Sunday, March 24, 2013

The Barefoot Philosophy Cornerstone #2: Merit-Based Pay

[This is part 3 of an 8 part series: The Barefoot Philosphy.  It is based on my experiences as the founder of a business—Barefoot Software—which I ran for 12 years.  Please start with the intro.]

Last week I promised I’d tell you why it doesn’t matter if everyone knows how much everyone else makes.  The answer, of course, is simple: pay everyone what they’re worth.

Netflix puts this slightly differently: they call it “paying top-of-market.”  They encourage their managers to ask this question:

What would we pay to keep that person?
— If they had a bigger offer elsewhere1

I don’t like this quite as much as the Barefoot philosophy.  For instance, Netflix goes on to say:

Some people will move up in comp very quickly because their value in the marketplace is moving up quickly, driven by increasing skills and/or great demand for their area
Some people will stay flat because their value in the marketplace has done that
—Depends in part on inflation and economy
—Always top of market, though, for that person2

This makes good economic sense, but I ain’t here to tell you how to make good economic sense.  I’m here to tell you, in case you’ve forgotten, how to make your employees deliriously, ecstatically happy.  Here’s how we did it at Barefoot.

In order to get hired as a programmer,3 you had to write some code for me.  I reviewed this code with you, and we talked about it.  I pointed out things you’d done wrong, and you told me why you made those decisions.  Perhaps there were factors I couldn’t know about just from reading the code.  Perhaps you just didn’t know any better.  At the end of this process, I knew how good a programmer you were, relative to me.

If you were, for instance, half as good as I was, you made half as much.  If you were twice as good as I was, you made twice as much.  Or whatever percentage was appropriate.  That simple.

There are a few technical details you have to contend with if you want to make this work.  First of all, you have to jettison the silly notion that you can only give raises once a year.  This is a terrible idea anyway.  It produces what I like to call “Academy Award syndrome.”  You ever notice that powerful, serious dramas only ever show up just before the Oscar nominations are announced?  Meanwhile, the period just after the Oscars are awarded generally sees the worst cinematic dreck ever regurgitated hit the movie screens.  Do you really want your employees doing their best work in the month or so leading up to raise annoucements, and slacking off all the rest of the time?  Hell no.

Barefoot had what I call “anytime raises.”  You could get a raise any time, for any amount.  From my Employees First manifesto:

I gave raises whenever you showed me you were better than your rate said you were, and the raise was for as much as you deserved.  I gave someone a 50% raise once.  I gave someone a raise once before he ever got his first paycheck, and I made it retroactive to his first day, because he demonstrated that he was better than I thought he was when I hired him.

Which also demonstrates that you have to be willing to admit you screwed up in your initial evaluation.  If you underestimated, you just fix that by raising their rate appropriately.

But what if you overestimated?  Well, I tried to be very conservative when evaluating for that very reason.  It’s always easier to give someone a raise than a pay cut.  But, if you’re committed to merit-based pay, you have be willing to do just that.  Not only can you overestimate someone initially, people can become less productive over time.  In an environment like the one we had at Barefoot, that was very rare.  The whole point of an employee-focussed workplace is to get everyone doing their utmost to push themselves to be the very best they can be.  But sometimes it just happens that people slip.  When that happens, you just have to sit them down and say, “I’m sorry, but I have to reduce your pay right now.  When you get back to where you were, we’ll put it back.  But right now this is what you’re worth.”  Remember: you aren’t doing anyone any favors by overpaying them either.

Valve does this even better, I think.  They use what they call “stack ranking,”4 which takes into consideration a lot of different factors, from technical ability to group contribution.  I’ll freely admit that they have a more complete handle on this, although at Barefoot certain things (such as productivity) were factored in to pay scale, but only after initial evaluation (which was completely based on technical ability).  But, the point is, your pay rate reflects your value to the company.

This is important, and it not only made it okay that you knew what everyone else made, it made it vital.  Your pay rate was what you used to assess your position within the company (in a couple of weeks we’ll talk about why you couldn’t use your job title for this).  By comparing how much you made to how much that guy over there made, you could tell if he or she was better than you.  At every other company I’ve ever worked for, the fight (silent, but deadly serious) over who is the “top dog” programmer was an ever-present distraction from getting the work done.  Not at Barefoot.  You knew who was better than you were, and you either accepted it, or you didn’t.  If you didn’t, there were two possibilities.  Either you thought the ranking was wrong, which inspired you to prove to me (or one of the other top guys) that you were better than we thought you were.  Or you knew the ranking was right, but you wished it was wrong, which inspired you to improve your skills and earn a better ranking.

And knowing where everyone stands in relation to each other is important if you want to have self-forming teams.  (Valve talks about this briefly.5)  Every team needs to know the relative skills of the people they’re recruiting.  And don’t worry about the less experienced people never finding a spot.  It only takes a very small amount of exposure to self-formed teams to figure out that you need those “junior” folks.  Lead singers get all the attention, but drummers and bass players keep the songs moving.  A team composed of nothing but “senior architects” is the most disastrously useless thing in the known universe: they all know exactly which direction the team should go in, and it’s never the same one.

Merit-based pay is also a self-regulating system.  People who aren’t that productive either get better, or they get gone—they have to, in order to pay the rent.  On top of that, with self-forming teams, the underperformers never get “picked.”  I never fired a single person in all my years at Barefoot, but that doesn’t mean that we never let anyone go.  There just wasn’t any work for them, eventually (which works because of next week’s cornerstone).

It also goes even further along the path that we started with financial transparency: there is no division between the people who run the company and those who work for it.  Myself and the various people who served as CEO, or COO, or CFO, got the same merit-based pay as everyone else.  We weren’t “special” because we had been there longer, or because we were nominally in “charge.”

This theme of equality also extended to benefits.  I note that Netflix figured this out as well: they offer a flat yearly amount for health care coverage; if you don’t use it all, you pocket the difference.6  At Barefoot, it was even simpler: we didn’t pay for any of your health care.7  Of course, our pay rates were higher to help compensate for that (but then, as we’ll see next week, they were higher for other reasons as well), but the point was the same as it is at Netflix: you pay for what you need, nothing more.  And you aren’t entitled to more health care just because you’re married, or you have children.  Why should you compensate your unmarried employees less?

Now that we know how to compensate employees relative to each other, we can look at the technical details of how to express pay rates.  Next week.

Part: << 1 2 3 4 5 6 7 8 >>

1 Netflix Culture: Freedom & Responsibility, slide 97.

2 Netflix Culture: Freedom & Responsibility, slide 102.

3 The process was different for non-programmers, but the philosophy remained the same.

4 Valve Handbook for New Employees, page 27.

5 Valve Handbook for New Employees, page 10.

6 Netflix Culture: Freedom & Responsibility, slide 109.

7 This is not quite true.  There was a way to get the company to cover part of it.  But we’ll come to that towards the end.


  1. enjoying this series. keep it up!

    1. Thanx Ed. I've already got the whole series written, so you'll definitely get the whole thing. Don't worry: no cliffhangers. :-D