Thought Leaders in Big Data: David Bernstein, Vice President of eQuest’s Big Data Division (Part 3)

Sramana Mitra: When you make that recommendation, what is happening in the back end?

David Bernstein: We bring forward a recommendation of the job or outlet to be used by our client – at which point in the year and for which positions [they should use certain outlets]. To get to that, we study the markets they are recruiting in. We look at recentness, frequency, response rates, and the level of the position and the way they describe the position in a posting title against the function the job encompasses. Then we look at the response patterns of candidates to help our customers understand which boards produce the most candidate flow or which boards are most appealing to our customers.

SM: The kind of data you are describing doesn’t look like big data to me.

DB: We receive more than a million job performance records a day through our system, on a global scale. We help customers know which the kinds of positions and the right kind of board set are on the market. Our initial offering to our customers is insight into how all of that data comes together in order to give a picture of the right set of forecast and the measurement of their effectiveness through their recruitment cycle.

SM: So, what you are saying is that [working with] these one million job records a day helps you develop certain heuristics within the system, and then you are basically running the new jobs, queries, or functions that your clients have to execute. You can then run that again through heuristics and model it.

DB: Correct. I was referring to the recentness frequency and response rate. Those million records a day are within the category of our job performance responses. We have approximately 200,000 unique candidates a day running through this system, viewing and responding to job postings around the world. That activity creates approximately one million new records a day. In a recent press release we explain how we were studying [the relationship between responses] and the days of the week and how this was changing over time. We analyzed the devices through which candidates responded, either mobile or desktop, and which job functions – health care, retail, or IT jobs, for example – people responded to. We started pulling together these studies to see was going on in those response patterns.  If this same study was going on in big data, it enables us to also provide predictive forecasts for our customers: Which boards or which collection of boards should they start their recruitment on, and how could they allocate their resources most efficiently?

SM: Let’s have a few more of these case studies.

DB: One of our customers is a large financial institution on the East Coast. They  relied on the board they had [because of] a lot of anecdotal evidence and past positive experiences. Over time, the list of job boards and subscriptions they had amassed was approaching 50. What they ended up getting was a portfolio of job boards wherein they had no understanding of which board they should be using at which time or for which positions. In this niche we often call it “post and pray” or “post and spray,” but it is an ineffective sourcing strategy where you are overspending. You do what you think is the right thing to do, which is try to create the best candidate flow so you can get more interviews going. It is an attempt to stay on top of things and meet the hiring demands of the company. But in the end it is a very ineffective sourcing plan. They ended up having to go back to a year-to-year request because the previous year’s budget is what they base next year’s budget request on. They ended up having a top-heavy budget request that didn’t ultimately satisfy the hiring cycle.

We sat down with them and did an effectiveness review. We were able to work with them and greatly reduce the number of boards they needed to use. They were also able to reallocate much of their budget. We also found some boards that they weren’t using but they ought to be using, in our estimation, and we created a more efficient advertising recommendation set for them, which created a higher candidate flow and led to more interviews and ultimately to more hires at a more cost-effective rate.

This segment is part 3 in the series : Thought Leaders in Big Data: David Bernstein, Vice President of eQuest’s Big Data Division
1 2 3 4 5 6