7 Tips for High Value Analytics

The analytics industry continues to grow, and more and more Webmasters and Web site owners are relying on analytics tools to provide them with information on how their Web sites are performing. This article will discuss some of the biggest issues faced by companies trying to get the most out of a Web analytics solution.


For many Webmasters, analytics activities are limited to checking basic traffic numbers, such as the number of visitors, unique visitors, and page views. These are great numbers to know, but this is not where high value analytics lie.

In addition, many organizations have their analytics work done in a silo, separate from the rest of their organization. The person, or people, who look at the analytics do their own thing, get some value from it, and then move on. Once again, this approach results in limiting the benefits of Web analytics.

Another issue is the tendency to assume analytics data is always accurate. It is, after all, a computer program measuring a variety of signals that are coming into it from computer hardware – so there should be no errors, right? Unfortunately, as I learned in a recent analytics study we did, this is far from the case. There are, however, ways to deal with the accuracy limitations.

Key Steps for Maximizing Your Analytics Investment

  1. Decide on Your Site Objectives: Before even beginning to decide how you want to use your analytics tool, you must have a very clear idea of what you are trying to accomplish with your site. For many people, this is something like “sell green widgets,” or “generate leads.” This may sound simple, but it’s a critical step.

    You want to focus your analytics effort on meeting the objectives of your Web site and not on anything else. Given the complexity of analytics, it is easy to get lost in investigating little mysteries instead of focusing on things that make you money. Being clear on your site objectives helps you get around that.

  2. Pick Actionable KPIs: Now that you are clear on your objectives, you need to decide what to measure. The key test for deciding whether or not you want to focus on measuring a variable is whether or not changes in that variable will cause you to take action. In other words, if the KPI you are measuring goes up or down by 20 percent, will that cause you to take action?

    This ends up being the second major component of maintaining your focus when using Web analytics. Cruising around reading canned reports is not the answer (this is what Dennis Mortensen refers to as report surfing). The way to go is to focus on those measurements that provide insight into your business and tell you how your Web site is performing. Attempt to do this up front, and then update your selected KPIs as you learn more about the information you can extract about Web site performance from your analytics program.

  3. Invest in a Business Analyst(s): Analytics pundit, Avinash Kaushik quite famously offered up the 10/90 rule. The basis of this rule is that 10 percent of your analytics investment should be in the software or service, and the remaining 90 percent should be on the people who use it.

    At the heart of this rule is one simple fact: High value analytics is hard. You can’t really think of it as taking an off-the-shelf tool, installing it, reading the reports, and then you’re done. Organizations that get the highest ROI on analytics are those that have really smart people focused on deriving value from their tools

    You need the right type of people to do this work, too. The business analyst who can do this work well will be a person who has a sharp technical aptitude, as well as a strong marketing and business sense. This person must have both pieces of the puzzle in place to be well suited for this type of work.

  4. Create a Data-Driven Organization: Eric Peterson was the first person to get me focused on the notion of data-driven organizations. The first three steps above were all focused on making sure that you use your tool in the optimum manner. But this won’t get you anywhere if your organization is not set up to act on the results.

    Not acting on actionable data is a common problem in many companies. To deal with actionable data in your business requires a cultural commitment to act upon the data as appropriate. For example, if the site’s sales conversion rate drops suddenly by 30 percent, people must be ready to act on this information.

    Perhaps marketing needs to investigate whether or not competitors have dropped prices, or management may need to reconsider a recent change in the terms and conditions offered on your site. Sudden KPI changes can affect many different departments. The key is to set up the organization so everyone is looking at the key numbers affecting them, and accept that responding to significant changes in the numbers is part of their job.

  5. Verify Your Implementation: One of the biggest sources of error in analytics is implementation error. This can be pages that do not have the analytics JavaScript tags on them, or pages that are incorrectly tagged. Either way, this can throw your data analysis into a tailspin.

    The best way to deal with this is to carefully verify your implementation. Treat the addition and/or modification of your analytics JavaScript tags the same way you would any other software development process. This is a source of error in analytics that is completely under your control, so take the time to eliminate it from the picture.

  6. Monitor Accuracy and Cross Check: Now that you have eliminated implementation error, you are indeed better off. However, there can still be significant errors in analytics. This stems from the poor quality of information coming into the software. For example, many proxy servers strip referrer information, and many people use services like AOL where the IP address they are coming from changes in the middle of a given session.

    There are many such sources of error, and each analytics package decides how to deal with these types of anomalies in different ways. The end result is many judgment calls, some of which are going to be wrong. In spite of these errors, the data you get still has incredible value, but you have to limit the scope of how these errors affect you.

    One way to do this is to use alternative types of tools to cross check your data as much as possible. For example, you wouldn’t use your analytics solution to count the number of orders you get from your PPC campaign, as it would miss some of the orders.

    For this, you are better off tagging the end of the URL used in your PPC ads with parameters that your Web application can read during the ordering process. Then you can use your Web application to tally up the total orders, focusing your Web analytics on other tasks it does better than absolute measurement.

    What’s that, you ask? Web analytics are great at relative measurement. If the tool provides you with a set of numbers over time, and you look at relative changes in those numbers, you will be looking at very useful and very accurate information. More on this in my next point.

  7. Focus on High Value Add Analyses: Bearing in mind the true value of Web analytics lies in relative measurement, you want to use the tools where that strength is leveraged. Here are some examples of such activities:

    • Segmentation: Break your visitors into segments. See how the behavior of the different groups differs. I went into the topic of segmenting visitors in much more detail in last week’s column, so check out that article for more on this topic.

    • A/B Testing: Comparative performance of different landing pages, different price points, different ad copy, different product offers – these are all gold mines. Most Web analytics packages make this easy to do.

      How much can you gain from this? Think of it this way. No matter how experienced a marketer you may be, the chances of designing the penultimate marketing campaign, ad copy, keyword buys, landing pages, etc., on the first try using only your experience is essentially zero. Virtually every Web site can benefit from doing this type of testing.

      Of course, testing is a non-trivial investment, and you need to have a really good business analyst on board to get the best results.

    • Measure Keyword Performance: You can use your analytics package to find keywords that offer a higher than average ROI, and you can use it to find poor performing keywords. With this information, you can tailor your PPC campaigns and drive up the ROI.


As I said, high value analytics is hard. Of course, for smaller Web sites, it is difficult to justify making the investment to do some of the things I outlined above. But as your site’s revenue base grows, analytics can offer a very high ROI on your investment in the tools and the people using them. Keep these seven key steps in mind as you begin that journey, and last of all, be patient.

Finding the optimum way to use analytics on a relatively large site requires experimentation and exploration. Start with the right approach, be patient, and evolve your strategy as you learn more about your Web site, your customers, and the analytics tool you are using.

Related reading

Simple Share Buttons