I have been thinking lately about the right intervals for revisiting and changing your site and business KPIs. I won’t bore you with all the back and forth, but merely share a few thoughts.

The stability of your business plays a role. If you are an established business, with established goals of your site, revisiting your KPIs too often suggests to me that you didn’t have the right ones in the first place. If, however, you are a newer, (somewhat) flying-by-the-seat-of-your-pants business, perhaps perfecting your approach, I can understand a much more fluid approach to what constitutes “success” and a need to more continuously evolve your KPIs.

My overall thoughts are that KPIs can’t and shouldn’t change every month, even for the latter business example. (Would it be unprofessional of me to say, “duh”?) You should be consistently measuring against the same yardstick.  However, I do think it’s good practice to take a look every three to six months and make sure your KPIs are 1) useful and 2) complete. Do you actually need all of them? Or, on the flip side, is there something new that should be included? Perhaps there are new capabilities you have developed that would allow a new KPI to be measured? After all, a year can allow for a lot of development in the analytics industry. Take advantage of new measurement options.

Recently, I have been involved in the re-evaluation of our KPIs. At the beginning of this effort, the website product team and the analytics team were involved in brainstorming new ways to evaluate the success of the site. Once we decided 1) what we should measure and 2) what we could measure, figured out the overlap of the two and selected from those, analytics began publishing the information. Now, a few months down the track, we’re at a point where our product managers are somewhat comfortable with the information, and the time has come to revisit. (After all, they can’t give feedback on something they don’t even understand or use yet. You have to give the information some time to allow for informed feedback.) Are they using the information? What’s missing? What’s overkill? I expect to do this every 3-12 months from here on out. Just like our site, I expect our measurement of it to be iteratively developed over time.

Parting hint: If you’re not sure if something is helpful or not, try removing it for a month. If no one complains, you have your answer. (But I never do this. Never. No, Really.)