Using Machine Learning to Maintain High Conversion Rates

It doesn’t happen often, but we managed to maintain or increase lead quantity without decreasing quality.

Instead of asking a lot of questions to tell us more about a lead’s company, we just scrape their website, and compare that content to what we’ve found from companies in known segments, and are able to categorize the lead appropriately.

The wordier explanation is here:

Our Problem

There’s usually a tradeoff between lead quantity and lead quality. Generally speaking, increasing your quantity of leads results in decreasing the overall quality of those leads.

This is often an issue when deciding on how  many fields to ask on a lead form. Ask more questions to get a better sense of who the prospective customer is and what her needs are? Or ask fewer questions to keep the form short and easy to fill out?

In this case at PitchBook, we had a problem where our Sales teams were unhappy with the types of leads they were getting because they weren’t necessarily in the customer segment that their teams supported. We could have asked more questions on our forms to determine what segment they were in, but from prior experience, that ends up resulting in lower conversion rate on the site, and fewer leads overall. At a company with growth targets like PitchBook’s, that would have been unacceptable.

First, some context around our customer segments: PitchBook Data‘s customer groups have different use cases, different needs, which can be better addressed by specialized Sales teams. For example, prospects who work at startups have different use cases than venture capitalists, and therefore they need to get routed to our sales team that focuses on startup customers.

To help our Sales teams, we decided to figure out whether a lead was coming from a startup or not. In many cases, we could do that fairly simply: we could look at the email address a person submitted, and looking at its domain, we could match it against the domains of startups listed in the PitchBook platform. This is one of the great perks of working at a company with the best database of privately owned companies.

However, there are some companies that aren’t known to PitchBook (or at least not yet!). Your cousin who just bought a domain last week and thinks he’s going to do something revolutionary by coming up with “like Uber plus machine learning for elective dentistry in VR” might not show up in our database yet.


Our Solution

Having received a person’s domain from their email address (unless it’s a Gmail, in which case we’re currently out of luck), we can visit their website, or more accurately, scrape their website. We can then take that data and compare it to other companies in known segments.

At this point, despite the fun I’ve had in learning how to automate various things, I’ll acknowledge that the technical explanation is beyond my area of expertise. Something-something machine learning.

Best to look at PB Demand Gen’s Erik Larson‘s explanation of how he made use of machine learning and Google‘s Content Classification API to solve this problem.

Because we didn’t have to ask any more questions on our lead form, we maintained conversion rate on our site, and we gave our Sales team what they wanted.

Sometimes you can have it all.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s