Bounce Rate and Rankings

[Rlilly]

Is it true high bounce rates effect rankings? A high bounce rate also means low page views in general.

However, Google says build a site for your viewers. But if you do, folks will find things quick and then leave immediately giving you a high bounce rate and low page views.

[tedster]

No - not "bounce rate" as it's defined in most analytics tools. Somewhat related, a quick "click-back" to the SERPs after landing from a click on a search result is one user signal Google (and other search engines) uses. Even that is pretty noisy taken on its own, and it's therefore just one of many related signals that are used in combination.

[deadsea]

I'm a big believer that Google relies on "user engagement" metrics quite strongly in determining rankings.

Google can measure:
1) Click through rate from SERPs. They track pretty much all clicks from the search results these days.
2) Re-click rate: how many people click on some other link after clicking on your site, having presumably backed out of your site not finding what they were looking for.
3) Re-search rate: how many people refine their search after viewing your site and not finding what they looked for
4) Time on site: through toolbar users, analytics, or adsens

In my opinion, click-through, re-search and re-click rates are probably the metrics that Google relies on most heavily, but its hard to tell.

A couple years ago, while looking into a large product information site, we were trying to determine why some product pages were ranking and others were not. For many products we were ranking very well, for others we couldn't get rankings no matter how much internal pagerank we were throwing at the product pages. We'd link to them from hundreds of other pages, and they still wouldn't climb the rankings. It soon became quite clear that pagerank was not the dominant factor here. The product info pages that were not ranking were "thin". They did not have as much information as other product info pages.

On the one hand, Google could be doing sophisticated page rendering and semantic analysis. On the other hand I think it is much more plausible they detect these thin pages indirectly by measuring how happy users are with the page through the metrics above.

Internally we started paying a lot of attention to correlation between bounce rate and the presence or absence of certain pieces of content on the pages. We found that pretty much every additional block of content that we had about a product could lower the bounce rate. Good photos were one of the items that could improve the bounce rate the most. Even the relevance of the ads that we were showing on the page had a big effect. A nearly empty product info page with not much more than the product name might have 70% bounce rate. A product information page full 10+ photos, videos, 20+ user reviews, professional reviews, related products, relevant ads, and links to articles elsewhere on the web, would generally have a 10% bounce rate (and a much better conversion rate).

We were able to reduce the bounce rate of specific poor performing products by getting more content for them specifically, starting with the content types that made the biggest difference. Rankings soon followed.

Five years ago it was enough to have pages with the right title and enough pagerank. That is clearly not the case now.

[deadsea]

Its also worth noting that we did not consider a user a "bounce" if they clicked on an ad, or clicked a link to another site. This is, unfortunately, NOT the same assumption that Google analytics makes. Google analytics sees users as bouncing even if they leave your site through UI that you provide. I would consider a bounce to be a user leaving through page abandonment: hitting the back button or closing the browser tab/window.

[travelin cat]

I believe that bounce rate is a very small part of the ranking algo. I base this on one of my websites.

On this site, we have a 4 word phrase. The landing page for this phrase has a form which, when completed, takes the visitor to another site, this creates a 70% to 90% bounce rate, yet we have ranked #1 for many years for the phrase.

Conversely, we have a 3 word phrase that uses an almost identical form, but the results page of the form keeps the visitors on our site causing a bounce rate of between 20% - 40%, and that phrase is always bouncing around #1 or #2 in the serps.

[deadsea]

Users that click through to another site didn't use the back button. That user shouldn't contribute the the bounce rate. They didn't bounce. At least they didn't contribute the re-search or re-click rates that Google actually cares about. I always measure bounce rate differently than Google analytics.

[tedster]

That's exactly why I like to use the term "click-back" as different from "bounce". Bounce metrics have a long history, way before Google, and they can be useful for the site management, even if they are noisy.

Click-back means returning to the SERPs after landing on that one page FROM the SERPs. It's still a noisy signal, but it's one that every search engine I know watches. It's mostly a kind of QA on their ranking algorithm - a way to check whether the results really line up with the user's intention.

Also note that a click-back to the same SERP and clicking on one of those results is a different signal than a click-back to the SERP followed by modifying the original query or entering a completely different query.

When it comes to other user metrics, Google has a lot to choose from just from their toolbar data or their Chrome browser itself. For example:

Time that the user spends on a page (dwell time)
Whether the user clicks a link on that page
Use of history, address bar, or bookmarks, to navigate to a different page
Whether the page is fully loaded
Whether the window gains or loses focus
If the page is scrolled, printed, bookmarked
Whether the user selects, cuts, or pastes from the page

Even more, different sequences of user behavior will mean something different than any one behavior on its own.

None of these above signals are dependent on the particular website's source code, but Google Analytics depends on JavaScript tagging, and GA is far from universally deployed.

So it doesn't make any sense to use a signal for ranking when there's not a controlled situation - where that signal can't be consistently accessed for every site that needs to be ranked. And Google doesn't even need to - there are plenty of other user engagement metrics that are available, and they're not nearly as noisy as bounce rate.

Also note - with regard to bounce rate, Google Analytics now allows you to tag certain actions as a conversion rather than a bounce, even if those actions involve a different website.

[Sgt_Kickaxe]

Embrace relativity, give Einstein a fictional hug today. What I'm suggesting is that I'm sure bounce rate plays some role in search rankings but it's important to compare the bounce rate of similar sites. A guide related site will have a high bounce rate because visitors find what they need and leave. A video site will have a low bounce rate since visitors want to see more than one video quite often. Measure your bounce rate against that of your competitors, if you can, because it's only them you need to beat.

Google tried to send a visitor to the exact page that meets their needs, if google succeeds you get a high bounce rate and that means you've done something RIGHT for Google to have chosen your site. Google isn't going to downgrade your site because you did something right, are they? :-)

[deadsea]

If you have a site that is trying to rank for "what is the state capital of Alabama?" type queries, then you would do well to try for a 100% bounce rate. For easy questions that can be answered in a sentence, the user should be able to easily find what they are looking for and leave quickly and satisfied.