I recently had the pleasure of working with Jim Sterne, speaking at the eMetrics summits in London and Munich. In both cities it seems the attendees had the same problems. They know they need web analytics, which is good. And they had the tools, which is also good. But they didn’t know what to measure, and seemed overwhelmed by the amount of data and range of reports you can get from an average web analytics system. Enterprise level companies seemed to have an even tougher challenge. They not only struggled with figuring out which reports to use, but also with who to give them to. I also listened to countless stories of analysts who, having
presented comprehensive reports, faced an uphill struggle to get the stakeholders to take concrete action.
To help clarify this dilemma I offer the following analogy described to me by Vincent Kermorgant, a guy who is heavily involved with web analytics strategy at Nokia.
Web analytics strategy can be as simple as driving a car
Vincent told me how he explains to people at Nokia what Key Performance Indicators (KPI’s) are. He described how he uses a car’s dashboard as the basis for the explanation. Allow me to elaborate.
Imagine your business is a car and you, the CEO, are the driver of that car. As the driver you control the direction the business is going in. As the driver, you need to know two things: First, what speed you’re moving, and second, how much fuel (resources) you have so that you can get from point “A” to point “B”. Other things that will interest you as a driver are warnings, for example, the oil light flashing.
Now start to think of your car’s systems as independent business units. Think of the engine and how many times the pistons moves up and down in the cars engine. The person in charge of the ´engine business unit´ will be interested in how many times those pistons move up and down because it’s an important gauge of the engine’s performance. The driver of course doesn’t care how many times the pistons move, only if the car works or not. Imagine how silly (even dangerous) it would be to report the number of piston revolutions on the car dashboard. This is the key distinction, and shows why giving all the information to all the people will just result in no action being taken - as no one is prepared to wade through numbers to define and locate the information that is relevant to them.
Giving the right KPI’s to the right people
Now that I’ve explained the analogy lets get back to developing KPI’s for websites. It’s all about reporting the right things to the right person. The CEO is interested in figures which tell him about the profitability of his business, not how his engine is performing. So give him the right gauges of performance. Give him the numbers that keep him informed about what direction the business is heading in and how long it will take to achieve business objectives.
The business unit needs information about how its particular operations are working. The information you give to the CEO is nice to know for the business unit but not critical to the day to day operations.
his article that the driver would be interested in warning signals as well - like the oil light flashing on the car’s dashboard. The reason for this is because the driver would logically then take action after seeing this warning by filling the engine up with oil, or have to face the negative consequences.
This can also be applied to a website KPI’s. I have personally used ‘page views per session’ as a critical flag of performance for websites. If you have a website which has a five page purchase process, then the minimum average page view per session has to be five pages per visitor or your website is not performing well enough.
Like the dashboard’s oil light coming on, you should pay attention when this sort of figure drops or rises for whatever reason.
These flags should also be applied across different audience segments. It’s very useful to know the difference between your website visitors.
KPI’s and behavioral flags
Eric Peterson recently wrote a book called The Big Book Of KPI’s, which is a great resource for businesses interested in determining KPI’s and flags for their own business operations. I am not trying to redefine what Eric wrote because he actually says in his book that reports should be allocated to the people that need them and also says that you should look at what’s important for your business. However, I do feel that there needs to be more distinction made between the business figures that the driver wants and the flags or trip wire metrics that the analyst sets up.
KPI’s and flags in my opinion should be treated differently. The KPI is a number I use to tell the business (or business unit) “driver” something effecting business performance. The flag is something that tells the web analyst working for the business (or business unit) that there is a problem which needs further investigation.
In order to get the stakeholders to take action you need to present concise reports containing numbers that are relevant to them. It could be that for the CEO a paper containing five or six key metrics is more important than giving him 100 reports across all website operations. As Eric said in his book, what’s important for one guy is not important for another. Simply align the KPI with the person and you will start seeing action being taken. Action taken on the metrics is what you want and is an indicator that you are using web analytics intelligently.