A great deal has been said and written about the good, the bad and the ugly of web analytics. Niel Bornman, Acceleration's Director of Online Analytics boils the debate into five of the biggest challenges we face as we implement analytics tools and the ways that we can deal with them.

1. Overselling and underdelivering

Our problems start when a slick sales person from a web analytics company sells you the solution to end all solutions and you buy into it without considering the practicalities of implementation and adoption.
This sets you up for disappointment from the start. Even worse, your enthusiasm for the solution may have led you to sell the tool to your boss and colleagues as the answer for all online marketing woes. Now you've sold the solution so hard, your boss and the rest of the company have completely unrealistic expectations of it. To top it off, your boss may know even less about online analytics than you, but was led astray by the same sales person and your support for the tool.

All is not lost, but you will need to move fast to correct unrealistic expectations. Arm yourself with the knowledge to understand what can realistically be achieved within your organization.

If you are working at a large corporate with many web sites, marketing initiatives, teams, etc, it might be difficult to effect change in the short term and you should take a longerterm approach. If your company is small and nimble, then it might be easier to reset expectations and you may have a shorter timeframe to delivering some business value.

2. Working with data discrepancies

Second on my list are data discrepancies. Nothing kills the success of an online analytics solution quicker than distrust of the data, especially if you also suffer from problem 1.

To add insult to injury, the need for financial balance requires all numbers to always balance out. If the numbers don't balance, people suspect that you're doing something sinister, that the product isn't working or that you don't know what you are doing.

When confronted with a data discrepancy between your email system and your web analytics tool, the first thing you'll want to do is hunt down the problem and fix it. After a couple of months and two nervous breakdowns, you'll probably come to the same conclusion that Avinash Kaushik did in his article "Data Quality Sucks, Let's Just Get Over It."

As a first step to addressing this problem, realize that you are probably not comparing apples with apples.
Reporting methodologies between tools differ and they may even track different things but they are given the same name. One example is web analytics solutions that claim to track click throughs from emails, but actually either track server requests or page views (in Net terms) long after the actual click through occurred.

Even though the data does not balance, the important thing is that it stays within an acceptable range that allows you to make decisions based on the trends. Arm yourself with knowledge to understand the differences and equip yourself to better deal with the next discrepancy. It does help to work with a consulting partner that has dealt with these situations before.

3. Not enough action

Everyone knows you should use data to make better decisions, that you should have leading indicators, that you should look at key market segments, but how many of us actually do? How many marketers progress beyond simply looking at data and sending out a report? How many actually take action?

Complacency and the inability to take action probably spring from too much data and too little focus. Take some time to identify your key performance indicators and how to improve them. Take action! Don't fear the unknown and failure. Learn to view change as a good thing and help others to see it in a positive light, too.

Ask yourself, "What would I do if I weren't afraid?" The worst that could happen is that you revert back to the old system because the changes you implemented didn't have the desired effect. You can always start again and look for a new answer.

4. Integration headaches

The fourth problem on the list ties in with the top three and is the lack of data integration between the various online tracking tools. Many vendors are promising integrations between all kinds of data. In reality, however, how the data is collected and when it is collected makes it difficult to integrate data sources as seamlessly as we'd like.

This problem has no easy solution, but many smarter vendors are working non-stop to overcome this hurdle. They realize the future of their products may depend on whether they are able to provide you, the marketer, with the ultimate solution that combines all the online tracking that is currently available and also supplies you with leading indicators and data mining on the fly capabilities.

Work with a vendor and consulting partner that can help you overcome some of your internal integration problems.

5. Reactive focus

This leads me to the final item on my list, automated data mining. Amidst all the terabytes of data and all variations in behavior, how do you find the critical segments and key behavioral patterns? How do you get to know whom the people are that interact with you or (better yet) don't interact with you?

Online data mining is currently driven by human analysis, as is most data mining. The problem is sifting through the volume of data and translating the results into meaningful action. Another key problem lies in understanding all the variables at play well enough to be able to use one of the more traditional data mining applications.

To predict what will happen, you need to understand what happened. But what if you could "accurately" predict the impact of a small change to your next campaign, or better yet, not having to know how a specific segment wants you to interact with them online, simply trusting your solution to know how and to do it?

Data mining and predictive modelling applications have been around for many years and there are some that can be used on web data. However, it would be great to see the leading web analytics vendors step up and start offering such solutions.

In the meantime, you can start investigating some of the solutions available and if you can overcome problem 4, you may start to see some viable proactive results.

Closing words

All these problems show that the industry is in its infancy and that there is a lot of education to be done. Accept the fact that the data will not match 100 percent, but educate yourself to know how to use it anyway. Take continuous action. Accept that everything will not work, but know that if you don't start making changes now using testing platforms you will continue to see the same conversion or worse than you are now.
By Neil Bornman,
Acceleration's Director of Online Analytics at Acceleration eMarketing
(http://www.acceleration.biz)