HRO Today July/August 2017 - 57

HRO Today Forum
" What happened with big data is that it became
so popular that it actually became a bit of a cliché, "
Snyder explains. " And today if you include big data in
a technology or finance job post, it actually detracts
from its performance, and those roles will fill much
more slowly. We're starting to see the term AI ride the
same curve. "
Conversely, there are also many keywords-which vary
by industry-that have gained traction in the past
year. For example, in the tech industry, the following
keywords became more effective in 2016:
* Gender identity
* Systems engineering
* Security clearance
* Minimal supervision
* Written communication
The solution to the cliché-keyword problem-not to
mention the most effective means of finding keywords
that work-was creating what Snyder refers to as a
" data loop. " This translates to continuously collecting
performance data from the 10 million new job posts
that Textio analyzes each month and using their
outcomes to create the criteria for the next 10 million
posts that come in. With this data, Textio has insight
into the best style and wording for attractive job
listings.
Once the data loop was established, the next challenge
was translating that information into actionable
guidance that could be understood universally by
recruiters and other talent acquisition professionals.
Many recruiters are excellent at selecting and
communicating with prospective candidates but
lack the deep technology background necessary to
generate and decode data sets. Understanding this
disconnect, Snyder wanted to ensure Textio made it
as easy as possible for recruiters to obtain a straightforward
analysis of the job listing that they've created,
answering the question: " Will this post attract the
perfect candidate? "
" I think that if you talk to anybody with a foundational
technology that uses machine learning, they'll say that
they're only as successful as their ability to build an
experience on the right data set, " Snyder says. " Then
the flipside of that is translating all of that data into
guidance that a real person can follow-even if they're
not a statistician or a linguist. "
In its three years as a business, Textio has assisted with
over 250 million job posts, and data analysis found that
companies that use Textio's recommendations to get
a high score (90 or above out of 100) see an average
of 25 percent more candidates qualified enough
to interview, 23 percent more female applicants,
and a time to fill that is two weeks shorter. Textio's
measurable success is one of Snyder's greatest points of
pride.
" The results are really statistically quantitative. You
can measure them, which I think is really unusual for
HR tech. I don't know of too many other products that
will actually promise you results for your pipeline, and
that really speaks to the power of the dataset and the
technology, " she says.
This data-driven approach has also informed Textio's
plans for the future. Data exchange customers-
customers who agree to share the outcomes of their
Textio-enabled listings-account for about 90 percent of
the company's business. Snyder plans to make it easier for
those users to share the results with Textio. Specifically,
she wants to introduce a closer workflow between Textio
and the job distribution systems or applicant tracking
systems where people publish their listings. Textio will
eventually be able to automatically publish the modified
posts it creates directly to those systems.
Research collected by the platform also revealed that
many Textio users are not content to only glean insight
into job listings. Rather, they have tried to use the
service to improve other forms of business writing
such as recruiter emails and social media messages. In
response, Snyder hopes to expand Textio's services to
include assistance with other communications.
" One of the biggest examples here is around candidate
recruiting emails: the emails that say 'Hey, I saw you
on Linkedin, I love your background, please come work
for my company,' " she says. " The kind of writing that
works for a good email is going to be pretty different
than what would work for a good job post even
though they're both in the recruiting context, so that's
a big area of focus for us. "
JULY/AUGUST 2017
| www.hrotoday.com
[57]
http://www.hrotoday.com
HRO Today July/August 2017

Table of Contents for the Digital Edition of HRO Today July/August 2017
