I googled the crap out of this but can’t find a single link that talks about talent management, recruiting or anything in the HR space that works like Pandora does for music. Maybe I’m all wet. Maybe I should have checked with Josh Letourneau. Maybe it’s a bagillion dollar idea that I’m giving away here but…
Talent Management and Recruiting Should Work Like Pandora
If you’re not up to speed on Pandora, it is a web-based (and mobile) music service that helps you find new music – music you will like. You put in your preferences (ie: music you like) and then it plays that music AND through a proprietary algorithm (love that word – makes ya sound so smart) it finds other music similar to what you put in and plays that as well. You then have the opportunity to vote “yes” or “no” on that new music to help the algorithm fine tune its recommendations.
Over time Pandora learns what you like and continues to serve up musical goodness with a lot of new music that you already like but weren’t aware of. It is eerily accurate.
Where’s Pandora for Talent?
Thinking about talent management, I couldn’t help but think that our lives would be much, much better if we had a system for talent management that worked like Pandora. We’d put in our “preferences” – the data on those employees that were in the top 10% of all reviews over the past few years. The system would then connect to say, LinkedIn, and pull those profiles with “similar” data. To be the negative me – you could do the same thing with the bottom 10% of your workforce as well to create a watch list for those profiles that historically have been non-starters in your company.
This isn’t a LinkedIn only idea but could work for just about any job board.
As you used the system (and as your company or culture changed) you could thumbs up or thumbs down certain profiles to help the algorithm learn your business requirements.
Identifying those traits, codifying them in a system and then applying a learning algorithm would seem to me a great time saver and help eliminate the “human” factor at the front end. Not saying we should eliminate the human factor in total, but in my experience, the first cut we take at any employee search is to use very specific and quantifiable elements to reduce the pile from “all resumes” to “some resumes.”
I’m just suggesting that we let a computer program that learns about us and our successes (and failures) do that front-end work for us.
Is that out there? Should it be? Why isn’t it?
Its a good idea.
I used to head admissions at a college and created a similar system for selecting students.
We would take the grade data of top current students, and feed all their admissions application data (education details, employment details, interests and hobbies, interview results, etc etc) into a program that would isolate the impact of each element on their future grades. For example, we might find that students with high undergrad results and 1-2 years of work experience would typically outperform students with low GPA but 4-5 years of work. (purely fictitious example)
Then we would apply this knowledge in assessing the new applications we were receiving.
Another aspect of your idea is that it can uncover biases in management thinking. If you realise that some of your managers are giving higher reviews to people from particular backgrounds, then you can address this internally.
Thanks Chris. Did your process work? Did it improve the decision making?
Paul,
Great idea and it really speaks to “triangulating” the real needs for future talent against what makes current talent successful. I believe Google does this (big surprise!) by breaking down their top hires into component parts (did they start a business, did they go to the right schools, etc.) and then building the Success = X + Y + Z algorithm and applying it to pre-hires based on their answers to questionnaires. Class regression analysis but of course you’d love to see what their R-squared values are. I believe this was presented during an ERE presentation back in 2006 or 2007 if anyone wants to look it up.
And so, in answer to your initial question, Google is already making a bazillion dollars off this idea since their employees are the ones driving the business and this would theoretically help them continue to hire the very best.
FOT rocks,
Matt Hill
If figures google was doing something similar… but I was thinking about how to apply that against existing databases like linkedin, etc. But google is pretty much there.
I know there are folks that do the “psychology” tests that predict performance and use that – but I’m thinking it would be based on the company’s internal ongoing assessment process (which are typically flawed so maybe the old garbage in-garbage out thing would apply.)
I’m just glad I didn’t get 100 comments saying – “hey dummy – they’ve been doing it for years. see this link.”
Thanks Matt!
Conceptual searching kinda fits intot his category. IF you’ve ever heard of an engine called Engenium, it’s a conceptual search engine that does an analsysis of resumes, builds up a knowledge base, and then lets you quickly compare /find similiar results.
So, what you could do was find the person and resume that was the “ideal” candidate for the job. You could then feed that entire resume into the engine, and it would return the best matching resumes to your “ideal”. We did a number of real world tests…and it was exceptionally good…i was always surprised at how similiar the returned resumes were. The key is you need to have a knowledge base built on 50,000+ resumes…that’s when the matching became good. In addition, if you could “group” people together into their own databases, it was even better…for example have a “Developers” database, a “Finance” database etc.
At a high level, it works like pandora in doing pattern recognition…so people are working on stuff like this!
Thanks Mike… that’s pretty close to where my head was.