[michellegreer1]

37signals started as a design firm. Which meant they knew their audience because they WERE their audience. http://37signals.com/about

This is an exception, but rarely the case.

In regards to bounce rates, yes, they are absolutely important and are one of the core metrics you want to reduce when you implement split A/B testing. Think of it this way: if you open a physical store and most of the people who walk in immediately walk out, is your marketing effective? How much did you spend to get that person to walk in the door? If you have a high bounce rate, you will inevitably have a low conversion rate.

[patio11]

Curses to the "no immediate replying deep in a thread" filter.

If you have a high bounce rate, you will inevitably have a low conversion rate.

If this were true, then you could use a low conversion rate as a proxy for making the decisions you are currently making based on bounce rate. If two metrics inenvitably say the same thing, ditch one of them. The interesting case would be when they differ. However, if your bounce rate and your conversion rates are telling you different things, then your conversion rate is right, because of the two metrics it is the one actually putting money in your pocket. If I were selling a product and the best thing I could say about it was "Either superflous, or wrong!", I would stop selling it. Everyone should stop selling bounce rate.

However, it is not the case that if you have a high bounce rate, you will inevitably have a low conversion rate. Many pages on my site have a high bounce rate, which is almost entirely due to my traffic acquisition strategy, which is organic SEO. Because a) Google is not infallible with regards to relevance and b) organic SEO is greatly helped by providing linkable resources, some of which can be consumed without leaving the page, this tends to increase my bounce rate relative to relying on, e.g., paid CPC traffic responding to an ad telling them they need to sign up for my free trial. Since customers pre-select for willingness to sign up, bounce rates among that traffic are much, much lower.

(In this case, as an artifact of an implementation detail, any interest in seeing the signup form will cause someone to be recorded as non-bounce. Of course, if I had implemented the form the other way and made that click do some JS wizardry to pop the form up in a lightbox, then the bounce rate would increase automatically. The fact that my bounce rate is very sensitive to my implementation details rather than any externally visible difference in user behavior should be a strong hint as to how useful it is(n't)!)

Of course, I don't have the luxury of telling customers on organic searchers "This site offers a free trial! Don't click here unless you're ready to sign up for it, you'll hurt my metrics!" and even if I did, wow, that would be a catastrophically stupid thing to do now wouldn't it. I make thousands of dollars off the organic traffic even if 70%+ on some pages satisfy their immediate desires without generating a second page view. (And if they satisfy their immediate desire and then link to my page from their blog or class home page then I am so far ahead of the game it isn't even funny.)

Bounce rates also reinforce a pageview-centric metrics model which was suboptimal but chosen for technical reasons years ago, and is now VERY suboptimal, given that we can have high engagement interactions these days without ever generating a second pageview, thanks to implementation details like Flash or AJAX. (For that matter, we can have high engagement intereactions without a FIRST pageview, because a good portion of the conversation occurs outside of our own websites. Given that I'm telling this to a social media consultant on a forum commenting about her blog I feel a heavy dose of Whoa This Is Meta at the moment.)

Folks interested in more detail on bounce rates and their inherent, ineradicable suckiness can read my blog post on it later this week.

[paraschopra]

>However, it is not the case that if you have a high bounce rate, you will inevitably have a low conversion rate

I would say the best way to calculate conversion rate is after you exclude bounced visitors. Bounce rate tells what percent of people found your website wasn't what they were looking for. Conversion rates should tell, how good job am I doing in convincing the visitor, if the visitor chooses to spend time on the website and hence showing interest.

These two metrics are related and but not as tightly. And of course website can (and should) be optimized separately for both of these metrics.

EDITED for clarification:

Just to summarize, following is a pretty good way of seeing things:

CONVERSION RATE = (Total # of non-bounced visitors who completed goal) / (Total # of non-bounced visitors)

For all practical purposes, we should exclude bounced visitors from analysis as it just creates noise. Bounced visitors are the ones who got to your site because of chance and you should separate them from the visitors who came to your site with full interest.

[cschwarm]

That's surely an interesting idea, but I doubt it's doable.

First, you assume to know why visitors bounce (the "website wasn't what they were looking for"). That's quite a jump since bounce rates are about behaviour, not the causes leading to that behaviour.

For example, let's say I'm a bad writer and designer. My page may be absolutely the one a visitor was looking for, but due to my bad writing and design, the visitor thinks it's the wrong one. People scan web pages, they don't read them.

As another example, let's say I'm shopping for a new mobile and your page sells mobiles. I may bounce not because I'm one the wrong page, but because your price is easy to find and it's completely over-the-top in my opinion.

Second, there's also the problem of defining and measuring bounces.

Third, the conversion rate is, to some degree, a relative measure. Any number, say 10%, is meaningless unless you have an idea what's usual (for webpages similar to yours). Therefore, it doesn't matter what base it uses as long as everybody uses the same base.

Overall, I believe bounce rates are rather useless compared to other measures.

[pbhjpbhj]

Google Analytics will allow you to compare your bounce rate with sites of similar size, traffic and with a similar sector. Visitors > Overview > Benchmarking

[michellegreer1]

Actually, I'm no longer self-employed. I am now the Director of Partner Marketing for the Rackspace Cloud. Sorry for not updating my "About Me" page but I've been pretty busy lately.

As a marketer, I always try to think of what's best for the customer. In terms of bounce rate, it's actually pretty annoying to THINK you are going to find the right page when you Google it, only to discover something useless to you. So I try to keep bounce rates low because it increases customer happiness.

Maybe it makes me a wuss to think that way, but I've honestly never had an issue producing ROI.

[patio11]

In terms of bounce rate, it's actually pretty annoying to THINK you are going to find the right page when you Google it, only to discover something useless to you. So I try to keep bounce rates low because it increases customer happiness.

I see what you're saying in the first sentence but I do not think that the second sentence follows from it. For customers who are in goal-directed mode, often times a bounce is the positive outcome and multiple page views are the negative outcome.

As a trivial example: suppose someone Googles [Rackspace affiliate program phone number]. Their intent is crystal clear: they want to call you. http://www.rackspace.com/information/contactus.php achieves their task and will almost certainly result in a bounce. If you wanted to reduce your bounce rate, you could put the contact link four clicks down and do several millions in usability research devoted to discovering new, innovative ways to get people not to click on it, but that would be prioritizing a meaningless and arbitrary metric over your customer satisfaction.

P.S. If there is anyone reading this from Google please don't get any ideas about new research projects from the above paragraph.

[paraschopra]

Yes, if you define bounce rate by number of pageviews, then this holds true. I wish I could have defined bounce as visitors who spend less than say 10 seconds on website but calculating exact time on page in realtime is very expensive operation which most web analytics tools don't do.

A typical example supporting your case would be a blog post. Most people come to your blog post via a link and they read the post and they go back to whatever they were doing. Even when web analytics tools will call them a bounce (because they just saw one page), I would say them non-bounce because they spent time doing what they came here for.

[ErrantX]

> If you have a high bounce rate, you will inevitably have a low conversion rate.

This is a worrying statement from someone who is supposedly a web marketing expert!

There are some inconsistencies too; now your claiming bounce rate is important for customer sanity - something I certainly would agree with! But I'm not sure you can logically extend that to linking with conversions.

As a programmer foremost I think I am a pretty good marketer too; so perhaps that's why I find what your saying disagreeable. Indeed I work with someone [as a consultancy] who describes himself as a "people person" and marketeer (and is supposedly a good one). He can't, alone, sell my software for jack shit...

I think there is room in these analogies for a middle ground where we can pool our skills; programmers to explain it, marketers to give it that needed gloss.

Also I feel none of your points (in the blog) are rocket science; it does something of a disservice to programmers, who tend to be very bright, to suggest that is where they fail. I think the key points we fail are as follows:

- difficulty in writing good, non-technical, copy

- looking objectively at the subject matter

- spotting where a "newbie" to a site/software might get confused on it's purpose how it works (and then explaining or working round that)

On the other hand we tend to be good at things like solid SEO, keyword marketing, spreading the word in a grass roots way - etc.

The only thing I would agree with is design; a lot of people try to sell bare metal work with the promise of a "custom design". That never works - your better off with a sexy design and a few missing, soon to appear, features :)

There are programmers who ARE struggling with the basics, I agree. But Im sure there are good and bad people in marketing.

[DougBTX]

An example of how to reduce bounce rates: http://www.marco.org/237630497

[rjurney]

No offence, but you're making more statements out of prejudice than she is.

[patio11]

I am genuinely interested in hearing which statement you believe is prejudiced. I think I was fairly thorough in the above post in supporting my contentions with logic and evidence from my own experiences, but if I missed something I will be happy to revise or elaborate.