Historically, decisions about salary involved
heavy binders full of data and some serious guesswork; HR professionals found
the best data sources and comparison data points available and then made ad hoc
adjustments to price the position in question. Today, compensation analysis should be a
science, not an art. With PayScale’s union of fresh detailed data and clever
statistical modeling in an easy to use software service, we're blinding the HR community with science! To understand how PayScale is rocking it, let’s
take a look at the science of PayScale.
a compensation data set that is both broad and deep is only half the battle.
What makes PayScale different from its competitors, who rely on old-school,
less specific, employer-submitted data, is that we use modern data mining and
predictive modeling techniques to yield the most highly accurate compensation
predictions possible, across a huge range of circumstances.
It's Poetry in Motion
So you want to know where we get our
data? We crowdsource it from visitors who come to our site looking for real
answers about what they are worth. Whether they are prepping to ask for a
raise, evaluating a job offer, or just want to know how they stack up against
others, they have an intrinsic motivation to not only complete our interactive,
online salary survey but also to give us honest answers.
We then run the data through a bunch
of filters to confirm its validity:
- Weed out the outliers: We discard data
that is too far outside of expectations.
- Defend against attempts to “stuff the
ballot box”: We automatically detect and reject too much data coming from any
- Standardize the data: You say
“computer programmer,” I say “software developer.” We have the technology to
realize we’re talking about the same job.
- Augment the data: PayScale knows a lot
more than just what we receive from our surveys. For example, if you tell us
you’re in Kuna, Idaho, we know that is in the Boise, Idaho metro area. Plus, we
know that the last census placed 606,376 people in the Boise area, and how
being in a city of that size affects compensation.
Once all that sifting and sorting is
done, we still have more than 35 million profiles – and counting. That’s a huge
pot of statistically relevant data.
As Sweet as any Harmony
PayScale finds the data that matters to you for pay, whether that’s showing you data that is near your location or that has the extra skill that is really driving compensation for the position you are pricing. PayScale's MarketMatchTM algorithm is kind of like online dating: we look at over 250 compensable factors and the relationships between those factors when finding the ideal matches for positions. For example, we know that most employers pay more for employees that have more experience or who are located in large urban areas.
Even with as much data as we have,
there can still be gaps and in those cases, MarketMatch makes
sophisticated mathematical predictions to get the answers you need. Say you want to know what a Java programmer with 7 years of experience
in Harlan, KY should be paid and we don’t have an exact match in our data. We
can determine how software developers are paid, in general and then adjust that
prediction to account for having seven years of experience, for being in a
small town in the upper South, and for programming in Java rather than C++.
words of Jim Cook, CFO of PayScale customer Mozilla, “I
want to see HR and its ability to track data come into the 21st century. The
minute you can properly track and trend data, HR becomes a true strategic
business partner vs. the traditional stereotype of administrative and soft skills.
Tracking the right information, the information that matters is critically
important. PayScale leads the industry in fresh and timely data. Combined with
their ability to create custom, trend-based reports, strategic HR decision
making is unleashed.”