We generally don’t like to pre-announce new features here at Prosper. But over the past few months we’ve put so much time and energy into a new feature called “Predictive Bidding” that we’re too excited to keep it under wraps any longer.

After mining our database of lender activity (don’t worry, no canaries were harmed), we detected some common patterns in lender activity before bids were placed:

• Lenders scrolling heavily on the listing page, back and forth between credit data and Q&A.
• Lenders clicking “Bid Now”, checking the estimated default, backing out, then clicking “Bid Now” again, then backing out, then sending a message to the borrower.
• Lenders reporting the listing, then asking the borrower a question, then inviting a friend to look at the listing.

In each of these patterns, the likelihood that the lender would then place a bid on the listing was upwards of 95%! In an effort to save our lenders some precious time, we decided that when we observe these patterns (among others), we should just go ahead and place a bid on the lender’s behalf. It’s kind of like a portfolio plan where the main criterion is your excitement level!

We’ll be rolling out this new “Predictive bidding” feature on April 1st. And in the following months, we look forward to extending predictive bidding with these improvements:

• Predictive Portfolio plans – lenders who run the same search more than 10 times in a week will have a portfolio plan created for the full amount of their cash balance.
• Eye tracking – using that webcam mounted on your monitor, we’ll detect when your eyes get big and excited about a juicy loan. If your account’s got $50 in it, you’re in.
• Predictive funds transfers – if your account ran out of money because of all those predictive bids, don’t worry – we’ll initiate a transfer for a fresh $500. You deserve it.

All these predictive features aren’t just for lenders, either. Based on some borrower data that is finally coming to light, we’re starting to see correlations between how a borrower fills out the registration form, and her credit grade, income level, and loan purpose. When we can predict these characteristics to 98% accuracy or better, we’ll bypass the Experian score pull, and auto-insert the income and listing description (everyone puts in $100/mo. for food anyway).

We hope these improvements to the marketplace are helpful to borrowers and lenders alike.

—

Andrew is a Product Manager at Prosper and today is April Fool’s.