ellipsis flag icon-blogicon-check icon-comments icon-email icon-error icon-facebook icon-follow-comment icon-googleicon-hamburger icon-imedia-blog icon-imediaicon-instagramicon-left-arrow icon-linked-in icon-linked icon-linkedin icon-multi-page-view icon-person icon-print icon-right-arrow icon-save icon-searchicon-share-arrow icon-single-page-view icon-tag icon-twitter icon-unfollow icon-upload icon-valid icon-video-play icon-views icon-website icon-youtubelogo-imedia-white logo-imedia logo-mediaWhite review-star thumbs_down thumbs_up

Why your conversion rates no longer matter

Why your conversion rates no longer matter Brandt Dainow

The ultimate measure of a website's success is its conversion rate -- the percentage of visits that resulted in a sale or an inquiry. It is supposed to measure the degree to which the site converts visitors into customers. As such, it is deemed to provide the ultimate assessment of whether a site is successful or not. In a single number, it captures the appeal of the design, the ease (or otherwise) of navigation, the effectiveness of the sales pitch, and all the other factors that affect a visitor's willingness to buy from you.

Except that it doesn't.

The conversion rate tells us absolutely nothing about the site. It's a completely useless metric. 

As a professional web analyst, my job is to analyze a site's web metrics and determine from those numbers where, and what, improvements are needed in order to increase the rate at which visitors convert into customers. Doing this over and over again has taught me that the conversion rate is of no help at all.

Let's go back to basics and work this through. I think if you follow along with my reasoning, you'll see there's no point calculating the conversion rate. Luckily, I have an alternative I think you'll prefer.

The problem with conversion rates
Firstly, the conversion rate is supposed to measure the success of the site as a sales pitch -- an attempt to turn a browsing visitor into a customer. The problem with the conversion rate is that it calculates the percentage of all visits that resulted in a sale. This means it includes people who bounced. By definition, a bounced visit is one that looked at only a single page then left. In other words, they never entered the site, they never engaged with the material within it. Thus, bounced visitors were never exposed to the site's sales pitch.

There are two steps in converting a visitor into a customer. When someone lands on your site the first thing you need to do is convince them to stay. Only after that can you try to sell them something. It's like the shop window: You've got to get people into the store before you can try to sell to them.

The overall average bounce rate is around one-third. It's extremely difficult to get a bounce rate below 30 percent simply because search engines are far from perfect and can list the wrong sites. In addition, people frequently scan a number of sites before deciding which one to return to and engage with. It's also possible to be legitimately listed in the search engines for content you don't have.
For example, I have a client who owns a chain of restaurants. Like most restaurant sites, it includes the menus available. Due to some extremely high-quality search engine optimization, the site accidently became No. 1 in Google for a series of menu-related phrases, such as "set menu," "Christmas menu," and "menu ideas." The result was thousands of people arriving who were looking for menus they could cook at home, not a restaurant. Needless to say, the bounce rate for these people was extremely high -- around 80 percent -- and none of those who remained converted into restaurant bookings. This dragged the overall bounce rate for the site as a whole up to around 50 percent, which proves how important it is not to look at the bounce rate for the site as a single figure, but look at the bounce rate for each audience segment individually. 

It is especially important to look at the bounce rate for different referring keywords because this tells you how the site appeals to different audiences. In this case, we found the bounce rate for people who were actually looking for a restaurant was down around 25 percent, so the site was doing a pretty good job of holding onto new arrivals. However, the key point for our purposes is that all this inappropriate traffic, with an extremely high bounce rate, pushed the conversion rate down to less than 1 percent.  This makes it look like the site is a disaster.

When I see a conversion rate calculated from the total number of visits, including bounces, I don't know how much it's been affected by that bounce rate. In other words, I don't really know if I need to improve my landing pages and engage more arrivals, or if it's my actual content that needs more work.

Next page >>

A more effective alternative
The solution I have come up with is to work with what I call the retained visits conversion rate (RVCR). This is the percentage of people who did not bounce and then converted. Retained visits is not a metric in the standards, and you won't find it in something like Google Analytics. It's my own term for the number of visits that did not bounce. You get this from Google Analytics by subtracting bounces from visits, or you can get it at the dashboard with this formula:

Total Visits x (1 - Bounce Rate)

You then divide the number of sales transactions by the number of retained visits to get the retained visits conversion rate.

RVCR tells you how well your site is performing as a sales tool. Using the old system, you could have a good conversion rate with a poor site simply by having a low bounce rate. Alternatively, you could have a low conversion rate on a great site simply because the wrong people are coming to the site. You can't tell anything about the performance of the site's internals if you're including people who never got to them. You have to strip bounces out.

These days my analyses separate all visits into two groups -- bounces and retained visits. I analyze bounces to determine how well advertising is hitting my target audience and how well the landing pages hold them. I analyze retained visits to see how well the site is working. In other words, bounces are marketing analysis and retained visits are sales analysis.

Case study
Working with the retained visitor conversion rate provides better information when it comes to determining where to direct your efforts. The best way to see this is to run though a case study. The figures I am using are derived from a B2B client I have. They're close to genuine; I've just rounded off the figures to make things a little easier. The point is, this is a real example and shows how much difference using the retained visitor conversion rate can make:

Our sample site is getting 40,000 visitors per month from Google PPC and another 40,000 from native search listings in Google. Both produce the same number of sales and the same income. According to the traditional method of calculation, both sources of traffic have the same conversion rate, 0.8 percent, and both have the same average order value of $200. This means we get 300 sales and $60,000 from each. In other words, both sources look equally valuable.


The figures are summarized here:

Let's assume I can't get any more visitors. Therefore, if I want to increase income I have to increase site performance. Where should I focus my efforts -- PPC or listings? When I look at bounce rates, I discover PPC has a 30 percent bounce rate, while listings have a 45 percent bounce rate. When I recalculate my analysis, using RVCR, the picture looks very different:

This clearly shows the conversion rate for the two sources is not the same if I remove bounced visits from the picture. The conversion rate from people who actually engage with the site is higher for listings than for PPC. In other words, visit-for-visit, listings are worth more. It's no longer so obvious where I should put my effort.

I can aid that decision by working out what I'm losing from each source due to bounces. Let's assume that if I reduce my bounce rate, the extra visits will convert at the same rate as before. Using this assumption, I can calculate a nominal value for bounced visits. I'm currently losing 12,000 PPC visits and 18,000 listings visits. If both these convert at the same RVCR, I'm losing around $26,000 in sales from PPC and $50,000 from listings. In other words, the lost listings traffic is actually worth twice the lost PPC traffic.

Looking at the picture now, it's obvious I should be focusing on reducing the bounce rate for listings. Visits from listings are worth 25 percent more than visits from PPC because their RVCR is higher. When I calculated bounce rate in the traditional fashion, I was actually hiding major differences in performance between the two sources. A new picture has now emerged that makes it obvious I should focus on improving my bounce rate for search listings. Any given reduction in bounce rate will generate more income from listings than from PPC. 

To me this makes it clear we should forget about our overall conversion rate. It doesn't tell us anything, and it's probably misleading us, as it did in the above example. 

Analyze your conversion rate only for retained visits -- people who actually engage with your site. This provides a much clearer picture of what is really happening. Get with the retained visits conversion rate -- that's real information you can make decisions with.

Brandt Dainow is an independent web analytics and marketing consultant working in the U.K. and Ireland.

On Twitter? Follow iMedia Connection at @iMediaTweet.

Brandt is an independent web analyst, researcher and academic.  As a web analyst, he specialises in building bespoke (or customised) web analytic reporting systems.  This can range from building a customised report format to creating an...

View full biography


to leave comments.

Commenter: pat mcgraw

2009, November 07

When you write "Due to some extremely high-quality search engine optimization, the site accidentally became No. 1 in Google for a series of menu-related phrases...", am I correct to assume that you are being sarcastic?

After all, "high quality SEO" wouldn't deliver poor quality leads - right?

Commenter: Chris Brinkworth

2009, September 25

Im following - but would suggest an additional layer:

Where are the majority of bounces coming from? IE - Yahoo, Bing or Google ? You should look by refferer at the bounce rate there also. IE - Bing may be sending perfectly relevant people - so conversion % is a true reflection. Google however are not, so that part of the equation need's adjusting. Also - WHY are they bouncing? Is there a pixel that you are serving or Google visitors that is causing latency on page load?

Using tools like TagMan (conditionally set logic on the fly, automatically, as the page loads) - help here. IE - if reffer=google and words = X, then do this.

Great article. Thanks. Made me think for the day!

Commenter: Chris Goward

2009, September 25

While I agree with you that conversion rates can be misleading, I disagree with your rationale.

Conversion rate measures should *absolutely* include bounced visits. Otherwise you're negating the importance of the landing page, and the landing page is critical to the sales process.

You said, "By definition, a bounced visit is one that looked at only a single page then left. In other words, they never entered the site, they never engaged with the material within it. Thus, bounced visitors were never exposed to the site's sales pitch."

Why would you think someone hasn't been exposed to the sales pitch until they see the second page? If you can't present your value proposition clearly on one page and entice the visitor to take an action, you've already identified your problem. Your landing page sucks!