It also says, SMITH is a long-form document matching system, which again points to the Search On announcement made by Google and the tweets from Danny Sullivan about how long-form content will get benefited from Passage Indexing.
How this works is almost similar to BERT. While BERT uses masked word language modeling, SMITH ups this by masking sentence blocks.
According to Google, their “experimental results on several benchmark datasets for long-form document matching show that our proposed SMITH model outperforms the previous state-of-the-art models including hierarchical attention, multi-depth attention-based hierarchical recurrent neural network, and BERT.”
When you compare SMITH to BERT, the former has the ability to process more words, which according to Google helps in increasing the ability to match documents.
“Comparing to BERT based baselines, our model is able to increase maximum input text length from 512 to 2048,” says the Google patent. It has to be noted that the maximum input text length for BERT is 512 words.
This is how SMITH is put into action:
Step 1: A document is split into several sentence blocks
Step 2: Language Processing Transformers will learn the contextual representations of each sentence block.
Step 3: The whole sentence block is then contextually represented by following the practice in BERT.
Step 4: Given a sequence of sentence block representation, the document level Transformers learn the contextual representation for each sentence block and the final document representation.
Speculations: Are We Entering Content Optimization 2.0?
Passage Indexing will change the way Google is going to show search results, and it seems like there is a lot to do for SEOs in the days to come.
Google considers meta titles and heading tags as essential signals to understand the context of the content. But that was until they launched the Passage Indexing.
With Passage Indexing, Google has started indexing pages by understanding the meaning of passages independently. As a result, individual passages can now rank for related search queries.
What’s more interesting is that even if the page discusses a less relevant topic, if the answer to the user’s query is buried inside any of the passages, the page will show up on the search.
Looking at various discussions on Passage Indexing with Googlers, it’s evident that Passage Indexing will have bigger implications.
For example, during an Office Hour discussion John Mueller, the later said, the results of Passage Indexing may now appear in the featured snippet area but over time passage indexing will determine normal search results.
So it might be that we start showing these in the featured snippets first because I don’t know we showed that example or maybe that’s the clearest way we can check this. And then at some point we start showing them more in the normal search results as well.
This is a big statement as we know that Google has been downplaying meta description for quite some time and there is a high chance that Passage Indexing will take control of what appears in the meta description moving forward.
Adding to this, I personally don’t think that the character limit of the meta description will be increased to accommodate the whole passage. Instead what Google might do is to extend the scroll to text feature that’s already implemented on featured snippet results to meta description.
This way, the user can navigate to the exact passage within a page that answers their question. But if you are relying on ad revenue, this feature is going to be a bane than a boon as Google will auto scroll to the exact section which means the users might miss seeing your ads.
Another reason why I see this coming into reality soon is that Google has a history of testing SERP features on a tiny scale before they roll it out on a larger scale.
The current example of passage indexing results appearing in place of the featured snippet and the scroll to text feature can be touted as part of an analysis to understand how useful they are for users.
John’s answer to Glenn Gabe’s question on whether Passage Indexing will just result in better answers for featured snippets is a revelation in itself about how important Passage Indexing is for Google.
Here is what John replied:
So my kind of taking a step back and just guessing at this, with my internal information. Usually what happens with these things is we will roll them out in one particular place, experiment a bit to find out how to best implement these, how they best work, and then find ways to roll that out a little bit more broadly.
But again kind of like with all of these newer changes in search. Usually, we try them on a small scale and then roll them out a little bit larger over time.
Even if Google is going to restrict Passage Indexing to the featured snippet area or not, you may find an increase in the number of zero-click queries.
This happens because users are getting the best contextually relevant information on the SERP and there is no need for them to visit the website that has curated the content.
So, once the Passage Indexing is made live, keep a close watch on the impressions you receive and the clicks. Also, there is a high chance that the Search Console will add an Enhancement Feature for Passage Indexing and showcase the click generated out of the new feature.
What this also means is that SEOs can now stop focusing on keywords and give more impetus to the topic-relevancy and conclusiveness.
Understanding the user’s concerns and addressing them with content solutions will play a deep role in the success of websites after the roll out of Passage Index.
What I mean to say is long-form content that takes a holistic approach on one particular topic stand a chance to benefit from the new Passage Indexing Algorithm.
Thanks to Google’s Natural Language Processing Algorithm – BERT, understanding the meaning and relevance of each passage is no more a troublesome task for the search engine giant.
Sometimes if you look at the algorithms and features that Google comes up with you can see synchrony in the way they work and how they support in achieving Google’s pristine goal of providing users the best possible results for a given search query.