A study in simple data visualization.
Scott BreudecheckBlockedUnblockFollowFollowingJan 7The tech scene in San Francisco is famously agile: companies pivot, products iterate, and employees change jobs.
It is not uncommon to work for a firm that celebrates an employee when they hit short one- or two-year anniversaries.
This happens at startups (typically only a few years old themselves) but also at the larger behemoths.
As a result of the frothy labor pool, employees have become used to the weekly -if not daily- barrage from Facebook’s recruiters, contract recruiting firms, and VCs reaching out on behalf of their investments.
It seems recruiters are the real winners in this industry: rewarded for poaching but rarely for retaining employees.
But surely some companies have figured out how to keep employees, and thus stop the endless LinkedIn-mining, phone-screening, Lever-reviewing cycle that can sap away a team’s time.
With that thought in mind, I was excited to stumble across survey data that Blind put out last year around this very issue.
Blind is an app where employees of trendy, large tech companies can anonymously weigh in on topics.
In this survey, Blind asked users two to respond to two statements:I’m compensated fairlyI’m interested in changing my job/companyThey then took those users’ stated employer and provided aggregated results.
Bearing in mind this is voluntary survey data on an unrepresentative sample of employees, it is still interesting to spot the variation among companies.
Blind portrayed this as a stacked bar graph.
While this graph has its advantage in that you can quickly find a company and see where the stack up, I found it (and the subsequent publications that picked it up: BI, Inc, Dice, SFGate) disappointing in conveying any real insight.
At best this is a scorecard telling us where a company ended up, but failing to tell us why they might be there.
So I dug into the data a little more.
First the distributions of company responses tells an interesting story.
On the left I’ve built a histogram of the number of companies and how their employee/respondents agreed with the “fairly compensated” statement.
Most companies sit in the 40–70% range with the median being right at 50%.
This means that the “typical” feeling at most companies is divided roughly evenly: employees (as a group) aren’t feeling taken advantage of, nor are they swimming in salary satisfaction.
On the right is the histogram for “changing company”.
Here the graph shifts heavily to the right, the median is closer to 70%.
Very few companies have a value less than 50%.
This means that most employees at nearly all major tech companies are at least thinking about leaving!But are these related?.In other words, does paying more get people to stay longer?.While the true gold standard here would be to run a controlled experiment (randomly pick half of employees, raise their pay, monitor retention), we still might be able to tease out some takeaways from the Blind data.
Blind attempts to address this with a double-bar graph filtered down to just 10 (of 30+) companies, but I find it too busy to take away any lessons.
By converting the data into a scatter plot, with each company plotted by “fairly compensated” agreement (x-axis) vs “considering a change” (y-axis), some trends start to emerge.
You can hover over individual points to see the company.
Right away you can see a strong downward trend, companies with more employees feeling well-paid see fewer employees considering an exit.
The correlation is quite strong at -0.
So pay has a strong impact, probably unsurprising to most of us.
Almost everyone at Netflix apparently is paid well (88% agree) and few are interested in leaving (27%), whereas at the opposite end Hewlett Packard Enterprise (HPE) has a pay problem (13%) and a retention issue (91%).
But I’d go farther: you have to pay well to keep your team around.
These survey data show no company has found a cheap, silver bullet that keeps people around; despite all the work on mission statements, wellness programs, and swag.
As a rule, every company with a compensation gap has a high change consideration.
This graph also shows that employees that feel well-paid (right side of the graph) can still have high rates of looking to quit.
At the extreme, NerdWallet respondents were one of the most satisfied with pay (78% agreed with “I am fairly compensated”) but also one of the least interested in staying (90% agreed with “I am interested in changing my job/company”)!.Intuit and LinkedIn had similar pay agreement (61 and 63% respectively) but Intuit employees were twice as interested in leaving (80%, compared to LinkedIn’s 40%!).
High pay by itself is not enough!Getting your company on solid compensation satisfaction seems to be a necessary, but not sufficient, strategy to address employee retention.
It should be considered table-stakes in the revolving door culture of SF tech.
I’d start by checking the pulse of your employees: if most people feel like they are overdue for a raise, expect to see an a regular exodus.
Some obligatory footnotes:Thanks to Blind for doing the survey.
It was a great PR move, I had never heard of the company until a podcast I listen to picked this survey up as a story.
But no content marketing-driven survey is perfect.
Data are now over a year old (so I’d guess Facebook’s seen some increase in “interested in changing”…), and have a response-bias as well as a selection-bias on Blind-users.
I could imagine a systemic issue where employees who are considering a job change engage with Blind to learn more about other companies.
On top of that, the numbers I used are my manual interpretation of Blind’s graphs.
It’d be cool to see how this compares to, say, Glassdoor review and salary data.
Finally, I’ve stored my data and python analysis notebook in github should anyone be interested in diving in.