Actionable Web Analytics: A Three-Step Process

Aug 22, 2014

250-x-188-Actionable-Web-Analytics-City-Bike-Riders
As Web analysts, your customers and their needs can vary greatly. One minute you are leading an analysis that will influence strategic decisions, the next you are distributing reports to folks who will never use them, but just have to have them. Sometimes, in the darkest hours, it can become more of the latter. That’s when you risk falling into a routine; when you risk forgetting the real power of Web analytics—effecting change.

At its best, Web analytics is progressive and actionable. It’s about testing hypotheses and taking action based on your findings. Yes, your customers will have vastly different goals, but by asking just a little bit more of them, you can really maximize the benefits of working with your Web analytics program.

“OK, sounds nice,” you might be saying, “but where’s the substance?”

Fair enough. So, let’s talk about three critical elements to actionable Web analytics:

  1. The goal-KPI relationship
  2. Hypothesis
  3. Action

1. Goal-KPI Relationship

The next time someone asks you for a Web traffic report, ask them to articulate their goals first. “Why is this report important to you? What are you trying to improve?” In other words, demand accountability.

Once you can clearly define the goals, you will need Key Performance Indicators (KPIs) to measure success. Basically, KPIs are the metrics best suited for measuring how well customers meet their goals. If a metric merely provides context, it may be suited for a dashboard or report, but it’s not a KPI.

Here’s an example:

Goal: Improve the website’s search functionality

Potential KPI: Percentage change in the Exit Rate on search results pages

Ideally, everyone who performs a search on your website selects a result. So, once you establish the current (baseline) Exit Rate, you can begin brainstorming ways to improve search.

Also remember that KPIs need not be based entirely on standard metrics found in tools like Google Analytics (GA). In fact, some of the most effective KPIs draw on data from multiple sources.

Here’s an example:

Goal: Increase the cost-effectiveness of our Web applications

Potential KPI: Cost per user

In this example you have two variables on which to take action. In addition to trying to increase Web traffic, you can strategize ways to reduce costs.

2. Hypothesis

KPIs must have a target. One approach is to set the target as a modest increase from the baseline (e.g. increasing sessions to the agency blog by 10%). What’s missing from this approach, however, is responsibility. Why will sessions increase? More importantly, if they do, what will it have meant?

Ask the customer what they are going to do to meet their goals. “What changes are we testing?” Failing that, suggest some hypotheses on their behalf.

Here’s an example:

Goal: Increase content consumption on the agency blog

KPI: Percentage change in sessions to the agency blog

Hypothesis: Getting 12 partner organizations to link to the agency blog will lead to a 10% increase in total sessions.

Having a hypothesis to measure is more important than where you set the target. The process is iterative, but it’s better to be testing based on actions and assumptions, rather than complete ambiguity.

3. Action

Not only do you need to ensure the customer takes whatever actions are defined in the hypothesis, but try to get a commitment for follow-up action as well. “If the hypothesis is correct, what are your next steps? If it’s wrong?”

The more you reinforce the need for action, the more you will enhance the perception of your Web analytics program.

So, the next time you are approached by an individual, program, or office, reference this three-step process and remember how powerful your work can be!Sam Bronson is the Web Analytics Program Manager at the U.S. Environmental Protection Agency (EPA).