This is part one of a three part posting summarizing a session I gave recently at Searchfest up here in pdx. The topic on advanced measurement for SEO analytics came about because I was annoyed by inaccurate measurement, silos of data, and inadequate reporting particularly from agencies (not their fault). In this first posting I will be talking about creating a measurement framework for SEM specifically, lifetime value and segmentation.
A measurement framework helps to prevent bad measurement. Examples of bad measurement are exceedingly common. For example, thinking that optimization only applies to PPC not SEO. Other common and nearly always mistaken techniques for optimization include:
- Tracking Clicks
- Tracking Conversions Without Assessing Value
- Tracking Engagement Based Solely on Page Views
- Tracking Engagement Based Solely on Time Spent
- Tracking Current –Session Behavior Only
The key to a measurement framework is being able to track key variables- keyword, program, placement, creative, etc.- against your measures of success. With SEM and Web Analytics, each half of this can be a problem. Many web sites don’t pass the necessary information to the web site about the key variables. Sometimes they are relying on external measurement systems that CAN access these measures.
However, there are good reasons on most large and complicated sites to use a web analytics tool. Many sites don’t have a good idea of how to measure their success. This problem is not acute in retail where the world revolves around the shopping cart, but in financial services, media, social, health, and government it is severe.
Success of your measurement framework can be defined in multiple ways. For most non-retail sites, it is necessary to develop a proxy for success. This can be the total value of pages viewed-media- or a measure of engagement- financial services, health, and government.
For all types of sites- including retail- it is ESSENTIAL that the sites measure not just the session-based value of a visit, but the anticipated lifetime value. This is one reason why a web analytics tool is essential, since 3rd party ad-tracking tools will not generally do this. Unfortunately, it is very difficult to measure LTV in real time.
So how can we track lifetime value? Start by assessing visitor lifetime value over a significant chunk of time. Analyze early-session behaviors to find predictive variables. For example, you may find that non-brand keyword visitors (let's call them "early stage researchers") exhibit a set of behaviors which is different than brand keyword visitors (let's call them "directed product researchers").
Create a framework that maps early session segments (early stage researchers and directed product researchers) to LTV segments (such as acquired, engaged, converted, retained or just one-time buyers and repeat buyers, for example). And use this classification to optimize campaigns.
This is a diagram that generally illustrates the measurement framework.On the left and right sides are brand and non-brand searches pouring in. The arrows show noise that must be subtracted from your model such as bounces (aka single page views) or job-seekers. Now you can map early stage researcher to the acquired lifetime value segment and directed product researchers to the engaged group. When you want to optimize your campaigns you are now asking the question, do I want to target more people at the very top of the funnel such as non-brand searchers or do I want to target folks who are familiar with the products? Each strategy has a different conversion goal and success metric.