Picking the right words is a key component of many jobs, and most of the time, we rely on our own vocabulary. But a new feature from Seattle-based tech startup Textio uses the power of machine learning to find the most effective words, thanks to a vast database and the power of a supercomputer.
Textio’s new predictive engine, called Opportunities, was announced today. Designed for job posts and emails to candidates, Opportunities adds to the existing Textio analysis tool that helps companies find diverse employees. The tool already spots language that historically attracts more male candidates and suggests replacements that are more neutral.
But the new engine learns from Textio’s users, evolving as language changes and once-powerful phrases begin to lose meaning. For example, Textio co-founder and CTO Jensen Harris points to the phrase “big data.”
“A year ago, if you wrote ‘big data’ in a job listing, software engineers would have been beating down your door asking for a job,” he said. “About six months ago, it started to lose its luster. Today, the term ‘big data’ is statistically an actual negative that will push engineers away from applying to your jobs.”
The nature of language means that finding the next “big data” is impossible for Textio’s staff to do manually. However, a supercomputer can analyze all the job postings made by Textio clients, along with data they feed back about who applies, and figure out which words attract the best talent.
Textio co-founder and CEO Kieran Snyder compares the new engine to Google’s AlphaGo project, which used supercomputing power to defeat human players. Textio’s engine spots all possible phrase you could use in your job posting.
“Before we began work on Opportunities, Textio recognized 75,000 or so phrases that have predictive power,” she said. “Sounds pretty big, definitely more than you can keep in a checklist. With this, though, the number of phrases that are eligible to improve your document are every phrase that can be written.”
And it’s not just technical words that describe the job. Textio has found that “love” currently draws in more interest than “like” for many jobs, whereas phrases like “passion for learning” draw in more women.
However, job listings and candidate emails are Textio’s focus for now. The rich data that comes with job listings allows the company to analyze text much more accurately than other forms of writing. An email to your boss or a post on your celebrity gossip blog doesn’t come with the same rich feedback as a job posting, which includes a wealth of data on the applicants, so suggesting new phrases for other writing is a much harder prospect.
Textio first hit the market less than a year ago, in July of 2015. Snyder is a linguistics expert who previously worked at Amazon and Microsoft’s Bing unit and conducted research on gender bias in performance reviews in the tech industry before starting Textio. Harris spent 16 years at Microsoft, heading up user experience teams for Outlook, Office and Windows.
The company raised $8 million back in December and partially used those funds to help build out this predictive engine, hiring researchers and engineers to execute a vision Snyder said they have had since the founding the company.
“Straight up, we are building the best machine learning engineering team in Seattle,” she said.