Dec 12 2011
An interview with Martin Eising
It puzzles me why so few businesses really attempt to optimize conversions. Why is that?
I think it has to do with the way people think about web sites. When someone wants to increase the action they get from their site the first thing that comes to mind is increasing the amount of visitors the site gets. This is just logical, and is naturally the first thing that comes to mind. Doing more with the same amount of traffic , on the other hand, is not something that most people immediately think of.
Also, when people commence SEO and marketing strategies to increase traffic they either hire someone in-house, or more often they hire an outside SEO/Marketing company. Many of these companies do not suggest to their client that they should consider conversion optimization: they just go ahead with the SEO campaign and get on with it. This is especially the case when the SEO/marketing firm does not have the expertise required to optimize for conversions.
So, how do web analytics tie into getting a higher conversion rate?
I see three ways by which web analytics can be used to increase conversion rates: identifying issues that should be addressed; identifying your most important web pages that should be tested; analyzing ROI data to determine where your PPC dollars are best spent.
By identifying and fixing web site issues users will end up having a better user experience. As a result they should be staying at your site longer, and the conversion rate will increase.
Some potential problems that web analysis can identify are:
- High bounce rates. Emphasis should be placed on especially important pages in a site. If an important web page has a high bounce rate then the content of that page should be examined. Is the page missing information that the visitor is looking for? Does the page have inappropriate messaging that is driving visitors away?
- Web site errors. Most sites have a custom error page, and if users are ending up at an error page then there is one or more problems with the site. Note that using a combination of web analytics and Google Webmaster tools is the best way to identify site issues like this.
Web analytics can also help to identify your most important web pages, which should undergo AB and/or multivariate testing. This testing can be done via Google Optimizer, which is a free, easy to use tool. The basic premise is that you define alternate content for the page being tested, and all possible permutations and combinations are then served up to end-users. Google keeps track of the conversion rate of each possible combination, and when the test is complete the combination of content that has the best conversion rate is then implemented for that page.
Web analytics is used to help identify which web pages to test (of course, common sense can also be used). I look at pages that:
- Have a high number of Pageviews.
- Have a high $Index value (the $Index value indicates how often a page is part of a visitor’s path that led to a conversion. The higher the $Index value, the more important the page is to the conversion process).
- Serve as a landing page (e.g. for a PPC campaign).
- Are within a funnel (note that funnels may not be appropriate for all goals/conversions).
The end result of all this activity is a better user experience, which in turn results in higher conversion rates!
Finally, I would analyze ROI keyword data to determine areas where AdWords expenditures result in the best ROI. For example, I suggest looking at paid keywords (the Keywords report in the Traffic Sources section) and examining conversion metrics. Keywords that are resulting in a low conversion rate can either be deleted or have their bids lowered to increase their profitability. Conversely, I would consider increasing the bid amounts of keywords that have good conversion rates.
I think you described multivariate testing fairly well, but not AB testing so much. Do you think there are some traffic limits for using multivariate vs. AB testing?
Let me clarify the difference between the two. AB testing refers to testing one page versus another to see which page structure converts the best. For example, a header with a 3 column layout versus a header and two column layout (oftentimes this is done before multivariate testing). Once you’ve determined what sort of page structure appeals most to end-users then you can implement multivariate testing for the winner of the AB testing. The end product is a web page that maximizes conversions.
As far as traffic limits go, I think that applies more to multivariate testing. It is important to limit the number of combinations to 25 or less, otherwise it is going to take a long time for the test to be completed. Of course, the length of time it takes the test to run is affected by the amount of web action you get, so this should be taken into account when “chunking” a page for multivariate testing.
Web analytics contain so much info and have lots of interesting charts and graphs sometimes they can be confusing and/or a rabbit hole. Can you give us some tips on how to take advantage of all the data in web analytics?
Well, when it comes to using analytics as a reporting tool I would encourage people to take advantage of report scheduling. This can save people time while keeping those that need to be kept in the loop up to date with regular reports. Oftentimes these reports involve the number of web visits, Pageviews and goal conversions.
People that have Google AdWords accounts and are using Google Analytics should also make sure that their AdWords account is linked to their Analytics account. This enables an AdWords sub-section within the Traffic Sources section of Google Analytics, and offer lots of valuable metrics regarding AdWords expenditures, clicks, etc.
Finally, to fully leverage analytics I would encourage people to define goals along with goal values. Increasing web site action is one thing; increasing conversions is another! Also, defining goals and associated values makes available lots of valuable AdWords related metrics such as ROI (return on investment), RPC (revenue per click) and margin. These metrics are very helpful when analyzing your PPC (pay-per-click) campaigns, and automated reports can be set up so that this financial data is sent to managers, financial officers, etc.
Martin Eising, the article’s author, has been in the SEO (search engine optimization) field since 1998. Since that time he has widened his knowledge base, and is also proficient in SEM (PPC like Google AdWords), web analytics (e.g. Google Analytics), advanced testing (e.g. Google Optimizer), and solutions-based, emotive response messaging.