Monday, September 8, 2008

Website measurement for Optimization - it's more than conversion

I see a lot of people A-B testing to try to optimize "conversion". I don't mind using conversion as a short-hand metaphor, but I hope everyone knows that there is a lot more to life than the top end conversion number. Ultimately, a website should be optimizing strategic performance. And while a major element of strategic success is normally profitability which can be correlated to conversion, there is more to it than that, and even top end conversion is too high a level to optimize to if you have significant traffic volume.

Examples of strategic goals which are not exactly top-end conversion:

  • Traffic - if you are going to sell your company at some point, the number of unique visitors can ultimately be worth a lot. Also, there may be economies of scale that come with being first in a market segment. And there may be long term strategic value in running competitors out of the market space by taking all of the traffic away at slim to no margin in the short run.
  • Return rates - it is possible to maximize conversion on a per visitor basis but hurt return rates. Since repeat traffic is often free vs spending to get traffic to your site, you may be hurting your long term profitability. If you are in a cash crunch, fine, but you should at least be aware of what you are doing here and do it deliberately.
  • Micro-conversion rates - the top-end conversion might get hurt, but some conversions within your funnel might improve. There may be things to learn and adjust by looking at small local conversions.
  • Segmentation - just because your overall conversion rate did not change does not mean that conversion in some segments did not go way up or down.
  • Long term profitability - When optimizing profitability, you really want to optimize the long term value of each visitor to your site so you can maximize returns. The only trouble is that it takes a while to get a good vintage analysis of long term value for everyone, and you can measure upfront conversion right away. Take care not to move too fast and sacrifice long term returns for the quick 10% increase in conversion. Make sure you look back after a while and see what the long term impact of a change is. And make sure you have the ability to measure it. It is typically one of the more challenging things to measure.

Of course, measure conversion too!

If you measure and use segments that your web server is aware of, you can show different behavior on your site to different visitors and optimize conversion for each little segment. Examples of things your web server can be aware of include:

  • Referral source - the url email link vs search vs display vs the red banner vs the green banner, etc
  • behavior on your site - visitors that looked at digital cameras, people that registered for your newsletter, folks that have previously visited, previously purchased, logged in currently, etc. There are even third parties that will tell you information about your traffic for a fee based on cookies which they have dropped around the web. I don't like using these for privacy reasons, but they are out there.

Well, this is a start on a measurement and optimization metrics list. I will edit and improve this list over time. Do you have example metrics besides conversion critical for your strategic success that you would like to share to help me improve this post?

No comments: