How Google uses engagement metrics for search rankings
How Google uses engagements metrics to rank pages based on user satisfaction and has done so for over a decade
It’s a tale as old as time.
A global search engine with complete market hegemony and a chequered past when it comes to being truthful has been disingenuous with the people who make it so.
And breathe…
TL;DR
Google has used engagement metrics to help evaluate page quality and ranking for over a decade. Patents from the mid-2010s and the recent Google Leak confirm this
Despite Google's official stance, experiments like Rand Fishkin's 2015 CTR test showed a direct correlation between engagement and ranking changes
Google's Chrome browser collects vast amounts of user behaviour data. This data is crucial for detecting anomalies and refining ranking algorithms
CTR, good vs. bad clicks, dwell time, and trending data play key roles in assessing user engagement
Does Google use engagement metrics to determine rank?
Yes, Google uses engagement metrics to help rank content effectively. It’s one of the aspects that truly moves the dial in SEO and has long been discussed and hypothesised by the SEO community that engagement metrics like click-through rate, click quality, bounce rate and dwell time can affect rankings.
Recently, Google has vehemently denied the use of engagement metrics. However, many former engineers and search leads at the search giant have confirmed their existence. If I had my tinfoil hat on, I’d say that the closer the SEO community got to establishing the use of engagement metrics, the more vehemently Google denied it.
Why?
Despite Google’s insatiable appetite for content, it cannot truly delineate between good and bad content. There’s a human element that an algorithm cannot deliver. At least not yet.
The world’s best and most effective AI models and LLMs can understand the differences and connections between entities. And like a chess computer, it can predict the ‘best’ next word based on the billions of articles that have come before it.
To bridge the gap between understanding connections between words and phrases Google has long used links as a source of trust. It’s why the rise of E-E-A-T has been important. It helps Google understand whether the wider industry trusts the content.
But it still can’t determine quality content like a user can. So if you have all of that user data at your very fingertips - good clicks, bounce rates, time on page, page interactions etc - wouldn’t that seem like a good place to start?
Google’s former engineer certainly thinks so…
What does Google say about it?
For around a decade, Google has denied the existence of engagement metrics playing a part in how they rank content. I don’t know why. I suppose they have tried to keep some parts of their algorithm closeted.
In 2015, Webmaster Trends Analyst John Mueller said “I don’t think we even see what people are doing on your website” and “So from my point of view, that is not something I’d really treat as a ranking factor.”
Interestingly in the same year, Google filed for several patents around user engagement. Most notably this patent on modifying search result ranking based on a temporal element of user feedback.
“…Obtaining user feedback associated with the quality of an electronic document; adjusting a measure of relevance for the electronic document based on a temporal element of the user feedback; and outputting the measure of relevance to a ranking engine for ranking of search results/.”
Google Patent, 2015, Modifying Ranking Based on User Feedback
Now, I’m not a detective, but the clues were there…
Then in 2022, Mueller again claimed that Google doesn’t use ‘engagement’ as a ranking factor. He even went as far as saying Google Search doesn’t know if your web page has a high or low level of user engagement.
We all thought Rand Fishkin had thrown the chessboard up in the air when he experimented with CTR as a ranking factor in 2014. Rand asked his Twitter followers to google a specific term (IMEC Lab) and visit his website from the SERPs. After a few hours, his website was ranked first for that given query, gaining six positions as a result.
Gary Illyses denied this by saying that clicks were “too noisy a signal” to be used as a ranking factor. Not only did Gary deny it, but he gave Rand the old one-two at a conference, naming his and his ‘experiments’ directly. If it wasn’t for people like Rand in the community, Google and SEO wouldn’t be what it is. Whether Google likes it or not, they need us.
Although Gary admitted that CTR was being used for personalisation. A modicum of comfort to poor Rand. Particularly as multiple former Google engineers (and even Senior VPs) admitted the existence of click data and user interaction.
The idea you would have access to all of this data and not use it to help determine the quality and value of content is not only absurd. But stupid. Intensely stupid. Stupid on a level that doesn’t compute with the world’s most successful company. Complete with the highest paid nerds of all time.
This article on Google and engagement by Arthur Andreyev provides really excellent background to the story. Highly recommended reading.
How the Google Leak contradicts this
Ah, the Google Leak. One of the great reveals of our time. A beautiful counter punch from the SEO community reminiscent of Mayweather’s pull and right hand to the nose. Particularly for those like Rand who had put their neck on the line for this.
Whether you believe it in totality or think the timing was quite convenient, it doesn’t matter. What matters is we were given a real look at 14,000 factors that influence Google Rankings.
If you want a better look at these ranking factors, this searchable database by GrowthSRC is absolutely incredible. Take the estimated weight with a pinch of salt of course. But for me, this has changed the game.
You can now see how prevalent things like CTR, good clicks and bad clicks are for ranking a page. These metrics directly influence a page’s ability to rank by indicating user satisfaction. Something we’ve long theorised but have continuously been shut down.
Well, since about 2014.
The search database even comes with a custom GPT that lets you search through the entire Google Leak database. One of the best ways to help you make evidence-based decisions.
Chrome’s role in data collection
In the URL bar, type chrome://histograms and press enter. Look at the sheer quantity of data collected from your browser startup to the previous page load. It’s jaw-dropping.
These thousands of data points are stored in histogram form. A format constructed by bucketing a range of values and dividing the range into a series of intervals. These intervals help analysts (or algorithms in a more modern, nefarious world) estimate the probability and value of each variable.
In all likelihood this is how Google understands and utilises;
User behaviour insights that shed light on how users interact with a page at a query and URL level
Performance metrics and their impact on positive user behaviour (also as a way of training their sandbox model)
Anomaly detection. Imagine a page with limited HTML on it, like a calculator. If a sudden change caused a dramatic drop in performance and user behaviour crashed, that’s an immensely valuable signal
I would highly recommend watching this video by Cundy Krum. It’s an incredible exploration of how Google may leverage Chrome data for indexing, the resource-intensive nature of executing JavaScript, and the impact this has on the search ecosystem. Long story short, your browser does the work so Google doesn’t have to. At least for the second round of indexing.
Google’s ability to store and use Chrome data has created this monopoly. One that the recent DOJ case might break up.
I only recently found this out, but you can filter the histogram data collected by Chrome. For example, chrome://histograms/UserActions shows all user action related metrics.
What engagement metrics does Google use to inform search rankings?
Many. Too many to go through in a single article. Particularly when the author loves the sound of his own keyboard. But using the Search Ranking Factors dashboard and my brain, I think the below play key roles at each stage of the search journey.
CTR
CTR measures the percentage of users who click on a link after seeing it in search results. For any who work in the news world, you should know how important this is in the Top Stories block. Immediate engagement is crucial. Your headline and image are everything.
For more traditional SEOs, it’s why shorter URLs, dates in title tags and in your structured data and a good quality meta description have a very real impact on page performance.
How can I influence it?
Title tags
Meta descriptions
Featured images (optimise across multiple platforms)
Brand building
Good vs bad clicks
Once a user has clicked through to your site, stored Chrome data lets Google establish how satiating each page is at a query level. Ranking pages, particularly for broad queries is tough.
If a page ranked in first position for the term pensions, the likelihood is that it has generated the best ratio of good to bad clicks compared to the competition. But this may change over a year as search habits change or a new and improved page is created.
I am including bounce rate and pogosticking under the auspice of good and bad clicks. The higher the proportion of good clicks over time, the better that page will rank.
How can I influence it?
Relevancy
Uniqueness
Giving people what they want when they want it - answering the question
Dwell time
Dwell time measures how long a user stays on a page after clicking through from the SERPs. In isolation, dwell time isn’t the best measure of page quality, Particularly if there’s a dearth of quality content available or the current format isn’t satiating the user.
But broadly, the longer someone sticks around once they’ve clicked through, the better.
How can I influence it?
High-quality, unique content
Understanding types of users (skim readers, visual learners, very thorough readers) and optimising for them
Breaking up walls of content
On-page visuals
Last good clicks
This is a great measure of current relevancy and works brilliantly when combined with the proportion of good clicks. By recording the last positive interaction, Google can assess the quality of recent site interactions.
By boosting rankings for pages with more recent positive interactions, Google can keep search results relevant and timely. It’s why pages a few weeks old can compete with content that has years of positive click data.
How can I influence it?
Page updates
Content freshness
Trending data
With so many queries, the SERPs change over time. Even over the course of a year. Around key holiday periods, why do you think e-commerce brands are prioritised? Because people are more likely to purchase products than at any other time of year.
By analysing pages trending in popularity and granting each page a ‘score,’ popular pages are elevated in search results. This is particularly relevant for news and other forms of time-sensitive content.
How can I influence it?
A contemporary understanding of the topic and the wider ecosystem
Changing your content to reflect seasonality
Make the most of Google Trends and build patterns of user behaviour
User comments
We’ve run a few tests on comments over the years. Although none particularly successful. However, fostering user interaction on the page helps Google understand how engaging pages are. Much like reviews, only the most engaged users comment. And not in a positive manner.
But that doesn’t matter. Engagement feeds the algorithm. That’s not restricted to social media. It’s why divisive content continues to be so prevalent. Maybe more so now than ever. Because it spikes emotion.
How can I influence it?
Understand your audience and focus your time there
Make it easier for people to comment (pop-ups, button placement etc)
Highlight your top comments; the comments that receive the most on-page engagement
Should I optimise for better engagement?
Most answers in SEO start with it depends. But this is one of the few that doesn’t. It’s evident. Particularly for publishers and companies who rely on the quality of their content and engaged users to survive. All of your content should be aimed at satisfying the user at a macro and micro level.
That means you want to balance brand affinity and creating evangelists with satisfaction at a query level. Essentially, have we answered the question?
Engagement and SEO - the verdict
Engagement is now a fundamental part of SEO. It has been for the best part of a decade. So it is less a question of balance and more one of integration. You need to establish the on-page fundamentals (title tag, h1, page structure, headings, quality images etc) whilst improving the user engagement.
This is one of the reasons SEO has changed so much in recent times. It’s less about satiating search engines (although that’s still important) and more about understanding the user.
At a page template or subfolder level, create reports that quantify on-page engagement. Identify weaker elements of your site and set targets based on a problem-solution approach.
Establish problems. Hypothesise solutions. Test.
Remember: Create content for people. Make it easy for search engines to understand.
