Managing Lead ROI With Predictive Modeling

By Jeff Liebl, VP Marketing & Product Sales, eBureau

The biggest challenge with buying sales leads from third-party affiliates and aggregators today is the difficulty in measuring lead quality and predicting return-on-investment. In some product/service categories, it may take weeks or even months before a buyer can clearly assess the quality of a given lead based on whether or not it converted. Moreover, quality can be volatile month to month, even from the same lead source. If you can’t reliably measure quality, how do you know if you’re paying the right price for leads?

Leave it to the online education sector to teach us new techniques. For-profit, higher-ed giants like the University of Phoenix, Corinthian Colleges, Kaplan University and many other schools spend about $1 billion combined each year buying leads from third-party aggregators and affiliates. And they are leading a quiet revolution that will change the dynamics of lead pricing in every product and service category. These companies are not only insisting on lead ROI, they want it to be quantified before they buy leads. Sophisticated predictive lead-scoring technology—much the same as that used for many years by offline direct marketers and financial institutions (think FICO credit score)—makes this possible by running leads through predictive models that leverage huge databases containing consumer-specific demographic and other data.

Here’s how it works. Before they actually purchase any leads, some online educators use third-party service providers to score the leads. Say a company is testing a lead supplier based on an initial 500 leads. The lead scores indicate they are a mixture of high and low quality. In general, there are more high quality leads in the mix, so the overall average price is set at, let’s say, $47. Then the company creates three pricing tiers: good quality leads priced at $55, average quality priced at $38, and the worst 8% of the leads at $5 apiece. The campaign begins to run at 100 leads per day, but the lead score indicates that the campaign’s quality is deteriorating. Fortunately, the pricing tier automatically reduces the average cost per lead to $37. The lead seller also sees the drop in quality within the first couple of days. Since the lead score reports the quality of every lead in real-time, the lead seller uses this feedback loop to trace the source of the bad quality back to a couple of its new affiliates. Those affiliates are shut down and the average price per lead climbs back to $47.

Throughout the month, the lead seller tries different approaches to improving lead quality and identifies some new sources that bring its average up to $50 per lead. These are better-quality leads and worth the extra $3 each. With the increase in volume, the buyer’s call center is now getting 250 leads more per day. The lead scores help to prioritize follow-up efforts. The highest-scoring leads are quickly routed directly to the company’s best call center agents and the $5 leads simply receive an email follow-up. Despite the spike in volume, the lead conversion rate goes up by double-digits because the focus is being placed on the right leads.

Class dismissed!

Jeff Liebl is VP, Marketing & Product Sales, at eBureau, leading the marketing, product management and product sales organizations. His experience includes management roles in marketing, product management and business development with Silicon Valley-based Fortune 500 and venture capital-funded startups. In his previous role as Vice President of Global Marketing & Product Management for Ubiquity Software, Jeff built a 20-person global team, branded the company and launched several new products. Ubiquity successfully IPO'd in March 2005