Comparing Visits & Clicks…

In this business, we often get distracted by technologies on the way to the business ends they are supposed to help achieve. The purpose of this post is to outline a basic process that should shed some light on the very thorny issues involved when comparing numbers derived from agency-side ad servers like DART aka DFA and client-side site metrics tools like Omniture.

Complicating matters is a situation whereby landing page tracking by the ad server cannot be implemented. What to do?
  1. Determine the Purpose. If the marketing objective is performance-oriented, or post-click engagement is essential then you are on the right track. However, if your campaign is focused on a branding objective, how important is it really to get numbers to match?
  2. Choose your Measures. Which measures make sense? Stefane Hamel provides a good backgrounder on Instances vs. Visits. If you are in the paid display advertising/non-search business then most likely clicks and visits are closest; unique clicks and visits/visitors even better. If you are working with search, instances may make more sense…view-through offers even more insight on post-click engagement.
  3. Leverage Standards. The Internet Advertising Bureau (IAB) & Web Analytics Association (WAA) are probably the most relevant industry groups that are working on measurement standards; terms and definitions vary.
  4. Manage Expectations. The reality is that numbers from different systems are unlikely to match; there are a variety of reasons but to make a long story short, they won’t match without serious integration and that has not yet happened. However, the recent announcement of a multi-faceted strategic alliance between agency holding group WPP & Omniture is very promising.
  5. Baseline. Given #1, the best alternative to is to create a simple baseline, e.g. an average over a safe period of time, e.g. a week or a month.
  6. Drop-off/Match. Breathe deep – acknowledge the causes are most likely latency, clickthrough URL parameter coding, landing page tag placement, application filters and counting methods that may never be in sync.
  7. Test & Debug. Understand that campaign trafficking and set-up processes are rife with glitches…clients, creative shops, media agencies and publishers rarely have the staff, or luxury of training to make all of this work flawlessly all of the time. One could spend significant quality time on process engineering these tasks.
Seems simple enough, right? Easier said than done!

Leave a Reply