I'd guess most of you who have web analytics configured for your site are paralyzed by reports. You're adrift in a sea of numbers, at risk of drowning; hoping management will be distracted by another colorful pie chart and not notice. Every analytic solution can generate a perfect storm of data that obliterates the horizon in innumerable details. Where's the insight? What's important? What should be done next?
Reporting isn't analysis. That's far from obvious for most of us. Our analytic tools are very good at reporting. It's a trivial effort to generate another report; it's a good deal of work to analyze the significance of the data and test your hypothesis. It's little wonder we're willing to confuse one with the other.
I'll use my own example as representative. Several years ago I was hired to run the digital advertising initiatives for a company that had pioneered digital projectors. I needed to manage eCommerce, display advertising, bulk email, SEM, the lot. I wrote specifications for web analytics and sold management on a high end solution that gleamed with more chrome than a 1949 Chevy Coupe Deluxe. The program generated an impressive number of reports which I dutifully copied and graphed and forwarded to management who dutifully ignored them. There was simply nothing they could use, nothing they could act upon.
For a metric to be useful, it must be actionable. If you can't
influence it, for good or ill, it's makes no difference.
The work of web analysis normally goes something like this. The performance of a particular metric randomly draws your attention. It might be an alarming number of visitors failing to complete your check out process or the lackluster performance of the landing page for an advertising campaign. You develop a hypothesis that attempts to explain the behavior that you're witnessing. You take action to either reinforce or inhibit that behavior, depending upon your goal. You test your hypothesis. Then you wait and watch to see if the behavior moves in the desired direction. If it goes as expected, you're a hero. If not, you quietly revert your test and hope no one noticed. Actually, a test that fails to confirm your hypothesis is just as valuable as a success. It can help inform your next hypothesis.
Choose Your KPIs Wisely
There are plenty of metrics you could mess with but you probably have time for only a few, a manageable few. Those are the six or seven metrics most important to your organization's goals, your key performance indicators (KPIs). A KPI is a stratospheric view of your site's health. To understand the reality on the ground, you'll need to get your hands dirty with the details. KPIs are directional, pointing to where you should start grubbing.
Your KPIs will vary with the purpose of your website and your business goals. For a commercial site, KPIs are most likely related to revenue. I've borrowed the commercial KPIs recommended by Jason Burby and Shane Atchison in their book Actionable Web Analytics (Wiley Publishing, 2007).
- Overall Conversion Rate
- Average Order Size
- Items per Order
- Step-by-Step Analysis of the Conversion Funnel
- Conversion Funnel Defections
- First-Time Versus Returning Buyers
- Effect on Offline Sales
Your purpose is to move the needle in the desired direction.
You will have established a goal for a KPI - or more likely,
your manager will have set the goal - but influencing the
trend is the real work of web analysis.
Conversion rate is immediately obvious - increasing the percentage of visitors who do what you hoped they would. Increasing the average order size and items per order both indicate greater yield per conversion and the efficiency of your online sales process. Analyzing the conversion funnel - the sequence of steps necessary to complete a purchase - and reducing the percentages of customers who drop off from one step to another can have a substantial influence on your overall conversion rate. Repeat customers cost less and provide a good indicator of customer satisfaction. And finally, your website may have an impact on offline sales, whether store front or phone center sales. Using discount codes or toll-free phone numbers that are available only on your website can help measure that impact.
You'll have noticed the importance of trending in the definition of Burby and Atchison's KPIs. Your purpose is to move the needle in the desired direction. You will have established a goal for a KPI - or more likely, your manager will have set the goal - but influencing the trend is the real work of web analysis.
Non-Commercial KPIs
For content only sites that don't sell anything Avinash Kaushik, author of Web Analytics An Hour a Day (Wiley Publishing, 2007) suggests the following KPIs:
- Visitor Loyalty
- Recency
- Length of Visit
- Depth of Visit
Visitor loyalty is the non-commercial equivalent of first-time versus returning buyers. It provides an indication of the success of your content. Recency measures the time between visits. Length of visit is the measure of how long visitors spent on your site per visit while depth of visit is the number of pages viewed per visit. Depth of visit is the KPI formerly known as page views which has gotten a lot of bad press recently but remains a powerful metric when placed in context. In all, these four are a good indication of your visitors' engagement with your site.
Segmenting each of your KPIs into several different groups-by referrer, for example, by advertising campaign, or demographic if you're using adCenter Analytics-is a powerful technique. Drilling down into the data and analyzing the behavior of groups of visitors defined by a common characteristic can provide a more granular insight into your visitors' intent and understanding intent is the first step in providing relevance.
KPIs change over time. They aren't chiseled in stone and delivered by Charlton Heston. They need to adapt to the competitive landscape and your organization's changing goals to remain relevant.
The Bottom Line
KPIs provide a view of the health of your website, pointing the direction for further investigation, analysis, and testing. They offer a manageable handle on the complex reality of your site, a handle useful for both analyst and management. Choosing the right KPIs has a big impact on managing the success of your site and your business.