Bounce Rates Count

Of late there has been a lot of discussion about bounce rates and whether or not the search engines count these in their algorithms. As far back as late 2007, there were reports that webmasters were seeing a difference in their rankings for major keywords within a few weeks of drastically changing their bounce rates. However, none of the tests and reports seem to be complete enough or repeatable enough to constitute proof.

As a result, there are plenty of naysayers who believe that such things as bounce rates are not now and probably won't ever be part of the search engine algorithms.

I am of the opposite view; bounce rates will certainly be part of the search engine algorithms and probably already are.

What I would like to do here is share with you some of the common naysayer objections and refute all but one of them. But first, for those who are scratching their heads about what bounce means, we are referring to people leaving a website. A bouncy website is the opposite of a sticky website, one where people stay a long time.

Objection 1: There is no definition of "bounce rate".

Response: This is the flimsiest of arguments. A bounce is when someone leaves a website, going back where they came from.

Objection 2: I don't like how Google Analytics defines a bounce.

Response: Sadly, Google doesn't ask me for advice, either. But cheer up, the bounce rate in Google Analytics might not be the same as they use in their algorithm, just as the little green bar is not necessarily the PageRank they use in their algorithm.

Objection 3: Many sites don't have Google Analytics turned on, so Google would have very incomplete data.

Response (scratching my head in confusion): What does Google Analytics have to do with anything? This is about Google (or Yahoo, or MSN, or Ask, or some other) tracking their own traffic and how their own users move about and - most importantly - how their users return to their website.

Objection 4: What is the threshold for a bounce? After 5 seconds? After 10 seconds? After 15 seconds? This is a mess! (This is often part of the how-do-we-define-a-bounce debate.)

Response: A bounce is a bounce, whether it takes a person one second or one hour to bounce back, it is a bounce. How the search engines choose to treat bounces with varying lag times is another matter. Let's be clear; they won't tell you, just as they won't tell you how many links on a page they index, how many they follow and how many they count in their ranking algorithms. Furthermore, it is a moving target. Just like every other algorithm input, bounce rates and bounce lag times will not be treated in the exact same way one day to the next.
Objection 5: What if people quickly click on an external link and leave my site? They found the site useful because they found a useful link on it, but they bounced.

Response: That is not a bounce, that's a referral. A bounce is when someone hits the back button.

Objection 6: What if the user quickly closes the window?

Response: That could be any number of things, but it is not a bounce. Who can guess how the search engines might treat that, or even if they treat it at all. However, it need not be considered a bounce unless the search engines believe it should be.

Objection 7: Doesn't a bounce mean the person has found what they want? Can't a bounce sometimes be good?

Response: Sometimes, perhaps, but rarely. After 5 seconds, a person has no time to read a page. After 30 seconds, they might have found something useful. So lag times matter. More importantly, the search engines can determine what a person does next. If a person returns to the search results and clicks on another link, that is a sign they did not find what they want. If they return to the search results and conduct a similar search, that might also be a sign they did not find what they want. If they return to the search results and conduct an unrelated search, that might be a sign that they found what they want. Search engines can weigh various bounces in light of the user's next action.

Objection 8: For some searches, people look for multiple sources, such as comparing prices, comparing products, seeking varying opinions, etc. Too many sites would be penalized if all those bounces were to be counted in the rankings.

Response: This is an example of false logic. If someone clicks on one website, then bounces, clicks on another website, then bounces, clicks on another website then bounces...all the high-ranking websites for that particular search query would be equally affected. Nobody would suffer a ranking disadvantage because rankings are relative.
Objection 9: Can't I just set up a bot to visit all my top competitors and leave their site after varying numbers of seconds to make it appear that their sites are all bouncy.

Response: Yes, you can. And you can get very creative. I have even heard of couriers in China travelling from one Internet café to another to click on a particular site as a means of increasing its rankings. I have no answer for this, other than that the search engines will have to control for this, just as they have found ways to control for automated link-building.

So have no fear. Good websites that provide what their visitors want or who help them find what they want will prosper. Sticky SEO looks at conversions and stickiness as integral elements to SEO.

Cheap sites that do a lot of link-building - bouncy SEO - counting on large volumes of traffic to offset poor conversion rates, will suffer - because the search engines will stop sending them that traffic.

It's just a matter of time. Or perhaps it has already started.