Archives for January 2010

Is Your Conversion Rate Wrong?

According to Web Analytics Association (WAA) standard definition, a conversion is a visitor completing a target action.

Conversion rate is calculated by total number of visitors completing a target actions divided by a “relevant denominator” (Sometimes it is total number of target actions divided by a “relevant denominator).

The key here is “relevant denominator”. This is often ignored by many organization and they tend to use the default denominator provided by their Web Analytics tool provides. The most common denominator used by most of the tools is either total visits or total visitors.

If you are an eCommerce site then potentially every visitors or visit is a target for conversion and “Total Visits” or “Total Visitors” as a denominator might make sense. However if your site is a non-eCommerce site and conversion to you means getting people to register on the site then using a denominator like total visits or total visitors to calculate conversion rate, wrongly assumes that all of the visits to sites have not previously converted (registered) and hence are potential conversion worthy. We all know that’s not the case. These sites get a mix of visitors, visitors who have registered in past and are not going to register again and, visitors who have not registered yet and hence are a potential target. Examples of such sites are facebook, WSJ.com, espn, yahoo etc.

Calculating the Right Conversion rate

In my examples I will assume that number of visits is equal to the number of visitors.

Hypothetical site in our examples is a portal that requires visitors to login (register) for some actions but also has content that does not require a login.

This sites gets an average of 1000 visits a day and gets 50 new registrations (conversions) a day. By most common way of calculating the conversion rate, the conversion rate for this site is 5% (i.e. Conversion Rate = Conversions/Total Visits = 50/1000 = 5%)

Let’s assume that every day this site gets 50% of the visits from the people who have registered in past. So that means 500 (50% of 1000) are registered and won’t register again. Since they are not going to register, why do we even consider them when calculating the conversion rate? By removing them from the calculation of conversion rate, the new conversion rate becomes:
50/500 = 10% (Immediately you doubled your conversion rate)

Why is this important?

Let’s look at two scenarios to illustrate the importance of choosing a relevant denominator.

  1. Scenario 1
    One day the customer retention department sends some emails to previously registered visitors that result in 500 more visits from people who had previously registered. As a result the site got 1500 instead of 1000 visits it used to get every day. Considering that nothing else changed on the “not-registered” visitor base, the site got 50 conversions, just like any other day.

    Using “Total Visits” as the denominator, it appears that the conversion rate drops to 50/1500 = 3.33%

    It looks bad, and might cause you to panic. Won’t it? However if you choose the right denominator you will find out that nothing really changed. The pool of people who had not previously registered is still 500, so the true conversion rate is still the same
    50/500 = 10%.

  2. Scenario 2
    The “customer acquisition” department bought a new email list and sends emails to this list which resulted in 500 extra visits. Overall the site got 1500 customers, 1000 who had never registered and 500 who had registered previously. That day the site got 60 registrations, resulting in a conversion rate of
    60/1500 = 4%.

    Look like our email list did not work because it caused the conversion rate to go down.

    This might cause you to wrongly assume that the email list that the retention department bough was not as good as the other traffic that you have been getting. However when you calculate the true conversion rate, it turns out that the email list actually worked; it resulted in a conversion rate of 6% (60/1000).

Note: Some might argue that why won’t this company tag all the emails with proper campaign identifiers so that we can track the performance of the emails. In theory this sounds great and that’s how it should be. In reality though, many departments work in isolation and never interact with the ‘Web Analytics” or follow their best practices. Now you know why marketing departments should share the marketing calendar/activities with “Web Analytics” group and involve them before running any campaign.

Hope this gives you an a reason to investigate what denominator you are using in your conversion rate.

In the next post, I will show you how you can calculate the “True Conversion Rate” using Google Analytics.


New Job: Analytics Associate at Huge (Brooklyn, NY)

Looking to fill your Web Analytics or Online Marketing position?

Post your open jobs on Web Analytics Job Board
————————————————————————–

Bounce Rate: All That Bounces is Not Bad – Part II

Most likely, you have all heard that high bounce rates are bad. Some people even believe that bounce rate affects your sites search engine rankings. Meaning higher bounce rate is bad for your organic rankings. I am not sure if that actually is true or not but it does show you that most analysts consider high bounce rates as a negative indicator of a site’s health.

Are high bounce rates really bad? Generally they are but you have to consider the other factors and put everything in context before you can come to that conclusion. In my post “All that bounces is not Bad”, I showed you that there are cases when the bounces are not really bad.

One of the causes of high bounce rate is the entry via “destination page”. A destination page is a page that provides the information that the visitors are looking for.

Destination pages are going to have higher bounce rate and no matter how hard you try you might not be able to move the needle on those pages. For this reason, most of the news sites, content sites, product info, support site, portals etc, are going to have a higher bounce rate than the ecommerce or lead generation sites. Visitors come with a mission e.g. to read an article, news, get information etc. and once they read the article they go back to where they came from (think search, twitter, facebook etc.).

Let me show you an example of a page that has a very high bounce rate. The page in this example has a bounce rate of 72%, which means that the page just sucks. Do you agree??? Let’s put this page in context and then evaluate.

Firstly, the page in question provides very specific information and is refreshed every week, sort of like a weekly horoscope page. To understand the visitor behavior I created a segment of visits that entered this page and then bounced.

As shown above 26.65% visits are “New” which means that 73.35% of visits had come to the site in past. That’s a lot of returning visits. It tells me that a lot of visitors are already familiar with the site and are coming back with specific mission and then bouncing.

To further understand how loyal these returning visitors are, I created another segment by adding additional criteria to just focus on the returning visits.

Over 90% of the visitors that bounced have come 3 or more times to the site. About 67% of these returning visitors have come 9 or more times on this site. About 30% have visited more than 50 times since they were first cookied.

About 90% of the visitors visited the site within last 7 days. Not bad. Since this page does not change within 7 days, it appears they might have visited the other pages on the pages on the site. However I have not found a way to prove that in Google Analytics.

What is this data telling me? It is showing me that the visitors that are bouncing off of this page are actually pretty loyal and are coming back to the site quite often. So what if they bounced in their latest visit? Now the 70% bounce rate does not seem that bad. What do you think?

So what’s the lesson? Don’t just look at the bounce rate and panic. Look at the other data points and then put everything in context before you jump to any conclusions. Remember “All that bounces is not bad”.

Comments?


New Job: Analytics Associate at Huge (Brooklyn, NY)

Looking to fill your Web Analytics or Online Marketing position?

Post your open jobs on Web Analytics Job Board
————————————————————————–

Predictions 2010: Web Analytics, Social Media and Digital Marketing

This is the time of the year to make few predictions for 2010. Here are my 5 predictions for Web Analytics, Social Media and Digital Marketing.

  1. Website Testing – What’s the point of creating a cool looking site that does not engage or covert the customers? Website Testing (A/B and Multivariate testing) is the key to those conversions, Website Testing won’t be optional in 2010. Website testing will become an essential part of website design and development. Website designs will make it easier to change the elements of page and hence make it easier to test them. Jobs in A/B and Multivariate testing are about to rise. For your reference there are currently 252 jobs for “A/B testing” on simplyhired.com.
  2. Web Analytics – I have two predictions about web analytics (generally known as web analytics):
    1. Adobe will start incorporating Omniture into its products.
    2. WebTrends will be sold this year.

  3. Mobile Analytics – Mobile usage will continue its upward trend and so will the number of mobile sites and mobile applications (apps). Just like a website, mobile sites needs to be analyzed and optimized and same goes for the mobile applications. There are already tens of thousands mobile apps and this number will continue to grow. It is critical that users download your apps, number of downloads is an important measure of success. But number of downloads do not equate to usage. If your app does not get used then what’s the point of paying thousands of dollars to develop them? Usage of your apps is even more important metrics. As the market gets crowded with all the apps, app providers would want to know if there apps are getting used and what features are getting used. The need to learn from the data and optimize the apps will rise this year. See above, Optimization is not optional.
  4. Twitter – Twitter will need to prove its value to the businesses. There is a lot of noise on twitter, most of the clicks (and I have proof) on twitter links are not from real people. Is anybody reading those tweets? Yes tweets are now appearing on search engines but are they adding much value to the search results? This will be the year that will define the future of twitter. Twitter will either be sold or dead by the end of this year.
  5. Privacy – Privacy will become a very hot topic putting a lot of pressure on behavioral targeting. Mobile Tracking will also be scrutinized. Privacy watchdogs will start looking at the data collection by Mobile apps (e.g. iPhone apps), desktop widgets/pass etc.

What do you think?
Related Posts


New Job: Analytics Associate at Huge (Brooklyn, NY)

Looking to fill your Web Analytics or Online Marketing position?

Post your open jobs on Web Analytics Job Board
————————————————————————–