There are some solid insights to be gained from Claire Cain Miller’s recent New York Times article, Can an Algorithm Hire Better Than a Human?
First a few words about business (B2B) vs. consumer (B2C) data. One can be at a bit of a loss when asked to apply data profiling to businesses the same way we do in the consumer space.
The big B2B list compilers’ selects are pretty much limited to: Standard Industry Code (SIC), annual sales, number of employees, years in business, geography, corporate tree, names, job titles, certain types purchasing history or inferred areas of decision-making responsibility – and flags for self-employed/home-based (SOHO) businesses, cottage businesses, franchise locations, etc.
Business “demographics” are barely a fraction of the 100s of data-points available on every U.S. consumer. So we can forget about all the predictive modeling tools we use on consumers.
That is, if we limit our targeting to the company level.
Once we dig down to the individual level, it’s a different story.
With LinkedIn and other tools, our targeting isn’t limited to company demographics – where you might infer that John or Mary is interested in a particular product or service based on their job title. With LinkedIn, we can find John, Mary and others like them who have joined certain interest groups (e.g., supply chain management, marketing automation, water engineering). They’ve self-qualified in their quest to stay current in their areas of expertise. And we can target them based on that.
Many LinkedIn members may not be aware that, as with Facebook, we are NOT the customers. We are the PRODUCT that is being sold. The customers are the advertisers – and with LinkedIn, the biggest customers are the executive recruiters. (And those of us trying to sell things to their members.)
Here are a few takeaways helpful to our peers in the data-driven marketing space:
- Hiring and recruiting often rely on the ability of humans to recognize human skills, like making conversation and reading social cues. But people have biases and predilections. They make hiring decisions, often unconsciously, based on similarities that have nothing to do with the job requirements.
- A new wave of start-ups — Gild, Entelo, Textio, Doxa and GapJumpers feel that software can do the job more effectively and efficiently than people can, which could make hiring faster and less expensive. And their data could lead recruiters to more highly-skilled people who are better matches for their companies.
- Another potential result: more diversity. Because the software relies on data, it tends to identify candidates from a wide variety of places, free of human biases. The interviewing process tends to be hugely influenced by similarity between the interviewer and interviewee , even though it’s not predictive of how the new employee will perform down the road.
- Gild uses employers’ own data and publicly available data from places like LinkedIn or GitHub to find people whose skills match that company’s requirements. It tries to calculate the likelihood that people would be interested in a job – and it even suggests the right TIME to contact them, based on the trajectory of their company and career.
- Textio, uses machine learning and language analysis to review job postings for companies like Starbucks and Barclays. Textio uncovered more than 25,000 phrases that indicate gender bias. Language like “top-tier” and “aggressive” and sports or military analogies like “mission critical” decrease the proportion of women who apply for a job. Language like “partnerships” and “passion for learning” attract more women.
Back to marketing data… As in hiring, we know that data plays a big role in campaign performance. And it’s often counterintuitive. That campaign with the super compelling creative may not perform as well as the one with the “ugly” design.
Not only is the HR space seeing the same thing via these algorithms, but these new technologies may provide further insights for targeted marketing. LinkedIn’s groups add a “psychographic” enhancement at the individual level, far beyond the old “demographics” inferred by a person’s job title.
And this new path of algorithmic discovery based on a job-hunter’s skills and experience may take it a step further. Will this shed new insights on an individual’s purchase-making decision process? We’ll see.