Utilizing Predictive Revenue Models in Paid & Organic Search
One of the more interesting concepts presented at last week’s SES show in San Jose detailed the use of predictive revenue modeling for SEO and paid search. A former colleague of mine, Dave Roth of Yahoo!, runs search programs on behalf of Yahoo!‘s properties. He discussed their use of predictive data models to gauge how they were performing against their competition.
In SEO it works like this:
For each keyword, create an “Opportunity report” that looks at your current organic rank and click volume versus your top competitors, and measure the gaps. Then map those gaps to Lifetime Value (LTV) to quantify their value. If, for example, you know how much revenue you made in position #1 versus position #2, you can quantify your revenue losses or gains based on rank. In other words, how much revenue are you losing by not ranking higher than your competitors? Nothing gets an executive’s attention faster than lost revenue, especially if you can see (at least roughly) how much revenue your competitors are earning by outranking you.
In paid search it works similarly, but is a little more complex:
For each keyword, create an LTV model based on past revenue performance, click volume, rank, CPC and click costs. By factoring in click costs you are capturing your efficiency over time because click costs account for your quality score. At Yahoo! Dave explained that they reforecast every month to stay on top of rapidly changing search markets, then communicate the updated numbers across all relevant internal departments so everybody knows what revenue gains or losses to expect for that month. At Yahoo! they use a “scorecard” to measure each Yahoo! property vs. another, all based on ROI (both immediate for that month and LTV). My favorite line in Dave’s presentation was a Yahoo! mantra: “If you can’t quantify it, it doesn’t exist.”
In a world where everything is quantified, all of your actions can be based on data and revenue models; and perhaps even more importantly, you can communicate potential fluctuations in revenue BEFORE they happen.
Make sure your toolset can read your search performance data across both paid and organic so you don’t have to marry two different reports together. Having that data in two separate systems leaves room for human error when combining the data sets. Don’t settle for anything less than a holistic approach to analyzing the performance of your search programs, and don’t rule out the importance of predictive revenue models for your organization.