As CTO at Cozy, I came up with a framework for compensation that I was able to mostly apply to the engineering/ops/fraud teams, and tried to impart onto other groups though not always successfully. There are a few parts to this framework, but nothing about it is complicated.
Some disclaimers: I’m not claiming anything here is so novel. It builds on the work of many better managers and researchers than myself, just one of whom I cite. I also want to warn that this is how I, personally, thought about compensation. It was never (unfortunately) agreed policy, nor were my actions as rigid or structured as it will seem here. I also don’t claim this system will work in your circumstances, but hopefully it provokes some thought about your own compensation practices.
I broke the framework down into a number of topics that build on each other:
- Salary Bands
- An Equitable Definition of “Seniority”
- Compensating Management
- Facilitating Career Growth
The first step in the framework is also the most formal one. Come up with salary bands, separating bands however you want. The most basic would probably be junior, mid-level, senior, and principal. I would also suggest not including “management” as a separate track or band. I will talk more about this in “Compensating Management.”
You should also have separate “tracks” for groups that you want to have different pay scales. This can be based on discipline, like bands for engineering, design, customer support, marketing, etc. I’d recommend, though, seeing if there is another way to split tracks, like between entry-level roles (customer support may fit here), professional roles (marketing, growth, trust & safety, graphic design), and specialist roles (engineering, maybe product design).
In some cases, you may want to build in some sort of modifier into a track for hyper-specialist roles, like Site Reliability Engineers or skills that are in particular demand, possibly local to your market.
You should come up with the actual salary numbers based on “market” rates, biased by what you can/want to pay. If you need to or want to pay under-market, that’s fine, but be honest about it and the value proposition you offer. There’s plenty of research showing that people want and are happier with fair/transparent compensation even more than additional compensation. And ultimately, you need to provide a fantastic work experience if you want folks to stay. Salary is a small part of retaining people, and if you’re bribing people to stay, you’re going to introduce dysfunction.
Most importantly, don’t base salary band numbers on existing employees, since one of the points of having a compensation framework is to reduce inequities in pay, and if you base it on existing employees you’re going to build that in. You can look at existing employees when defining “Seniority” next.
An Equitable Definition of Seniority
People generally agree that we should be paying people and awarding seniority based on their “value” and “contributions.” This sounds fine, but I find it leads to all sorts of dysfunction. I have what I think is a better option, but first I’ll explain the problems with paying/defining seniority by “value.”
The problems with value-based rewards
Value-based compensation in most industries is inherently subjective. Managers determining compensation determine value. And since most managers think they are doing very important work, people that align more closely with that work will naturally be valued more. I have seen first-hand how this manifests, and the dysfunctions it creates:
- Employees with more institutional knowledge are quick to provide solutions and are often the first point-of-contact for managers. Rewarding institutional knowledge incentivizes knowledge-hoarding.
- Employees who are responding to live issues solve immediate problems for managers. Rewarding bug-fixing may or may not incentivize bugs, but it definitely incentivizes folks to stop what they’re doing and fix live problems, rather than incentivizing management to create a sane system to handle inevitable live issues.
- Employees who are not working on the current strategic directive at a large company are undervalued, even if what they are doing is extremely important. They become frustrated working in a backwater and leave.
- Employees who do work unfamiliar to their manager are not compensated effectively because the manager doesn’t really understand what the employee does. This leads to overpaying or underpaying employees, both of which are inequitable.
Notice that none of these factors are in the control of the employee! They are generally going to be results of the work they are assigned to do or what they are naturally inclined towards, strategic changes, or unfortunate management arrangements. It is really frustrating for employees to be compensated this way. It’s also a likely vector for discrimination to sneak in, since it’s highly subjective.
One negative result of value-based evaluation is that managers end up slicing compensation adjustments very finely: 8% to this person, 10% here, 9% there. You believe you have a weigh to rank an individual’s value, so you feel capable of making fine-grained salary differences. Over time, these differences breed resentment and ill-will (and, knowledge of unconscious bias would tell us, more discrimination).
In the face of all of these arguments that value-based pay is problematic, what about its main benefit: that is, value-based pay incentivizes producing value? Well, there is extensive evidence performance-based pay does not work. See this Phil Haack article for a great rundown and supporting material. If you are really a staunch believer in performance-based pay, I don’t quite know what to tell you. I’d love to hear why you support it.
Predetermined, Transparent Compensation Schedules
I’d argue that adding a predetermined, transparent compensation schedule into a salary band track is a better choice, and introduces far less dysfunction, than “value-based” pay. It does, though, require a certain type of environment that not all companies will be able to achieve.
Predetermined and transparent compensation schedules means that you only need to subjectively decide someone’s “seniority” (place in the salary bands of a track) once. It’s generally a function of how many years experience someone has, and their skills/expertise. Almost anyone with 0-2 years experience is going to be in the lowest band. Someone with 6 years of experience can be in the senior band, or someone with 10 may in the mid-level band. This is also going to be different for folks that are moving around internally vs. hired in externally. It’s a coarse-grained assignment so you’re much more likely to get it “right” than yearly fine-grained value-based adjustments.
Once someone’s initial band is determined, you should have a clear idea of what their salary change is each year of employment. I found it very helpful to give about the same dollar adjustment to everyone in a track. This is a larger percentage change at the junior end of a track, reflecting the faster growth I expect of those employees. It’s a lower percentage change as folks become more senior, since I believe their contributions and growth are “baked in” to their salary at that point. In other words, what I’m compensating at the junior end is a different set of expectations than the senior end, not some absolute measure of “value.”
I know this is a topic I need to explore more. Ultimately, I believe percentage-based adjustments are fundamentally inequitable, and likely to lead to higher turnover due to inappropriate salary growth.
I’m a big fan of quarterly salary discussions, rather than a single yearly adjustment for the entire company. One reason is that, you will inevitably end up with folks that were placed in the wrong band, have growth quickly to another band or higher in their existing band, or need a few extra months to grow into their next band. I have found that because you’re generally working with coarse-grained schedules, it’s relatively easy to find where someone sits, and it will be clear if you sat them wrong. But, you should have a regular system to make those adjustments in, not yearly or adhoc.
“Manager” should be a salary/seniority band. Management is, to me, an alternative role for someone, not itself a level of seniority. There are a number of mid-level employees I’ve worked with who would make great managers, and many senior folks don’t want to move into management.
In the worst case, someone who doesn’t do well as a manager can go back to being an individual contributor without a pay cut or being overpaid.
In practice, becoming a manager generally increases someone’s impact and responsibilities and they see a somewhat increased pay cadence into the “next band” once they have proven their abilities.
Facilitating Career Growth
I’ve discussed predetermined and transparent compensation enough in the positive case of someone moving through the bands more quickly than usual. But what about the negative case, where someone is struggling?
This has been, to me, where this compensation system shines.
If someone is struggling, you can have two types of discussions.
The first is that they are struggling, and you are paying them less for it. This is a catastrophically bad discussion to have:
- I have never, not once in my career, seen someone struggle for a year without seeing tons of systemic problems and management mistakes. If someone is not going to work out, you generally know very quickly. If they have made it a year, it’s due to management’s shortcomings. Don’t punish your employees for your mistakes.
- Most people are not aware of their struggles. Telling them their performance is poor at the same time you’re punishing them makes is antagonizing. They didn’t get a chance to prevent a situation they’ll now have to deal with for a year (or more likely the rest of their tenure).
- People need confidence to succeed, and you’re telling them they are not competent. They will wonder if you even want them to succeed or if you’re hinting they should quit.
- Most of our companies are committed to inclusivity, yet women/LGBTQ/minorities are much more likely to struggle. You will end up paying them less, frustrating them more, and they’ll be most likely to leave.
The alternative conversation is something like this: we love having you here, you’ve grown and accomplished a lot. You’ve earned this! I want to make sure you’re on the right trajectory for the future, and over the next year I want to focus on X, Y, and Z.
Employees should walk away from this feeling supported and that their manager is their partner, not an opponent.
The cost of the alternative approach is generally a few thousand dollars in the short term. In many cases, you will have put a capable employee back on track, in which case the cost is really a hugely positive investment. In some cases, you’ll still have to fire them, and a few thousand dollars compared to the cost of employee turnover is almost nothing. The case you avoid, though, is the most common and damaging case: the employee struggles along for another 9-12 months and then leaves.
Many companies provide equity and benefits. Great.
Fred Wilson has some excellent material on deciding how much equity to give employees, including this blog post. He advocates including equity-converted-to-dollars when you make an offer to an employee (ie, 15,000 shares at $1/share, for additional compensation of $15,000). I concur in this case, as it helps a bit making equity apples-to-apples between companies.
That said, I would never factor in equity or benefits into dollar compensation numbers after that. Great benefits are expected in our industry- if you offer poor or mediocre benefits, it’s a huge strike against you. “Refresher” grants are generally going to be smaller than an employee’s initial grant, and they are less likely to ever be able to exercise them, so they should be considered more like an extra potential bonus. If you start to apply accounting gymnastics and give $3,000 salary increases but with $5,000 in equity, and say it’s $8,000, you are insulting them. My experience is that they’d rather be treated honestly and equitably- if you need to award equity instead of salary because you need to cut costs, be honest about it!
One question I’m left with: why not use a “compensation calculator” like multiple companies publish? I’ve thought about this from the start. I’m applying an arbitrary amount of formality to what has traditionally been a very subjective process (with expected inequitable results). I don’t have an answer. Lack of imagination, maybe. If you’re brave enough, I would encourage you to experiment with whatever you can stomach (either full transparency like Buffer, or a calculator like Stack Overflow.
I hope it’s clear that I’ve not advocating my approach for every business. At Cozy, we built a rewarding and cohesive culture that people wanted to be a part of. On engineering in particular, folks of all experience levels were constantly learning and growing. We had great satisfaction and retention results. I can’t say how much of that, if any, was due to this salary framework. Maybe it was a net negative, though I have only evidence it was a positive.
I have no idea if this has been helpful to you. It was helpful to me to try to structure the system that’s been mostly only in my head for the last couple years.
Also published on Medium.