We take a look at the HireVue press release from July 23rd (it can be found here) in our usual commentary style.

Here we go:

SALT LAKE CITY, Utah – July 17, 2014 – HireVue, a leading digital recruiting and talent interaction provider, today announced the first predictive candidate and interviewer recommendation engine, HireVue Insights. For the first time, companies can use the power of big data to identify their top candidates and best interviewers based on interaction, hiring and performance attributes. It helps organizations optimize their hiring model and gives talent professionals a competitive edge by reducing guesswork and helping them discover the right candidates more quickly. For candidates, it levels the playing field and gives often overlooked talent a chance.

According to the Talent Board, less than 6% of job applicants get an opportunity to interview for a position. The average position receives approximately 100 applicants, leaving 94 candidates in a black hole – often wondering where they stand and why. Many have the skills, personality and potential to do the job but don’t get a chance to tell their story. Until now... HireVue Insights uses the power of big data and personalized digital interactions to recommend candidates based on 15,000 interaction, behavioral and performance attributes – and how they correlate to the organizations current top performers.

MyPOV – Good description of the problem from an enterprise and recruiter perspective – it is really all about finding the best fit candidate. With the application load the error rate at finding the best candidate naturally goes up and what is better to solve that than looking at BigData and (true) Analytics to help find the best candidate. (Here is my definition of true Analytics and 6 common misunderstandings on the topic).

HireVue also does a good job to keep stressing the candidate experience. Not really clear how this product will change it – the answer is of course more and better automation to tell candidates where they stand in the application life cycle. But it also means that each criteria captured about a candidate will be part of the interview process funnel and part of the hiring decision – a better situation for candidates out there than leaving it to human, usually information overloaded and error prone decision making.

HireVue Iris™, a patented deep learning analytics engine that powers HireVue Insights, analyzes a unique data set of interactions, feedback and outcomes that never before existed. Developed by HireVue’s data science team, Iris was built based on over 3 million interview responses. Each candidate interview contains 100,000 times more bytes of data than the resume or profile traditionally used for identifying job candidates. The platform examines attributes in three major categories: interview attributes, behavioral attributes, and performance attributes. Iris’s proprietary algorithms discover patterns and learn which attributes predict performance, then scores each candidate on how they compare to existing top performers. And, similar to a batting average for hiring managers, Iris also scores interviewers based on how their historic ratings and feedback correlated with hiring and performance outcomes.

MyPOV – A nice insight on how HireVue Iris is trying to solve the problem. What is not mentioned – but key to make this a very interesting analytical offering – are capabilities that were demoed to me back at the HireVue user conference (my takeaways here) in early June: The recruiter can improve analytical models by rating past hiring success – a key trust creating measure in my view – to get business users to trust the magic coming their way via an analytical application. And HireVue smartly provides a layered model delivery, starting with a default model, fine tuning that to an industry model and then all the way to a position model. The right approach to makes there is immediate value from analytical applications. Business users do not expect (some of them fear) for the analytical software to be right away better, but by easing them in to the software’s usage, when they can see that the software is adaptive, and learns (from them), they should come on board and comfortable using the analytical software.

And HireVue rightfully highlights the other perspective – the model can also be used to predict how thoroughly and ultimately successfully a recruiter works.

“Recruiters and hiring managers rely heavily on instincts, hunches and memory to choose the right candidates, but there isn’t a lot of data to help them predict who will become a top performer, or decide who should be interviewing candidates,” said Mark Newman, CEO of HireVue. “This could be the most important innovation in recruiting in the past 25 years. HireVue Insights analyzes over 100,000 times more data than a resume, all within the context of your organization, your positions and your feedback. It gets smarter over time to become your own personal data-driven hiring model.”

MyPOV – Well said. Not sure HireVue needs to stretch the 100k more – assuming this is the math behind the 2-4kb of the average resume vs. the average length of the interview. But at the end it does not matter – as long as the Isis model can show a recruiter better candidates.

The data science team at HireVue worked closely with customers to develop and configure Insights to meet the unique needs of each company. Customers like Chipotle Mexican Grill and others are already realizing big improvements within their processes. […]

MyPOV – Good to see the model was not build in the data scientist’s ivory tower but with customers. Extra points for a well known brand like Chipotle.

Overall MyPOV

Well done by HireVue to apply BigData and (true) Analytics to a very hard problem, recruiting. It would be great to know what behavioral / body language etc. algorithms HireVue might be using to skim through the video file, but understandably that is proprietary. With end user feedback consideration and stepwise model progression HireVue has implemented two key mechanisms that are important for analytical application success.

Now we can only wish HireVue and their customers luck in the early phase to have the successful showcases everyone hopes analytical software will deliver. The good news is that from interacting with the HireVue analytical brain trust at the user conference, they have the ability to address and fix things quickly - should that be necessary. And then it would be great to consider model thrashing and multi-model decision making – not just using the powerful and pretty ubiquitous scoring mechanisms. But one step at time.

More on HireVue

  • First Take - 3 Takeaways from HireVue Digita Disruption Conference - Day 1 Keynote - read here

More on Recruiting

  • Musings - How Technology Innovation fuels Recruiting and disrupts the Laggards - read here
  • Musings - What is the future of recruiting? Read here
  • HRTech 2014 takeaways - Read here.
  • Why all the attention to recruiting? Read here.

And  more on Payroll:

  • Could the paycheck re-invent HCM – yes it can – read here.
  • And suddenly, payroll matters again! Read here.

Find even more coverage on the Constellation Research website here.