Dwell Time: Does This Ranking Factor Really Live Up to the Hype?

I'm an "SEO" with 6+ years experience; founder of The SEO Project; "link building" enthusiast; regular Ahrefs contributor; avid drinker of red wine; self-proclaimed steak expert; and all-round cool guy. I'm also shorter than you (probably).
    “Dwell time” is the latest buzzword in the world of SEO and right now, it seems like the entire SEO community just won’t stop talking about it. 

    The problem: “dwell time” is a confusing (and generally misunderstood) metric.

    I mean, what the heck does “dwell time” actually mean anyway, and what’s all this business about it being a ranking factor?

    Even if we assume that “dwell time” is indeed a “ranking factor”, it still begs the question: is it really that important, and should you actually be optimising for it?

    In this post, we’ll be answering all these questions (and more) once and for all.

    Oh, and we also reached out to some of the biggest names in the SEO industry to get their take on the matter.

    Let’s get started!

    What is “Dwell Time”?

    Dwell time” is the amount of time that elapses between you clicking a search result and returning back to the SERPs.

    For example, let’s assume I do a search for “whitehat link building”:

    whitehat-link-building-google-search

    Naturally, I click the first result and I spend a few minutes reading the content on that page (5 minutes 14 seconds, to be exact):

    whitehat-link-building-results

    Because I have a somewhat unhealthy obsession with learning, I then decide I want to know even more, so I head back to the SERPs (via the “back’ button in my browser) to look for more content.

    My “dwell time” on that page, then, was roughly 5 minutes 14 seconds (i.e. the amount of time that elapsed between the moment I clicked through to that page in the SERPs and the moment I returned)

    OK, so we understand what “dwell time” is, but why does this matter for SEO?

    A Brief History of “Dwell Time”

    Dwell time” was first mentioned by Duane Forrester (Senior Project Manager for Bing) in his 2011 post on the Bing webmaster blog:

    duane-forrester-dwell-time-bing

    Here’s what he said:

    [“dwell time” is] the time between when a user clicks on our search result and when they come back from your website tells a potential story.
    […]

    A minute or two is good as it can easily indicate the visitor consumed your content. Less than a couple of seconds can be viewed as a poor result.

    Duane Forrester
    Duane Forrester, Senior Project Manager for Bing

    But still, why does this really matter to a search engine?

    Here’s what Duane said:

    Your goal should be that when a visitor lands on your page, the content answers all of their needs, encouraging their next action to remain with you. If your content does not encourage them to remain with you, they will leave.
    Duane Forrester
    Duane Forrester, Senior Project Manager for Bing

    OK, makes sense.

    He’s basically saying the longer a user spends on your website (after clicking through from the SERPs), the more likely it is that they found your content useful.

    After all, Google’s primary goal (along with Bing and other search engines) is to provide users with the most appropriate, relevant and useful search result, so it’s easy to see how looking at “dwell time” could provide some insight into whether or not a page fits the bill.

    Here are a few “dwell time” examples and how they could be interpreted:

    • 2 second “dwell time”: The user probably didn’t find what they wanted/expected from your site (more on this later) and went back to the SERPs pretty sharpish looking for better content.
    • 2 minute “dwell time”: The user found your content pretty useful and stuck around a couple of minutes to read it.
    • 15 minute “dwell time”: The user found your content super-useful and was heavily-invested in what you had to say.

    02-dwell-time-illustration

    It’s not too far-fetched, then, to imagine that search engines could be using this information to help determine whether or not your page fulfils searcher intent for a particular query, and therefore should rank higher for said query.

    We’ll talk more about “dwell time as a ranking factor” later in this guide but first, let’s clear something up:

    Dwell Time’ vs. ‘Bounce Rate’ + ‘Time on Page’: What’s the Difference?

    If there’s one word that embodies these three metrics as a whole, it’s this:

    Confusion.

    I’ve actually seen a few SEOs using these metrics somewhat interchangeably (important: these three metrics are not interchangeable), and it seems that even some of the smartest and respectable SEO’s can’t help but get mixed-up from time-to-time:

    For example, here’s what Brian Dean has to say about “dwell time”:

    Google pays very close attention to “dwell time”: how long people spend on your page when coming from a Google search. This is also sometimes referred to as ‘long clicks vs short clicks’. If people spend a lot of time on your site, that may be used as a quality signal.
    Brian Dean
    Brian Dean, Founder of Backlinko

    I’m genuinely in awe of Brian’s work and while this definition isn’t entirely wrong (in fact, it’s one of the better ones I’ve seen), unfortunately, he doesn’t mention the part about the user returning to the search results, which is actually super-important.

    He also directly follows this definition with a video about reducing bounce rate:

    dwell-time-bounce-rate-brian-dean-backlinko

    This could be perceived as somewhat confusing.

    So — in the interest of clearing up any potential confusion — here’s what these metrics mean (in plain English):

    • Dwell time”: the amount of time that elapses between the moment a user clicks a particular result in the search results and subsequently returns back to them.
    • Bounce rate”: the percentage of single-page sessions (source) — i.e. visitors who chose only to visit one page on your website before leaving (important: these people may have returned to the SERPs or simply closed the page, it doesn’t matter which. It also makes no difference whether they stuck around for 2 seconds or 2 hours, it’s still technically a “bounce’).
    • Time on Page”: the amount of time a visitor spent on your page before going anywhere else (this could be back to the SERPs, to another page on your website, to a bookmarked page; literally anywhere)

    It’s also worth noting that while “time on page” and “bounce rate” are easily accessible metrics in Google Analytics:

    google-analytics-bounce-rate-time-on-page

    You won’t find any such metric for “dwell time”.

    If Google (or any other search engine for that matter) is actually using some kind of “dwell time” metric into their algorithm, they’re certainly not sharing this fact (or any of the data) with us.

    Because of this, SEO’s can only speculate as to how Google may be calculating “dwell time” and how they may be using it in their algorithm.

    Is “Dwell Time” a Google Ranking Factor?

    Right now, there’s no official statement from Google regarding whether-or-not “dwell time” is a ranking factor (although I personally believe this would make sense — more on this in a minute) but either way, there’s one incredibly important point you should understand:

    Dwell time” would only kick-in as a (possible) ranking factor once you rank in the top 10.

    Here’s why: if you’re not on the first page, you’ll be getting very little (if any) traffic as it is, which means there’ll be hardly anyone “dwelling” on your page — not even for a couple seconds.

    So, no matter how much you “optimise for ‘dwell time’”, you’re not going to get any additional traffic (or see any other benefits whatsoever).

    The bottom-line is this: if you’re not already on the first page, don’t worry about “dwell time” — worry about other (more important) factors that will get you to the top 10.

    OK, now that’s out of the way, let’s talk about why it could potentially make sense for Google to use “dwell time” as a ranking factor:

    It’s a Good Indicator of Relevance (User Intent)

    Let’s take the search term “paleo diet for beginners”:

    paleo-diet-for-beginners-results

    It’s clear from this query that we’re new to the concept of the “paleo diet’ and we’re looking for a definitive “beginners guide” to get us started.

    Now, ranking in position #1, we have NerdFitness:

    nerdfitness-google-result

    If you’ve ever read this guide, you’ll know it’s insanely extensive and covers pretty much everything you could ever want to know about the paleo diet (a fact backed up by the insane about of comments it has):

    nerdfitness-comments

    I spent roughly 15+ minutes reading through this (this would be my dwell time) when I first found it before returning to the SERPs; it was:

    • Incredibly extensive
    • All on one page (no need to visit more pages really)
    • Well-presented
    • Well-written

    Or, to put it simply: it fulfilled the search intent perfectly, as it truly is the definitive “paleo diet for beginners”.

    In contrast, let’s take a look at this page (currently ranking #6 for the same query):

    fitness-magazine-paleo-diet

    It doesn’t take long to realise that the content is nowhere near as extensive as the guide from NerdFitness (and nor is the overall UX particularly great).

    Here are just a few of the issues:

    • Content is relatively thin at around 500 words (not really the in-depth “beginners guide” we were looking for)
    • Information is pretty basic
    • It’s plastered with ads (this is usually enough for me to close a page immediately)

    To put it simply: although it’s semi-relevant from a topical perspective, it doesn’t actually fulfil the intent of the user (as we were looking for something a lot more in-depth, not just a basic overview).

    Because of this, I doubt my “dwell time” was any longer than 30 seconds on this page as honestly — as soon as I realised it offered very little value and the ads started to annoy me — I headed back to the SERPs in search of better content.

    Here’s an overview of my “dwell time” + overall experience for these two sites:

    NerdFitness: I stayed 15+ minutes before returning to the SERPs (so a 15-minute “dwell time”) because I found the content super-useful and was heavily invested in what it had to say (i.e. it perfectly matched user intent).

    FitnessMagazine.com: I stayed <30 seconds before returning to the SERPs (so a <30-second “dwell time”) because I didn’t find what I really wanted (i.e. it didn’t really fulfil user intent and was only semi-relevant)

    In theory, then, “dwell time” does appear to be a good indicator of whether-or-not a result is both relevant and serves the overall intent of the query.

    It (Potentially) Trumps Bounce Rate as a Ranking Signal

    I’m still not saying that “dwell time” is a ranking signal but if it was, it would make a lot more sense than looking at “bounce rate” (in my opinion).

    Why? Because “bounce rate” is pretty sketchy as a “ranking signal” to say the least.

    Here are two reasons why it’s a poor “signal”:

    • Users often “bounce” for any number of reasons and therefore, this metric alone doesn’t offer any truly definitive insight as to whether or not the user had a good or bad experience on your site.
    • There’s no real way to determine whether or not the information on a page fulfilled the overall intent of the query solely from looking at “bounce rate”.

    To illustrate this, let’s go back to our previous example (the “paleo” one).

    Because I only visited one page on each of the sites, I technically “bounced” on both (even though my experience on each was completely different).

    Not a good start for “bounce rate”, then, and it gets even worse when you consider how these visits would look in GA:

    Here’s my 15+ minute visit to NerdFitness:

    bounce-rate-google-analytics

    And here’s my <30-second visit to FitnessMagazine.com:

    bounce-rate-google-analytics

    Not only do these two (in reality, very difficult) visits appear to be 100% identical, they also both state that my “time on page” was exactly 0 seconds.

    Clearly, this isn’t true, so what’s going on here?

    In order for GA to calculate the “time on page”, it needs two clicks: an entrance click and an exit click. If there’s no exit click (e.g. the user clicking through to another page on your website), GA can’t calculate the “time on page”.

    Here’s a great explanation from AnalyticsEdge.com:

    For sessions where the user only looked at one page (a “bounce”), the Time on Page and the Session Duration is 0. This isn’t because Google knows they left right away — it is because they didn’t have any indication of when the user left so they couldn’t calculate the Time on Page, and they consider the lack of a value means 0.

    They go on to say:

    It [“time on page”] could have been 10 seconds or 10 minutes; they don’t know, so they say 0. Did the user read your web page? They don’t know. Maybe. Maybe not. All we know is that they didn’t look at another page on your site within the next 30 minutes (that’s how long a default session lasts).

    And that’s not the only potential issue with using bounce rate as a ranking factor either as in order for Google to do so, they would actually need to mine Google Analytics data (as there’s literally no other way to figure out the “bounce rate” of a page).

    But, Google’s official stance is that they don’t use any GA data in the algorithm:

    google-analytics-twitter

    So, by Google’s own admission, “bounce rate” is not a ranking factor.

    Sidenote.
    Clearly, taking Google’s every word at as gospel would be pretty foolish but in this instance, I believe they’re probably telling the truth. 

    Here are two good reasons why:

    • Not everyone uses GA: It was reported in 2012 that 10+ million websites had Google Analytics installed. If we assume that figure has grown ten-fold over the last 4 years (which it probably hasn’t), it still means that only around 10% of all websites have GA installed (according to current estimates from InternetLiveStats.com).Would Google really be able to decipher anything of true value from analysing “bounce rate” on just 1/10th of the world’s websites? Maybe, but I suspect the data would be predominantly useless.
    • GA is often incorrectly installed: If you’ve ever done an SEO audit, you’ll know how important it is to check that GA is installed correctly. Mis-installations are more common than you might think, which can lead to inflated “bounce rates” (amongst other metrics).This would be inaccurate data for Google.

    The bottom-line is this: even if Google was secretly using “bounce rate” data from GA in their algorithm, it’s highly probable that they would be able to decipher almost nothing of value from these “bounced” visits.

    IMO, this is where “dwell time” (potentially) trumps bounce rate as a ranking signal, as it would be super-easy for search engines to both collect and incorporate “dwell time” data into their respective algorithms.

    For example, let’s assume you Google “iPhone 7 review” and click one of the first few results:

    iphone-7-google-search

    Google can simply start a virtual stopwatch the moment you click your chosen search result.

    If and when you return back to the SERPs, Google could click the hypothetical “stop” button on their virtual stopwatch and, hey presto, they know exactly how long you spent on that site (i.e. your “dwell time”).

    Now, I know what you’re thinking, “how does Google know I’ve gone back to the search results?”; here are a two possible methods:

    • Chrome browser data: According to the latest figures from W3Schools, 72.4% of people now use Chrome. Being Google’s own browser, they probably know when you click the “back” button and return to the SERPs.
    • Next click” analysis: If you go back to the SERPs, it’ll probably only be a couple seconds before you click-through to another result. Google could wait for this click and thus, decipher a rough “dwell time” for your previous click.

    It’s clear then that with a bit of data mining, Google could almost certainly uncover some useful information from these methods and potentially use anything they learn in the algorithm.

    Google’s has (Possibly) Experimented with “Dwell Time” Before

    Some of you may remember that back in 2011, Google experimented with adding the option “block all example.com results” to the SERPs:

    google-search-block-all-results

    It was speculated in the SEO community at the time that Google was primarily using “dwell time” as a means to show this option.

    Here’s how it worked: if you clicked through to a page and stayed for a few seconds (i.e. a short “dwell time”), you would see the addition of “block all example.com results” on that result upon returning to the SERPs.

    Note: This post also states that Google showed a “plus” option when “dwell time” (supposedly) was high; I can’t find any screenshots of this, though, so here’s a mockup:

    google-search-plus-one-mockup

    Google also experimented with the “more by [AUTHOR NAME]” feature in the SERPs back in 2012, which looked like this:

    more-by-author-google

    Here’s a more detailed overview of how it worked:

    • User clicks through to an authorship-enhanced result in the SERPs
    • User spends a good chunk of time on the page (~15 minutes)
    • User clicks the “back’ button to go back to the SERPs
    • User is greeted with a “more by [AUTHOR NAME] button under the same search result they previously clicked.

    Right now, it looks as though nothing became of these experiments (as authorship is pretty much completely gone), although I personally have a hunch this isn’t the case (keep reading!)

    (Potential) Caveats for “Dwell Time” as a Ranking Factor

    So at a first glance then, the idea that Google may be using “dwell time” as a ranking factor not only seems pretty logical but also quite likely.

    But, as the saying goes: if it seems too good to be true, it probably is.

    Here are a few (potential) issues with using “dwell time” as a ranking factor:

    It doesn’t work well for simple question queries

    Let’s take the search query “when did robot wars stop airing?”, for example:

    when-did-robot-wars-stop-airing-google

    Right now, this doesn’t bring up a knowledge graph result and thus, you have to click through to one of the results to find your answer.

    If you click through to the top result (from Wikipedia), you find your answer in the first sentence:

    robot-wars-wikipedia

    Following this, you’ll probably then return to the search results quite fast (meaning “dwell time” will be pretty low — likely <10 seconds).

    However, this doesn’t correlate to a bad UX; you found the information you were looking for (and more) pretty sharpish.

    Here’s what Eric Enge had to say on the matter:

    There are many scenarios where SHORTER dwell time is an indication of quality. For example, anytime someone is looking for a quick piece of reference information, such as a zip code or phone number for a business. For informational searches like these, you want to design your pages so users find what they want pretty much immediately. 
    Eric Enge
    Eric Enge, Founder of Stone Temple Consulting

    So, in this example, although the “dwell time” was pretty low (which may theoretically indicate a potential problem), this was almost certainly the most appropriate and relevant result for the #1 spot.

    It doesn’t work well when you’re searching for a specific page

    A couple of months ago, I listened to a great interview with Richard Branson.

    Last week, I decided I wanted to listen to it again but couldn’t remember exactly which one it was.

    All I remembered was that it was a pretty long podcast about Virgin (obviously), so I went ahead and Googled “Richard Branson podcast virgin”.

    The result: many results!

    richard-branson-podcast-virgin-google

    I clicked the #1 result (naturally), but after a few seconds on the page, I realised this wasn’t the one I was looking for; I head back to the SERPs and click the next result.

    Nope, not that one either.

    Finally, after sifting through a few results, I find the one I was looking for.

    But, here’s the thing: because my “dwell time” was extremely low on the first few pages I clicked, this could have potentially had a negative effect on the page/site itself (assuming Google uses “dwell time” in their algorithm).

    However, there was nothing wrong with the first few results I click; they simply weren’t what I was looking for.

    It’s highly likely that most others searching for this phrase are simply looking for any interview with Richard Branson, not a specific one.

    It doesn’t work well for “AFA” (Accidental False Advertising) pages

    AFA” results are those that — at first glance — appear to offer exactly the solution/information you were looking for but, upon further inspection (which takes time and thus potentially increases “dwell time”), you realise this isn’t the case.

    Here’s an example:

    I was recently looking for a Google Sheets template that could scrape Google search results from within Google Sheets itself.

    I searched for “google results scraper google sheets” and clicked the first result, which happened to be from SEER Interactive:

    google-results-scraper-search

    At first glance, this looked perfect:

    seer-interactive-google-sheets-scraper

    I clicked the link to the included Google Sheets template, (which opened in a new tab, so I never left the SEER website), made a copy of it, and tried it out.

    But, after a couple of minutes, I realised the spreadsheet no longer worked and was giving an error.

    seer-interactive-google-sheets-error

    I clicked back to the SEER tab (I never closed this btw) and scrolled to the comments section to check if others were having the same issue; they were:

    seer-interactive-google-sheets-comments

    I went back to the search results and tried the next few results.

    So, although my “dwell time” on the page was quite high (>5 minutes), it didn’t actually fulfil my needs at all and — in its current stage — doesn’t really deserve to rank anywhere in the top 10.

    It is easy to manipulate for Black Hat SEOs

    Even if we assume that Google genuinely does use “dwell time” as a ranking factor, we’re still in the dark about exactly how they calculate this.

    However — and I certainly don’t claim to be an expert here — I get the feeling it probably isn’t that difficult to figure out.

    Here’s my best guess:

    01-dwell-time-illustration

    If this happens to be anywhere near accurate (which it probably is), you could quite easily use bots that mimic human behaviour to artificially inflate “dwell time” and thus, rank higher.

    This means that — should Google be using “dwell time” as a ranking factor — they’re going to have to be super-careful not to give it too much weight.

    So, while the die-hard “white hat” view may be that simply not knowing the exact algorithm (for calculating “dwell time”) could theoretically make it more efficient as a ranking factor, this isn’t actually true and who knows, we could end up seeing a post like this a couple of years from now (like we did with guest posting):

    matt-cutts-dwell-time

    Sidenote.
    While this screenshot is clearly fake and meant as a joke, you really never know what might happen in the world of SEO

    Should You Be Trying to Improve “Dwell Time”? (and if so, how?)

    So, assuming “dwell time” actually exists (by this, I mean assuming Google uses it somewhere in their algorithm), there’s no 1–2 punch tactic you can use to magically improve your “dwell time” overnight.

    Google is most likely looking at a lot of different user engagement metrics (“dwell time”, CTR, etc.), so my advice would be this: don’t get obsessed over these metrics.

    Instead, focus on creating great content and offering a great UX.

    If you can do this, you’ll almost certainly improve your “dwell time” indirectly without really trying.

    I reached out to Danny Sullivan (founder of SEL, MarketingLand.com, etc.) to ask his opinion on all this and here’s what he had to say:

    I think Google probably tries to measure and use engagement as part of its ranking algorithm. I think precisely how it does this isn’t known, I think too many SEOs obsess that it must be clickthrough rate. It largely doesn’t matter. As marketers, you want people engaging with your content first and foremost. So focus on that, and you’ll probably align with what Google wants
    Danny Sullivan
    Danny Sullivan, Founder of SearchEngineLand

    Exactly.

    So, instead of listing a bunch of ways to “improve your ‘dwell time’”, here are a bunch of ways you can improve your overall UX and, as Danny suggested, get people engaging with your content:

    #1 — Create better content

    Pretty obvious, right?

    However, I should stress that this doesn’t necessarily mean creating longer content; sometimes the most deserving result for the #1 spot is the most succinct.

    Here’s Erics take:

    I bet if you ran an experiment to measure the average dwell time on millions of websites and their ranking positions in the SERPs that you would see a strong correlation between dwell time and ranking. Does that mean that I think dwell time is a ranking factor? NO. It just means that there are more searches where a long dwell time means a user is happy than there are searches where a short dwell time does. There are also likely many searches where dwell time is irrelevant as a measure of quality too. 
    Eric Enge
    Eric Enge, Founder of Stone Temple Consulting

    For example, Google the question “is it Christmas?” and you’ll see isitchristmas.com ranking #1.

    If you click through to the page, you’ll soon see that it’s a one-word, one-page website:is-it-christmas-website

    Now, your “dwell time” on that page is likely to be no more than a few seconds but as Eric pointed out, “dwell time” isn’t always a good measure of quality.

    It’s clear that this content is by far the most worthy of the #1 position.

    #2 — Make sure to target the right keywords (and don’t ‘clickbait’)

    Sometimes keywords may seemingly make sense to target, but if you don’t fully understand your target audience, you may be making errors.

    For example, let’s assume that I put together a post entitled “the advanced guide to SEO” and in it, I write about putting keywords in your title tags, meta descriptions, and building links from directories.

    Clearly, these aspects of SEO are far from advanced and thus, don’t belong in an advanced SEO guide.

    Anyone coming across that page by googling something like ‘advanced SEO guide’ may very well be enticed by the title, but as soon as they click through to the page and see the (poor) content, they’ll most likely head back to the SERPs within seconds.

    Here’s what Eric had to say:

     In the long run, what Google wants to see is who are the types of people that represent the very best match for your site. It’s obvious who those are – your prospects. Serve them extremely well, and you align your goals and those of Google in the best way possible.
    Eric Enge
    Eric Enge, Founder of Stone Temple Consulting

    #3 — Improve page load time

    According to this infographic by KISSmetrics, 47% of people expect a web page to load in 2 seconds or less.

    Sidenote.
    This infographic was published 5 years ago, so people’s expectations are probably even greater now. 

    If your website is too slow, people are going to click back to the SERPs and they will never get a chance to see or engage with your content (if this happens, “dwell time” will be zero).

    #4 — Make use of a value-adding internal linking structure

    A good internal linking structure isn’t just “good for SEO”, or “increasing dwell time”, or “reducing your bounce rate”, it also improves things for your users.

    Why? Because a good internal linking structure will direct visitors to pages they genuinely might want to read.

    For example, if a visitor reads a post entitled “how to build muscle” on your website, it’s not too farfetched to imagine that they may also be interested in your post about “10 best protein powders”.

    #5 — Make sure you have a clear layout

    There’s nothing worse than clicking through to a website and feeling confused, bewildered or simply overwhelmed.

    Therefore, you should make every effort to ensure that your layout is simple, your content easily digestible, and that all important elements are in prominent positions.

    Remember, people are always one click away from thousands of other results; they’re not going to bother reading your post if it’s written in yellow Times New Roman text on a white background, no matter how good it is.

    #6 — Remove overly-obtrusive ads/pop-ups

    Nobody likes ads; it as simple as that.

    By removing any obtrusive ads from your site, you’ll improve your overall UX and trust me, your visitors will be much happier (and yes, this means no “welcome mats” loading the second someone visits your page, no broken links, etc.)

    Oh, and please, stop it with the “content gates”.

    #7 — Keep your content updated

    Some content will benefit more than others from this one, but you should strive to update things every few months (as a minimum).

    Why?

    Because people are probably more likely to trust advice given by a page clearly stating that it was “last updated: 2 weeks ago” over a page showing “last updated: 3rd March 2011” (especially for time-sensitive queries in ever-changing industries, such as “SEO”)

    #8 — Implement pageless scrolling

    People are lazy; many can’t even be bothered to click the “next page” button these days.

    If you have a lot of multi-page content, introducing a pageless scrolling design (like Facebook/Twitter) will do wonders for your overall UX.

    #9 — Use responsive design

    Believe it or not, I still visit pages on my phone that have no responsive design.

    The result: I go straight back to the SERPs (“dwell time”: zero) and look for another result (if I’m on my mobile, I mean; it doesn’t affect my desktop searches).

    I know, this isn’t going to be news to most Ahrefs readers, but I really recommend double-checking that this is actually working properly for any sites you’re involved with. I see a lot of issues with plugins blocking on-page content, weird font styling, and other similar stuff, so get it fixed!

    My (Probably Way-Off) “hunch”: Could Google Be Using ‘Dwell Time’ for “Private” Results?

    Before I wrap this up, I want to mention one last thing that popped into my head while writing this post.

    I’m sure you don’t need me to tell you that these days, Google search results are personalised by default.

    Note: You can verify you’re looking at “private” results by looking for this icon in the SERPs:

    private-serps-icon

    If you also have this turned on, you may have noticed that Google has started adding a “You visited this page” message to any results in the SERPs that you’ve visited before:

    serps-you-visited-this-page

    There’s also the “You visited this page on XX/XX/XX” one:

    serps-you-visited-this-page-on

    And the “You’ve visited this page X times. Last visit: XX/XX/XX” variation:

    serps-you-visited-this-page-last-visit

    From my casual browsing observations, it seems that Google ranks these pages higher when you return to the same SERP at a later date (only for you, though).

    My hunch (and it really is nothing more than a hunch — I’ve done no testing here) is that Google may well be using “dwell time” data to decipher whether or not it should show these previously visited results higher in your personalised SERPs.

    However, just because you’ve visited a page, it doesn’t automatically mean that it’s the best result; here’s why:

    1. You might have only stuck around for a few seconds before returning to the SERPs (due to the page not offering the information you were looking for).
    2. You may have visited another page on the same SERP page and preferred that result.

    It would, therefore, make little sense for google to blindly favour the pages you’ve visited before in your personalised search without taking into account certain metrics, one of which is almost certainly “dwell time”.

    OK; Let’s Wrap This Up!

    Dwell time” is clearly generating a lot of hype in the SEO world these days but hopefully, this post helped you to understand it isn’t some kind of individual ranking factor you should obsess over.

    Instead, focus on providing your users with the best possible experience; this means making your content a delight to read and your website a pleasure to visit (i.e. things that most SEOs and marketers have been advocating for years).

    At the end of the day, “Dwell time” is just another possible way for Google to determine whether or not your page deserves to rank, so give them what they (probably) want: happy users.

    I'm an "SEO" with 6+ years experience; founder of The SEO Project; "link building" enthusiast; regular Ahrefs contributor; avid drinker of red wine; self-proclaimed steak expert; and all-round cool guy. I'm also shorter than you (probably).

    Get notified of new articles

    51,255 marketers are already subscribed to Ahrefs blog. Leave your email to get our weekly newsletter.

    • namely to increase the waiting time needed to publish a summary of a longer article Does seo work and similar issues when? Thanks for the information.

    • Thanks for the best, sanest discussion of dwell time and its possibility as a ranking factor I’ve seen anywhere. Well done! 

      One more scenario in which dwell time might be a false flag for content quality and user satisfaction: shopping. Often when I’m shopping I may click back and forth rather rapidly among multiple results because I’m just at a stage where I’m comparison shopping, maybe for price or certain features.

      Because of that and the other problems you mentioned, if Google is using dwell time in any way, I suspect they are trying to discern user intent, and only using it where it would be more likely that dwell time is a reasonable indicator that the landing page satisfied the user.

      • Joshua Hardwick

        Completely agree about shopping, Mark. I guess that’s where Google could be using Rankbrain to judge things on a query-by-query basis??

    • Tobias Dean

      So what we need analytics software providers is a ‘time to bounce’ stat the measures only the time on site of users that bounce.

      • Except bounce is not the same as dwell time, as the article explains. The problem is there is no way for us to track once someone leaves our site where they go next. We can’t tell if they went back to the same SERP, which is what dwell time is concerned with.

        • Tobias Dean

          I understand, but analytics software providers can track ‘bounces’ they can track ‘time on site’ so why can’t we segment bounces with the time they spent on page? Might not be 100% as there will be other actions than bounces to the serps but would provide more insight as ‘dwell time’ is increasingly being considered a factor — rankbrain etc

          • I suggest you reread the article. It’s entire point is that it is quite uncertain, and leaning toward doubtful, that dwell time is an actual ranking factor, and even if it is, its effect would be so minor in the mix of things that you’d be silly to try to “optimize” for it. On the other hand, gaining more time-on-page and time-on-site is in itself a good metric, and an indicator that you’re creating real value for users, which is good for you even if it’s not a ranking factor.

            And RankBrain has nothing whatsoever to do with “dwell time.” It’s role is to “translate” queries for the rest of the algorithm as to user intent and meaning, and sometimes to help select the right mix of ranking factors that should influence the query.

            • Tobias Dean

              Fair enough my basic understand of rankbrains role (none of us know), is the increased machine learning aspect, understanding previous user behaviour in interactions with a page to better determine which page to serve next time. Hence where a factor such as ‘dwell time’ might play a part. But of course I agree in the obvious aim should always be to improve usability and generally onsite experience etc.

            • Actually, we DO know some things about RankBrain: what Google has told us, which is what I listed in my previous comments. There is NOTHING in RankBrain that is about “understanding previous user behavior in interactions with a page.” Nothing. Never even brought up as anything RankBrain does. People are crediting all sorts of things to RankBrain which no one at Google has ever said it does, and many that Googlers have outright said have nothing to do with it.

            • Tobias Dean

              Thanks Mark, appreciate your take

      • Joshua Hardwick

        Tobias, it’s impossible for webmasters to see “dwell time” information (if it even exists). Just focus on creating the content that best serves the users query, and you’ll do fine.

        • Tobias Dean

          i appreciate we can’t see that data now might point is if we’re accepting this factor is playing an increase role and I think most would agree user data is weighted significantly heavily than a few years ago [insert one of various high correlation studies] then there may be some value in one of the various analytics providers investing into unlocking some of this data.

          • Joshua Hardwick

            The problem is that Google owns the data, and there’s no way for a third-party analytics software company to get their hands on it. Why? Because there’s no way for them to know when a user specifically goes back to the Google SERPs.

    • Joshua Hardwick

      I’d imagine that Google can filter outliers such as yourself with machine learning and whatnot. I.e. if they see someone clicking multiple SERP results within the space of a couple seconds, it could be automatically discounted.

      That’s assuming “dwell time” is even a thing, of course

    • I really enjoyed this article Joshua. You dug deep, and it shows — thanks for creating such a clear and concise piece.
      *I didn’t see the term ‘pogo stick’ though — which is synonymous with a bit of what you were explaining (I think) 🙂

      • Joshua Hardwick

        I’m glad you enjoyed it, Andy! 🙂 And the term pogo-sticking somehow slipped my mind while writing this it seems!

    • A great article, and great disambiguation with bounce rate and time on site (which are topics I obsess over).

      In order for GA to calculate the “time on page”, it needs two clicks: an entrance click and an exit click. If there’s no exit click (e.g. the user clicking through to another page on your website), GA can’t calculate the “time on page”.”

      That’s not entirely correct; you can fix this lapse with some easy js extensions for analytics which send timer events such as “riveted”. While it won’t update the “time on page” metric within Analytics, it will give you a page-by-page breakdown of time spent on page.

      • Joshua Hardwick

        Great tip, David 🙂 

        However, the main point I wanted to make with this post is: don’t obsess over such metrics (e.g. “dwell time”, time on page). Instead, focus on creating content that perfectly fulfils the search query and you won’t go far wrong.

    • Kuba

      Joshua, great text. Thanks. 

      My personal observation: examples, comparisons, and visuals related to the content are great at increasing dwell time. People may skim they text, but the stop over when they see examples.

      • Joshua Hardwick

        Good point, Kuba 🙂 I’d say that adding visuals is part of creating high-quality content and, like I said, that should be your main focus.

    • Just Web

      What is that screenshot app you use?

      • Joshua Hardwick

        Skitch 🙂

    • Great Golf Deals

      For an ecommerce site, price shoppers would usually generate a short dwell time as they usually click through to get the price then go on to the next vendor.

      • Joshua Hardwick

        I agree; I imagine Google looks at “dwell time” on a query-by-query basis.

    • Jörg Bachmayr

      Thanks for this excellent article, Joshua. My point of view on the caveats:
      It’s about the average dwell time for that specific query (or, proabably, a set queries with similar intent). On average the query “when did robot wars stop airing?” will have a short dwell time. It doesn’t matter what the (average) dwell time for “paleo diet for beginners” is, that’s not what it is compared against. Also, Google is seeing dwell time in combination with the subsequent behavior on the SERPS. Are other results clicked? What’s the last result that is clicked on before the particular query is abandoned?
      Frankly, I strongly disagree with the point that dwell time can be easily manipulated by black hats with bots. Google’s ad business heavily relies on being able to identify bot clicks on ads. I think it’s safe to assume that they have developed very advanced ways to do so. One visible outcome of these efforts is reCAPTCHA.
      All in all, I find it very likely that Google uses dwell time.

      • Joshua Hardwick

        I completely agree, Jörg. My examples for “when did Robot Wars stop airing?” and “paleo diet for beginners” were meant as two separate examples. 

        I would imagine that with the “robot wars” query, a shorter “dwell time” would actually be a positive thing (as it means the user found the information they were looking for quickly), but with the “paleo diet for beginners” term, a longer “dwell time” is probably a positive thing, as it’s clear the user is looking for a lengthy guide.

        Like you said, it’s all about the average “dwell time” for a specific query (or set of queries). It wouldn’t make sense to compare the dwell time of completely unrelated search queries.

        Even if Google can decipher bots (which, I agree, they probably can), you could still manipulate things by paying people to click and “dwell” on certain results (would probably be pretty easy/cheap to do with mechanical turk and similar services)

    • Completely agree with you. You just have to create content which is interesting enough for user to spend some time on your site and like that dwell time issue is taken care of. I have I doubt though. If after searching for particular keyword, I have opened different sites with multiple tabs from SERP. and I simply hop from one tab to another to check different site instead of going back to SERP. How dwell time will be calculated in this case?

      • Joshua Hardwick

        In that case, I’d guess that the algorithm would simply discount your data (that’s assuming “dwell time” is real and works roughly how we imagine, of course)

    • Loved the write up.

      I would disagree with this number, however:

      #8 — Implement pageless scrolling

      People are lazy; many can’t even be bothered to click the “next page” button these days.

      If you have a lot of multi-page content, introducing a pageless scrolling design (like Facebook/Twitter) will do wonders for your overall UX.

      It doesn’t always improve the overall UX. In fact, when implemented poorly or used in a case where it is unnecessary, it will damage the overall UX

      • Joshua Hardwick

        Kyle, I’d argue that the same applies to anything (i.e. poorly executed implementation, or implementing where unnecessary). I agree it’s not the solution for every page/website though.

    • Interesting, I had always thought that Google denied observing anything beyond the click on the SERP.

      • Joshua Hardwick

        Google denies a lot of things, Douglas 🙂

    • Chris Ward

      I read your blog post. They are some very interesting finds. It makes sense for Google to only use dwell time for certain types of queries, and disregard data below a certain threshold. The closing advice in the article is solid. You should write some more articles for your blog, I was hoping to find more of the same!

    • Chris Ward

      Another factor that could be important for dwell time is video. Having a highly relevant video included on a web page will, in all good theory, naturally increase dwell time. If you’ve also published that video on YouTube with a link back to the web page in question, it may give Google another piece of the puzzle — perhaps it incorporates ‘YouTube dwell time’ into the ranking signals. 

      Google will always be looking at an incomplete picture to some extent. Links, dwell time, youtube playback time — none of these factors on their own can give Google the full puzzle, but the more data it has of each for a particular web page, the better informed its guess is, and the more likely you will achieve your deserved ranking.