Cross posted from the Tink Tank.
My girlfriend now runs a company of her own, and one difficulty she’s been running into is setting up a compensation plan for her people.
What follows is some observations I’ve made on salary:
1. Compensation is not about motivating, it’s about loss-prevention.
I liked to think that before I when I wasn’t a full-time manager, the twelve years I spent as a individual contributor to six tech startups, I was a pretty decent performer in a creative field (engineering). I wanted to become the manager that I wished I had, so when the tables were turned and I became a manager again, I would learn what motivated me to guide my decisions toward my own people.
One of those observations was that extrinsic motivators are not actually motivators but, at best, a form of loss-prevention. What motivated me as an individual contributor was the challenge of the problems in front of me; what de-motivated me was when I was when I was being treated unfairly.
“We’ve always take the position that money is only something you can lose on,” Cannon-Brookes told me. “If you don’t pay enough, you can lose people. But beyond that, money is not a motivator. What matters are these other features.”
This paragraph popped up in passing to serve another purpose in the book, but one weakness I noticed in the book was not in investigating what a simple observation arrived independently by two engineers-turned-managers actually outlines a framework for providing effective compensation for organization.
To restate the above: giving someone more salary, bigger options, or a nice job title won’t actually give you better results, but giving too little will demotivate people and you will lose their productivity (or them entirely).
The philosophy is good compensation is not a fount of positive outcomes but damage control against the negative outcomes that can come forth from it.
2. Salary should not be tied to performance.
One time my manager gave me an annual performance review. Because of the criteria of the review structure, I received a mediocre performance review. This was the same year I took my own initiative in writing a product that saved the company and because of it, I was promised a title change by the VP of Engineering and given a larger pool of options by the CEO. I still can vividly recall the feeling of disgust as I got a change in salary that didn’t outpace the rate of inflation and had to sit there listening to my manager rationalize this steaming turd that was just dumped on me.
The first time I had to give annual performance reviews, I noticed I hated to do it and the part I obviously hated most was being required to follow the review with an announcement for what their salary adjustment would be.
So instead, the first year I told my people their adjustment before I gave the review; the second year I ignored the company rules and informed them their adjustment in a separate meeting; the third year, the company policy had aligned to my mode of thinking to allow me to completely decouple the two.
It took the company two years to figure out what I learned so many years ago: even the appearance of salary being tied to performance is wrong. Compensation adjustments and annual reviews occurring around the same time because of budgetary reasons was an error that was never fixed at my last company.
Avoid this mistake.
3. Salary should be set by an overriding sense of fairness.
Your reaction may be, “But if it’s not tied to performance, how is that fair? What should it be tied to?”
The nature of that question actually answers itself. Accepting the first rule and applying it to the second rule creates the indignation in the reaction reflected in the question. The second rule seems to violate our sense of fairness, so let’s make fairness the criteria, not performance.
In the example that influenced me, I just saved the company because without my initiative, they would have run out of money. According to my performance, I did poor because in taking initiative, I didn’t do what I was ordered to do which was the main criteria that an “Engineer – Class 3” is evaluated on. I correctly didn’t get an raise because my salary was tied to that poor performance. But the same exact “poor performance” was what caused the two top people at the company to separately compensate me and spoke to some sort of unfairness. According to “performance” I didn’t deserve a raise; according to “fairness” I certainly did. In fact, this clear violation of fairness standard was so demoralizing to me that I left the company later that same year for someplace else at double my salary.
That’s what I mean about this being about “loss prevention.”
4. Fairness is a shared standard.
People in creative disciplines work with each other and respect each other. People will dislike unfairness whether that means they’re making too little money or too much money relative to their peers. Unfairness is demotivating.
How to implement fairness? While I’m sure there are better ideas, my solution what to do what every mid-level manager is trained to do: I made a spreadsheet.
My spreadsheet had a row for each of my reports (past and present): it had columns for their requisition number, start date, location, job title, and previous years and current salary or hourly wage extrapolated yearly. It also had three other columns. The first was an adjusted effective pay based on formulas for cost of living, contractor/salaried status, etc. The other two contained a log of all possible related special considerations about the person such as recent promotions, promises I made, associated laws that affected them—two simply because the second served as an archive for when the first became so large as to be unreadable.
What I did then was sort the people by their effective salaries and input salary adjustments within HR constraints that would do the most to adjust the sort order based on the simple concept that, “If any of my reports somehow saw this spreadsheet, would they say, ‘Yeah, that’s about what I expected?’” Any area I saw where I’d think one of my people would point and say, “That’s not fair!” I made the largest possible adjustments in order to try to fix.
In this way, I ensured fairness, because fairness is a collective standard.
5. Determining compensation is distasteful work because it crowds out fulfilling work.
I have this long-standing disagreement with a good friend. He’s of the opinion that it’s important to be as open as possible with compensation adjustments like salary and bonuses and options, while I disagree.
Why has it been so difficult for my girlfriend to set up a compensation system? Why did I procrastinate on giving annual performance reviews until after they received their compensation adjustment?
Openness is one way of ensuring fairness, but I think it’s a lazy way that causes more strife than it solves. If I were to actually open the spreadsheet I made to my people, it’d be a signal, “this is worthy of your attention.” But the whole purpose of the first rule is to try to say, “compensation is not worthy of your attention, in fact it belittles your importance because it presupposes you are reducible and motivated solely by a number: your salary.”
I eventually gave access to the spreadsheet to the future managers of my teams. My belief in the importance of fairness without openness meant that I didn’t even have an archive of this spreadsheet when I left.
You may be doing the same nominal thing as poor companies that are secretive with compensation, but the reasons for arriving there are quite different. That difference is everything.
6. Do distasteful work
It is my belief that if you want to be a good manager, you should be value your people. This means valuing their time so you don’t waste it with things unworthy of it. This means doing distasteful work so they don’t have to.
Ignoring the impracticality of doing this with 30+ people and still having time to do anything else, many of my colleagues looked down on the fact that I didn’t have biweekly 1-1’s with each of my reports. However there were two 1-1’s I did every year and did for longer both in preparation and action than any other manager though it would take me months to complete it: the first was their annual review, and the second was a compensation discussion. In fact they were the were the only two clear violations of the rule I kept with my people, “You don’t report to me, I report to you. You to call me in for a meeting and I make time for you, not the reverse.”
In the compensation meeting, I didn’t spend much time talking about their actual salary. In fact, I make a point to not know it—the only thing I recall is the days I spent sorting and adjusting it on a spreadsheet. Instead I give them a two page FAQ (included below) and I spend an hour with each of them explaining how I arrived at my way of compensation. There are only three things I want them to get out of this discussion:
- When it comes to their compensation, fairness is of paramount importance to me.
- I spent a lot of time agonizing over their salary and I think the whole concept of determining compensation is disgusting.
- I hope they think I spent too much time agonizing over it, because if I did, then it means they don’t have to.
And isn’t that what you want to do as a manager? Do the things that need to be done so that you enable your people to do greater things than you could ever do.
WMF Features Engineering Compensation Discussion Notes 2014
If you received a salary adjustment, it will be effective on 7/1/14. You will see it on your paystub on 7/31/14 with an adjustment reactive to 7/1/14.
Globally: How average salaries are determined?
WMF contracted a study from Radford, the premier tech salary survey. This allowed them to do the following:
- Eliminate gender pay differences
- WMF compensation overall are in the top quartile of non-profits
- WMF compensations are in the 50th percentile of tech companies in the Bay Area not considering incentive plans (bonuses, options, etc).
Philosophy of compensation in Features
People are happiest when salaries are fair. Within some limitations, I control Features comp only, so my goal was to imagine how people would react if everyone saw each others salaries. Would you think it fair, or not fair?
How this affects Features
Salary bands from the Radford study should be “under the hood” and only made available to managers and recruiters. However, at this time, even I don’t have access to the actual Radford data or the bands, so I couldn’t use them.
Instead, I was given guideline target as to a total percentage salary increase across my department. This guideline was 6% overall, but reduced by discounting new hire eligibility.
I made adjustments within guidelines, but because of over budget in Tech, a second pass was made involving COLA adjustments for non-US and a reduction again in recent hires, as well as a class of people who had already received their COLA adjustment.
The overall change in Features ended up being much less than 6%: It ended up being less than 5%.
Process in Features
First, in Features, I created three classes of salary ranges based on position/title:
- Senior Engineers & Leads (people with title)
- Engineers who are junior in nature (recent hires)
These salary ranges have some overlap. Also, there were two issues:
- Group (1) and (2) were overlapped, and were both toward the bottom of the salary band reserved for them.
- CA law passed effective Jan 2013, made it so that nearly the entire class of Group 3 salaries were over the minimums for overtime exemption. I opted to just raise the minimum range of Group 3 (their salaries) to compensate, instead of pay them hourly.
The first impact were adjustments related to the ranges. Then, if people belonging to the group, fell outside the minimum of the range, then I opted to use the largest increases to attempt to bring them into the range.
The above are baseline settings/targets. Not everyone was able to get there, but I actually got most in there.
After that, I used increases to try to account for special cases (marriage, children, etc) within a given range. For example, if you recently had a child, then you’re still within your salary band for your group, but you’re more likely to be toward the top end, if I had room to do so.
Finally, I made adjustments within the range, based on your annual review feedback. This would mean that not accounting for the previous, if your contribution was more.
What else went into target setting?
I adjusted 15% for people paid hourly as opposed to salary (this is how HR computes the current benefit difference for contractors). I adjusted based on COLA using the same tables that HR uses to compute contractor salaries: Use state-by-state tables to compare California to your state; Use UN data to compare the United States to your country.
Keep in mind:
- This method will make people move into the appropriate pay range over time. There are larger jumps for promotions and corrections, but it can’t be done all at once.
- Despite different starting salaries, similar performance will converge on the same part of the pay range. If you received a large increase right now, the increases may slow over time.
- Over time, even small differences in pay can create significant differentiation over time (think compound interest).
- Promotion is any move to a higher grade or pay range. Movement (like between Senior Engineer to Lead) is considered a lateral move from a salary/compensation standpoint.
- All of us will take a “hit” relative to industry because of equity (lack of Wikishares ;-)).
- The WMF has a policy of trying to make extra-salary compensations (health benefit, etc.) egalitarian and available to all which affects things.
- I have a maximum I am allowed to increase an individual
- I have a limit that was placed in Features
- I don’t think the COLA adjustments are accurate or fair. I asked that HR if we can standardize on a better set of tables in the future
- After, I set it, some adjustments were made because the rest of Tech was overbudget. Unfortunately, I was not rewarded for this, and some adjustments were made in my targets that threw things off by a little here and there. Luckily, it’s a small amount overall, but I don’t think the adjustment was the most fair.