Brandt Dainow of Think Metrics reports on his area of expertise -- metrics, and on what's working, and what's not.

 

Useful links:

Recap of past tests 

Metrics overview 
View entire campaign 

This article is going to focus on the Sugarshots campaign iMedia has been using as a case study for the last few weeks. We at Think Metrics are providing the metrics for the site, so it is an excellent opportunity to share the process of web analysis with you. I’m going to walk you through some analyses step-by-step and show you how to cook with web analytics. This site is an excellent example of contradictory tendencies. It’s not my aim to give you answers or conclusions, but to pose questions for thought. This is a recipe, not take-out.

Preparation:

It always helps to know what you’re dealing with, so go to the Sugarshots site and look around:

www.sugar-shots.com. If you like, you can try going to www.sugarshots.com.  Notice you end up in the same place.  Boy, those guys at Shadomedia who built this site know a thing or two!

Now go to our reporting system: InSite. (If you see an error page, simply click the "Log-in again" link, and you'll get there the second time.) We’ve removed the security for the Sugarshots reports so everyone can see the metrics.  Feel free to click on a few things.

Set the Time

The reports in InSite default to the current month, so we need to change the date range to cover a common time period.  It’ll help if we’re all looking at the same time period, and thus the same data.

At the bottom of the screen you’ll see a link to “Change Date Range.” Click on this and set the time period to cover 1st June to 30th June, then click the "Set Range" button. 

Key Key Performance Indicators

Web metrics systems must present a wide range of information because they never know what you’re interested in at any given moment. This means there are always more items of information available than you need. These are all Key Performance Indicators of something. The secret to gaining understanding is to know what to focus on. You don’t want to look at everything every time. At the recent Emetrics Summit in London someone came up with the concept of Key Key Performance Indicators (or KKPIs) to reflect this. KKPIs are the Key Performance Indicators you’re interested in at that moment. The secret is to know what your KKPIs are before you look at your report, and then home in on them, ignoring all the other stuff around them.

First Impressions

Let’s start by getting an overview of how the site performed in June. Select the "Summary Report." The first KKPI here is the relationship between Visits and Unique Visitors. Notice the two numbers are almost identical. This means very few people make repeat visits.  A few lines down it indicates only 189 people out of 3,210 visited the site more than once.
Whether this is good or bad depends on the site.  At this point we’re not drawing conclusions, only gathering impressions.

Fact #1:  Most people only visit the site once.

OK, this site is not getting people to come back. So how committed to the site are they when they are here?

Commitment & Bounce Rate

I can see from the Summary that the average number of pages read in a visit is three, and the average time of a visit is three minutes. As a fact alone this doesn’t tell me much. Averages are dangerous because they can be skewed easily.  People fall into the mistake of confusing “average” with “most common.”  An average of three minutes does NOT mean most people spend three minutes on the site. In order to find out how long people are spending, we have to be a little more sophisticated.

Click on the "Visits" link. This will show you the first 20 visits in the month, listing how many pages, the visit duration and where each visit came from. Clicking [details] shows you an individual visit in a pop-up.  

Twenty visits isn’t really a statistically valid set, so use the sub-control on the left to display the first 100 visits.

Space forbids that I walk you through the process, but by exporting this table to Excel, I determine that 62 percent of people who visit this site only view one page. That’s the traditional measure of Bounce Rate, but I don’t think a site can achieve much in less than 60 seconds. 

Commitment vs. Bounce -- What is It?

This does raise the question of the relationship between the size of the site and what constitutes a committed visit. This site is only five pages in depth, so is it fair to demand 60 seconds as an indicator of commitment? Let’s understand the process in a visit -- firstly someone arrives and has to decide whether to stay and read or leave and look for a better site. If they decide not to stick, they have bounced. Controlling your bounce rate is about matching the initial appeal of the site to what people are searching the web for. You’re not telling them about your products, you’re telling them about your site. You have to sell the site before they can hear about the product. Once they’ve committed to the site you can start to sell the product. If someone bounces, they didn’t think the site offered what they wanted. If someone commits, but doesn’t buy, they didn’t think your product offered what they wanted. Different issues. 

I don’t think the number of pages in a site makes a difference here. Online reading speed is between 150 and 200 words per minute. The smallest page is the products page, at 190 words of copy. Allowing for entering the site, absorbing the navigation structure, clicking into one of the pages, making a commitment decision, I think we can argue that 60 seconds is the minimum time it takes in order for someone to absorb enough information for the sales process to commence. One of these visits read three pages in 13 seconds (row 93). That’s not a serious visit, and I can’t blame my sales pitch for not getting a sale from them, they never saw it. They were scanning the site to see if it was worth reading. Thus I think the key metric for bounce versus commitment is time, not the number of pages. Think about someone who spends five minutes reading only your home page, then leaves. They didn’t bounce, in my view that’s a committed visit.  For this site I’m going to include visits less than 60 seconds in the bounce rate.

Fact #2:  The site has a bounce rate of 70 percent.

Or to put this another way:

Fact #3: The Committed Visitor Index is 30 percent.

In other words, around one-third of the visits to the site are of value.  Further analysis of committed visitors tells us about their behavior.

Fact #4: Committed Visitors average five to six minutes, reading seven to eight pages. 

In my experience, it is usually the case that your committed visitors will spend about twice the overall average for duration and page views.

Sources

It’s important to know whether behavior varies by source or not. If it is the same, then the major determinant of behavior is site design.

The Referring Sites Report shows me where people are coming from. I can see that 45 percent of my visitors come from http://view.atdmt.com. (You cannot go to http://view.atdmt.com – it’s a tracking system, not a website.) This is the ad campaign Doug Schumacher’s been running as a research project through Atlas DMT. The next most popular source is “No Referrer” -- in other words, people who came directly to the site by typing the name in for themselves -- around 20 percent. This can include people who have bookmarked the site, or added it to their Favorites menu. Unfortunately it’s not possible to differentiate between these two. Google is next with around 10 percent. These top three sources account for 75 percent of all our visits, so we can focus on just these three for now.

If you click the "Visits" link to right of http://view.atdmt.com, you’ll get a list of the first 20 visits from that site. Use the sub-control on the left to select the first 100 so you have a meaningful sample.

Export this to Excel and run your favorite analysis on it and you’ll find the bounce rate for the Atlas DMT ads is 90 percent. In other words, only 10% of the people who clicked on an ad committed. Using Doug’s Atlas DMT numbers, I estimate this means 1 committed visitor for every 2 million ad impressions. Let’s hope the client never does that calculation!

Using the same process I can determine that the bounce rate for No Referrer’s is 65 percent, and 70 percent for Google. So it looks like the bounce rate for the ads is way too high, and it’s off to talk to Doug about how we can save this before we have to report to the client. But it does leave a question unanswered -- what’s a reasonable bounce rate for this site?

Baseline Bounce

Let’s consider the best possible case -- someone goes to a search engine and looks for this site. Surely these people are the ones who will have the lowest bounce rate? The Referring Search Phrases Report lists the phrases people used in all the different search engines to locate the site. Number two is “sugar shots.” These are people who were obviously looking for this site. Clicking the "Visits" link for this group, and then grabbing the first 100, reveals a bounce rate of around 20 percent. That’s the best we’ll ever do.

By comparison, the bounce rate for “liquid sugar” is 50 percent. The site is highly relevant to this search, yet still a high bounce rate, so it looks like we’re always going to have a high bounce rate.

Conversion Rates

The aim of the site is to sell. For the purposes of our study we’ll measure this by views of the purchase page. The Pages Report lists the number of views each page has. Row 4 is purchase.html.  Drilling down on the "Visits" link gives me a report on the visits which included this page. An analysis of the first 100 visits reveals that Google accounts for 30 percent, followed by No Referrer at 28 percent, then Atlas DMT at 10 percent.

The following table shows the conversion rates for our three main traffic sources:

Conclusions

It is typically reported that the average conversion rate is two percent and the average bounce rate is 40 percent. This is typically what I see, so in my experience this is an unusual site. It has a very high bounce rate and a very high conversion ratio. What does this tell us? I think what it tells us is that the marketing is picking up a lot of rubbish, but within that rubbish there is a seam of pure gold.

If I were responsible for the financial success of this site as an ecommerce site, I’d be concerned. The high conversion rate confirms what I believe when I look at the site -- it’s a good design. The high bounce rate tells me the marketing spend needs to be very intelligently targeted.
 
Please play with the metrics in InSite and tell us what you think -- about the site, about the campaigns. This is Open Source Marketing, let’s all learn together.

Brandt Dainow is CEO of Think Metrics, creator of the InSite Web reporting system. InSite delivers real-time sales and marketing information at the click of a mouse. Easy to customize, InSite comes in a shape to suit everyone. Brandt can be reached directly via email or on the web.