Bounce Rate In Google Analytics

[bwnbwn]

Website in GA has a bounce rate of 85% time on site is 1.14 min. in Audience Overview.

In GA Behavior Site Content All Pages the average time on the page is 3.12 min. I looked at 1000 pages this average includes mobile.

To me this just doesn't make sense.

What is the difference in the two and why such a large difference in the two?

I thought the code might be corrupt so it was updated to the Universal GA code. Since the change this is the time I checked.

[JD_Toims]

Bounce Rate In Google Analytics

...is irrelevant as far as rankings go.

More to the point of your question: GA records any landing on a page where there is not another interaction (page view) as a 0 time on site for bounce rate and some other metrics, whether the visitor is there for 1 second or 10 minutes. I haven't researched it, but my guess is Behavior uses the actual time someone was on the page even if they don't visit another page.

[netmeg]

#1 I'm not sure it's an accurate number and #2 I'm not sure it's a useful metric by itself. Honestly, I don't pay a lot of attention to it.

[rish3]

The discrepancy is probably a number of sessions with a page depth of 0. Go to Audience -> Behavior -> Engagement and click on the "Page Depth" tab. See if there are a high number of sessions with "less than 1" page view.

There's not much info on this, but I suspect a page depth of 0 indicates some sort of error condition, website problem, or unusual client (like a bot that runs javascript, but not well).

You can drill down on those "0 depth" visits and see if they are clustered around certain browser types, ip addresses, etc.

1. Go to "Audience -> Overview",

2. Click the "All Session - 100%" button with the blue circle on it.

3. Click the red "+ New Segment" button.

4. Select "Advanced->Conditions" and add a condition of "Behavior -> Page Depth < 1".

5. Name it "Page Depth 0" and save it.

Now you should be able to drill down into those problematic visits and see browser type, geography, etc. You'll probably find some kind of commonality that will give you a clue as to what the issue is.

[bwnbwn]

netmeg I don't either but this has been bugging me and doing some searches I really couldn't find any good info on it so I posted here to see. I am going to look at what rish3 has suggested. Thanks.

JD_Toims actually bounce rate is a metric in the algo not sure how much weight it is given but I do know it is up there and a metric Google uses to gauge user experience.

[JD_Toims]

Bounce Rate In Google Analytics... is irrelevant as far as rankings go.

And where's the source for bounce-rate in-and-of-itself being a factor in rankings? I've had pages with a +90% bounce-rate rank #1 and others with 25% rank #1, so you'll need to provide some sort of data that says it's actually a factor on it's own before I'll buy it -- They've always said it's too noisy to use and could only be a minor factor if it's used at all.

[rish3]

The most definitive source I could find where Google is using something like bounce rate is in a patent(#8775924) they were recently granted. The title of the patent is "Processing web pages based on content quality".

They are using the term "amount of time spent on that page", which isn't exactly bounce rate, but there's certainly a close correlation.

See [google.com...] for the full text. Here's the relevant portion:

A web page may be suggested for review and/or its content quality value may be adapted based on the amount of time spent on that page. For example, if a user reaches a web page and then leaves immediately, the brief nature of the visit may cause the content quality value of that page to be reviewed and/or reduced. The amount of time spent on a particular web page may be determined through a variety of approaches. For example, web requests for web pages may be used to determine the amount of time spent on a particular web page.

[JD_Toims]

A granted patent doesn't necessarily mean they use it, and time on page is totally different than bounce-rate -- Bounce-rate is one-and-done, but doesn't in-and-of-itself indicate "good" or "bad".

10 seconds on one page > 10 seconds on another page > back to the results lowers bounce-rate, but looks way worse than 2 mins on a page and then ending the search or even back to the results, even though the 2 mins on a single page increases bounce rate while 2 10 second page views lowers it.

Not trying to sound rude or anything, but I really think it would be good if people [not only you] stop talking about bounce-rate being any type of stand-alone factor until there's some proof, because Google's said repeatedly it's too noisy to make much of a determination from and if you think about it, it really is, because 10 seconds on a page and 2 mins on a page are both bounces, but they give totally different signals about whether the query was answered or not.

[rish3]

Sigh. I specifically pointed out that it was time on page, and also specifically mentioned that it wasn't bounce rate.

[JD_Toims]

You also said time-on-page is a close correlation to bounce-rate, and it's really not. They're two totally different things, which is why I explained.

I actually have a one page site with 100% bounce-rate as far as Google's concerned. People can either buy or not. The shopping cart is on PayPal's site. The thank-you page is on the supplier's site. It's been pinned at number 1 for 7 years now, because it gets the job done -- Thinking bounce-rate is a factor and trying to lower it would kill the ease and simplicity of the user-experience.

[rish3]

It's a close correlation to what people really want to know when they ask about "bounce rate".

They are asking if Google tries to assess quality based on low user engagement.

Neither is really a good measure of engagement, but Google can only use what it has access to.

[JD_Toims]

It's a close correlation to what people really want to know when they ask about "bounce rate".

Wow, all this time I've thought when people asked about bounce-rate they were asking about bounce-rate, not time-on-site, because I keep thinking if they want to know about time-on-site or user-engagement as a metric they'll ask a question about those things rather than asking about bounce-rate.

Finally starts to understand why there's so much confusion, FUD and misinformation surrounding some topics -- It's because some of us try to answer the question people actually ask rather than assuming they're asking a different question. Oops, I guess.

[petehall]

10 seconds on a page and 2 mins on a page are both bounces, but they give totally different signals about whether the query was answered or not.

I think the signal is someone returning back to the results to find an alternative source of information. The quicker they do this, the worse the score.

I'm sure it is a noisy signal but in numbers will provide a trend of how relevant a page is.

It would make sense to let 'the people' do a chunk of Google's work for them... :)

[JD_Toims]

In your example, if you added a ga-event on page scroll, your "bounce rate" would go from 100% to 0%.

No it wouldn't -- My stats would be skewed, but people would still only view a single page on the site and my bounce-rate would still be 100%.

And, if there's a definition of bounce-rate that's anything other than "percentage or number of visitors who only view a single page on a site", please let me know what it is.

Bounce rate (sometimes confused with exit rate)is an Internet marketing term used in web traffic analysis. It represents the percentage of visitors who enter the site and "bounce" (leave the site) rather than continue viewing other pages within the same site.

[en.wikipedia.org...]

Bounce Rate is the percentage of single-page sessions (i.e. sessions in which the person left your site from the entrance page without interacting with the page).

[support.google.com...]

[1][edited by: JD_Toims at 8:35 pm (utc) on Nov 19, 2014]

[rish3]

Okay. You win. Congrats.

For what it's worth, Google's definition of bounce rate is fundamentally different from wikipedia's

"Bounce Rate is the percentage of single-page sessions (i.e. sessions in which the person left your site from the entrance page without interacting with the page".

But, I can see that this issue is way more emotionally charged than I thought, so I give. I'm out.

[JD_Toims]

For what it's worth, Google's definition of bounce rate that is fundamentally different than wikipedia's

If you read the second paragraph on the Google help page you'll find they are really talking about a single page view, just like the WikiPedia definition, and the only one I've ever known to be generally accepted.

[bwnbwn]

If the bounce rate wasn't important to Google why even show it. There is limited space for them to show site metrics bounce rate, time on site are all part G uses to best determine a user experience.

I really don't care if you think it is or isn't part of ranking I do so what else matters.

BTW rish3 the tip you gave me has helped a lot. :)

[JD_Toims]

If the bounce rate wasn't important to Google why even show it.

Because it's expected of a stats program.

I do so what else matters.

You're posting on a community message board where people come to learn and get quality, accurate information -- Posting an unfounded belief you hold which is statistically refuted in a number of instances as if it's fact is not helpful or useful to those people, in fact, it's detrimental, because it adds confusion to the already confusing topic of SEO.

[bumpski]

bwnbwn

I do not understand why your post was not immediately attacked as off topic and relocated to:
[webmasterworld.com...]
the analytics forum.

Please do see the following two posts, one after being approved and published in the Google SEO forum, but then relocated to the Analytics thread by ? after an "this post is off topic" comment by EG.
The second thread was intentionally posted to the Analytics forum to help webmasters who are relying on Google Analytics for their SEO:

Title:
Highest Quality Pages, Google, & Google Analytics
[webmasterworld.com...]

Title:
Google Analytics Bug(s), Time On Page grossly Inaccurate
[webmasterworld.com...]

[bwnbwn]

Ahh didn't see area never have posted or visited it. Thanks for the links I apologize.

[bwnbwn]

Ahh didn't see this area never have posted or visited it. Thanks for the links I apologize.

[bwnbwn]

JD_Toims, if you would do some searches and read some highly respected educated post from this search I think you might understand why I feel it is part of the algo. I said

I do so what else matters.

because you seem to think just because Google doesn't come out and say this is in the ranking factors it isn't. This is called a rookie mistake. I know this is a community forum so with that said let the community be educated. Living in a box is not how good SEO's make a living.

Good search to help you gain some insight.

is bounce rate part of google algo

[JD_Toims]

because you seem to think just because Google doesn't come out and say this is in the ranking factors it isn't.

No, I don't need Google to come out and say it -- I happen to code a bit and when I think about using bounce-rate as a factor on it's own from a coding/algorithmic perspective, I run into absolute noise, because it doesn't tell me anything definitively.

A 100% bounce-rate on a page could mean people found the answer.
A 100% bounce-rate on a page could mean people didn't find the answer.

A 0% bounce-rate on a page could mean people found the answer and it was in a split article they continued reading on another page.

A 0% bounce-rate on a page could mean people didn't find the answer, but there was a prominent link that looked like the resource linked would contain the answer, so they followed it.

And, all the other percentages in-between say about the same thing, which is why MC as GG said it was too noisy to use right here on these very forums years ago -- I think he qualifies as "respected and educated" or at least did as GG.

---

I work with some people who "bought into" what the "experts" say and bounce-rate being a super important ranking factor, so they insisted on a new design -- The time on site nearly doubled. Bounce rate dropped by almost 50%. The site completely tanked -- Oops! So much for "what the 'experts' say" being accurate and truthful.

---

If you really want to change your GA bounce rate because you think it's a factor, put an iFrame on the page with your GA code on the page and the iFramed page. Heck, put 10 iFrames. You'll have 11 page views with whatever time someone spends on the page hosting the iFrames for every visit and your site should just plain skyrocket!

---

As far as "the respected, educated posts" go, meh. I don't really need them to tell me what can be used and what can't, because I can think for myself. Besides, there are some "highly respected SEOs" who still talk about keyword density and have neat little analyzers on their sites even though it hasn't been a factor for over 5 years now, but they have to make money somehow...

I actually wrote a KWD analyzer myself 8 years or so ago when KWD was important -- It did exact match and stemmed comparisons, checked spelling, showed top phrases up to 3 words and a bunch of cool stuff.

And, as far as being a "rookie" goes, I guess if only having done this and not had a "day job" for a bit over a decade makes me a rookie, then you got me.

[rish3]

Coming back in since the heat factor seems unchanged without me anyway.

I do think that the "without interacting with the page" bit makes a difference.

Out of the box, GA's "bounce rate" would be useful, and likely highly correlated with "time on page" for sites where the goal is to get people to a landing page, and then get them from there to another page. That's very true for most ecommerce sites, many informational sites, etc.

For sites where it's not as useful, you can use built-in functionality within GA to define events that count as an interaction, but aren't pageviews.

Google seems to recognize this. From their help page on event tracking:

The term "Non-interaction" applies to the final...parameter that you can use with the _trackEvent() method. This parameter allows you to determine how you want bounce rate defined for pages on your site that also include event tracking.

They seem fine with "fixing" bounce rate such that it's more useful.

For example, suppose you have a home page with a video embedded on it. It's quite natural that you will want to know the bounce rate for your home page, but how do you want to define that? Do you consider visitor interaction with the home page video an important engagement signal?

If so, you would want interaction with the video to be included in the bounce rate calculation, so that sessions including only your home page with clicks on the video are not calculated as bounces.

And, they even seem fine with making a single page view + exit NOT count as a bounce, provided some interaction happens. In their example, playing a video. I assume they would feel the same way in other examples, like a one-page landing site injecting an event if an outbound link that feels like an interaction is clicked.

Which leads me back to the original question. When people ask if "bounce rate" counts in Google's algorithm, what's the right answer?

You could go with a literal answer of either:

- Google doesn't pull GA data at all for it's algorithm.
- Out of the box "bounce rate" is often meaningless

Both are true, but I'm not sure how helpful that is.

Might someone asking the question be open to a broader answer?

- High bounce rates in GA *could* indicate there's a user engagement issue. Especially if your goal is drive users across pages.
- There are things you can do (tweaking GA being just one of them) to see if you have a user engagement problem, or are just seeing the flaws of "out of the box GA / bounce rate"
- Google would likely use "user engagement" as a signal if they could
- Nobody can really prove or disprove what signals Google is using
- Improving user engagement is worthwhile regardless

[JD_Toims]

Google would likely use "user engagement" as a signal if they could

I'm not arguing with that at all, but bounce-rate as a factor by itself is unusable from an algorithmic programming perspective -- Even coupled with time-on-page it's incredibly noisy, because determining if someone spent time on a page said to have the answer due to finding the answer and reading it, or not finding the answer and looking for it because the page was returned in the results isn't something that can really be algorithmically determined.

Time-on-page + bounce-rate relative to other results is "less noisy" and could possibly be used, but that means the bounce-rate and time-on-page for a given page returned for a given query together have to be within "norms" relative to the other pages returned and since the stats for other pages are unavailable to us, it's unknown whether a site/page needs to increase it's bounce-rate and decrease the time-on-site or increase it's bounce-rate and increase the time-on-site or decrease it's bounce-rate and increase the time-on-site to be within "the standard" levels for other returned results.

Nobody can really prove or disprove what signals Google is using

It's very easy to disprove bounce-rate/page-views as ranking a factor. Just put an iFrame or 10 or 30 on a page with GA code on it and the iFrame page(s) and see if there's a ranking impact -- I've done it before. There is no impact whatsoever, but don't take my word for it, just try it yourself.

See my above post for another, "Wow, that really didn't help..." situation that wasn't even a "manipulation" of the numbers GA presents.

[edited by: JD_Toims at 12:23 am (utc) on Nov 20, 2014]

[JD_Toims]

Improving user engagement is worthwhile regardless

Not always as far as rankings are concerned. Again, see my previous post where someone I work with "bought into" that line of thinking, told me to make the change and it totally backfired -- The site completely tanked even though the "super important" user-engagement numbers all got *way* "better" than they were.

Sometimes, less is more...

[rish3]

It's very easy to disprove bounce-rate/page-views as ranking a factor. Just put an iFrame or 10 or 30 on a page with GA code on it and the iFrame page(s) and see if there's a ranking impact -- I've done it before.

I'm not saying they are using Google Analytics stats directly. That would be a bad idea for many reasons...most of them not technical at all.

I'm saying that they probably want to measure and include "real bounce rate". "Real bounce rate" meaning people landing on a site, then not interacting...for the various definitions of "interacting".

Also, I don't know that A/B testing SERPS is always useful. I would guess that that when they choose to use signals that aren't 100% clean (hello negative SEO) that the logic is more complex that just giving the signal a weight and using it.

[JD_Toims]

Let me give you some examples where "real bounce rate", even coupled with time-on-site gets totally noisy.

---

A person searches, clicks on Result 1, clicks back to the results two seconds later, clicks on Result 2, clicks to another page after 20 seconds, they spend 30 seconds on it and click back to the results, clicks on Result 3 and ends the search.

Which is the better result if the query was 2+2?
Result 1 which has 2+2=4 in 40pt type at the top of the page.

Result 2 which the searcher spent 20 seconds reading, then clicked the "find the answer" link and still couldn't find it.

Result 3 which verified 2+2=4.

---

A person searches and clicks on Result 1, they spend 1 minute on the page, view another page and click back to the results, clicks on Result 2, spend 25 seconds and only views one page, then clicks back to the results, clicks on Result 3, spends 20 seconds on the page, clicks back to the results, clicks on Result 4 and ends the search.

Which is the better result if the person was searching for a price on a TV?
Result 1 where they had to read the page, then click the "get the price link" to find what they were looking for.

Result 2 where the price was prominent next to the picture of the TV and easily found.

Result 3 where the price was prominent next to the picture of the TV.

Result 4 where the price wasn't present and the searcher decided to go to a store they know.

---

The point is, bounce-rate *cannot* be used on it's own. It doesn't "say" anything by itself -- And, trying to "lower the bounce-rate" and/or increase the "user engagement" isn't always the right answer for rankings, because the only way to even use "user engagement" at all effectively is based on "norms" or "standards" relative to other results returned for the query, so if other pages returned are giving the answer quickly and easily "forcing" or "enticing" another page view and more time on the site could/should *harm* rankings, not help them -- I've seen it happen.

[rish3]

Let me give you some examples where "real bounce rate", even coupled with time-on-site gets totally noisy.

Good examples, and I agree. But, there are other signals that Google openly admits to using that are very noisy as well. I can only assume that the algorithm is capable of some noise filtering, dynamic weighting, using some signals only when other signals are present, etc.

[JD_Toims]

I can only assume that the algorithm is capable of some noise filtering, dynamic weighting, using some signals only when other signals are present, etc.

I would guess this is about where they are, but where I get "frustrated" is when people [not you specifically] try to say bounce-rate *must* be a signal/factor on it's own, because as far as "noisy" goes, bounce-rate is nothing more than crazy-loud unintelligible clatter by itself, and it's really not possible to determine "good" or "bad" from it independently.

I guess basically what I'm saying is: people who try to manage their bounce-rate thinking it's an independent factor and a lower number must be better for rankings, rather than simply doing a better job of providing the answer/result a searcher is looking for will likely lose in the long-run, because two page views v. one page view doesn't necessarily make a site or result better or a page rank higher, it could actually do the opposite.