Semantics Noun – the meanings of words and phrases in a particular context
In this case the “context” is a resume and a job description. Matching algorithm’s that promise to change everyone’s life….and they don’t.
Big data comes along and says, “Hey if you have more information about them before you hire them, you’ll make a better hire!” So primarily they crawl and add additional social data sets to resume data sets and then match them up against your data set – the job description.
Doesn’t work does it? Ever wonder why? It’s because the basis for the data is poor context.
Let me back up…
My background is AdTech/eCommerce/Big Data. Specifically a form of display advertising called retargeting. You go to a website, you don’t buy, you get an ad based on your behavior on the site. Everyone in retargeting uses algorithms and they work. The technology drives clicks, serves efficient impressions, and matches product to people. Hundreds of relevant data points over millions of users crunched to find efficiencies. Genius? Not really, it’s just great context and structure of data.
When a human shops on a site, it’s natural. They don’t comply to a standard, they navigate and filter based on preference. And the eCommerce site is structured, prices are in the same place, product pictures are in the same place, etc. Your natural responses are tracked, recorded, matched to additional data sets, and action is taken in controlled environments. Maybe you get an offer on site. Maybe you get one off site. It’s behavior based, normalized, structured data.
So how can this be related to HR?
A resume is the antithesis of this. An ancient open format created for the purpose of being a conversation starter that morphed into a “end all be all” search and approval document used to filter. What gets matched against it? A job description. Another open format document initially created for legal/compliance reasons that morphed into a matching data set that describes perfection in a person for a company.
If you wanted to buy a pair of shoes, can you imagine documenting shoe history in an open format from 50 years ago, submit it to Zappos, who takes their return policy mixed with shoe names and them shipping 2% of the matches a box of shoes to buy? That’s what roughly what is happing with matching algorithms….but hey, they are proprietary and patent pending!
eCommerce created a structured data path trading privacy for a digital footprint along the way, to facilitate more transactions. When transactions don’t happen, the data gathered re-targets that prospect with relevant messaging created from the first visit. eCommerce is match making of a different kind.
What can be done to improve it?
Semantic data has little to no structure and in this case both sides are semantic. So having more of it (Big Data) is a big waste of time. Trying to match it, another big waste of time. What works is gathering better data, structured data, intent data, and private data. Resumes and job descriptions are none of these things, so why gather more? And why attempt to match it?
A big confusion is equating a job board to an eCommerce site. Yes people shop for jobs like they shop for shoes (sort of), but the job isn’t the product here, the person is. Disagree? When you go to buy shoes, who pays? You do. When you go to hire someone, who pays? You do. See the correlation?
It’s time we march toward relevant data gathering. It’s time this data is housed on shopping engines. And it’s time we stop investing in semantics for the sake of tradition and start investing our time in data that will have a positive impact on building great teams.
Author: Chad Porter is the Co Founder of Invisume, The Invisible Resume and an eCommerce veteran applying the principles learned at his time at eBay with hiring. He’s been featured on some of the top sales blogs in the United States and spoken at conferences regarding display advertising and big data.