Part 3: Compensation for Startups: Implementing and “Defending” Your Compensation System
This article is part of a series on compensation for startups. It covers implementing and defending your compensation system. The four parts are:
If you are actively designing and implementing any of these systems, we also have some Do-It-Yourself packages you can buy that include a recording of us explaining all of this in more depth as well as dozens of templates.
Ok, friends! We’re back with the next installment on building scalable compensation practices in a startup. This post is focused on how you implement a compensation system once you have designed your philosophy. If you haven’t already, go read the post on designing your philosophy as it’s the right place to start.
Implementing a compensation system basically has two phases: 1) initial implementation and 2) maintaining the system as you grow (or defending, depending on how you think about it). This post will mostly focus on initial implementation but I’ll also talk a bit about how to maintain and defend the system so you don’t end up with a big mess.
Before we get started, I just want to make one additional philosophical / design point (on top of the ones I made in the Philosophy post) that is critical to implementation: Simple is sustainable. Let me explain… The worst case scenario is that you do all this work – design a new compensation philosophy and make all those hard decisions, do the work to map your employees to your new system, communicate your new philosophy and system to the company, etc. – and then, because of day-to-day decisions, your new system falls apart. How does it fall apart? Through a lot of small decisions. A recruiter offers someone a title that doesn’t match the framework. A manager approves a salary that is out of band. You make a counteroffer to a valuable employee who got an offer from another company.
When you’re designing and implementing this system, you have to realize that a lot of the day-to-day will be implemented by recruiters and managers through hiring and through conversations with employees. That means that simple is your friend. As you’re designing, you want to think about the most junior recruiter on your team and ask, “With a little training, can that recruiter maintain this system?” The more complicated you make it, the more likely that the answer is no.
I’ll talk about this more at the bottom of the post, but in addition to simplicity, I would also argue that some amount of centralized control is a good thing to maintain the integrity of the system. You want someone to “own” the system as it moves from initial implementation to a living and breathing – ongoing – thing. I would have that owner put themselves in the middle of important decisions like approving exceptions, etc. If you leave the ongoing maintenance in the hands of recruiters and managers, it is very likely to end up devolving into a mess.
Ok, on to implementation!
Initial implementation of a compensation system
As I mentioned in the compensation philosophy post, the earlier you can design and implement this system in the life of your startup, the less mess you have to clean up. I usually recommend doing it around 30 or 50 employees if not before. I have done it at 200 employees… it just takes longer to implement and the chaos – people outside of bands, titles that are all over the place, etc – is more significant.
There are basically 7 steps you have to go through for the initial implementation, and I will walk through those below. If possible, I try to pair this initial implementation with a performance review cycle. I start the compensation design process about 1-3 months before the next cycle (more time if you have more employees or if you have less experience with this stuff), use the performance cycle to confirm decisions around levels and comp with managers, and then finish the performance cycle by rolling out the comp philosophy and any comp changes to all employees, along with their reviews.
Later in August, I’ll share my standard performance review process and it will explain better how this all works together. At a high level, I try to always tie performance and compensation work together. That’s because (1) I like to reduce the overall time burden on the organization (otherwise you are just constantly in some sort of review as a manager) and (2) I design my compensation systems to tie performance and compensation together. (I talked about this in the philosophy post and will talk about it more in the performance post as well.)
Ok, so here are the 8 steps you need to take to implement a compensation system for the first time:
Create or update your leveling and titling system.
Design your compensation philosophy.
Get market salary and equity data from vendors.
Decide how many “functional groupings” you are going to have.
Design compensation bands.
Map current employees to your new levels and comp bands.
Add in geographic multipliers.
Start making new candidate offers using the new bands.
Let’s go through them in detail…
(1) Create or update your leveling and titling system
There is lots more detail on how to set up levels in the leveling article. Make sure you have the number of levels that works for you, and if you’re using the framework template in that article, customize the language to match how you talk about things like scope/impact at your company. Next, agree on which titles go with which levels, and how to name ICs versus managers.
(2) Design your compensation philosophy
There are a number of significant decisions you need to make that are outlined in the compensation philosophy article. Before you can implement anything, you need to make those decisions.
(3) Get market salary and equity data from vendors
Popular comp data vendors in the tech world are places like Pave and Radford. At some point, every company will need to purchase comp data, but early-stage companies may be able to get it from their VC firms. Consultants can supply it as well. Wherever you get it, if you are using a market-based philosophy, you need to buy market data. Don’t just survey your friends 🙂
The vendor will send you a spreadsheet or access to a website with comp numbers for each job title & seniority level. Radford and Pave both use a similar leveling system to the one in my leveling article, with 8 levels, so another benefit of that system is that you don’t have to do a ton of adaptation or job matching. You’ll be able to just copy and paste data as you go through the following steps. 🙂
To effectively determine your market data, you need to pick a set of companies to benchmark against. Make sure you are benchmarking against companies that you compete with for talent. Typically, you choose your benchmarks either based on a similar amount of money raised or a similar amount of revenue. Products like Pave allow you to sort and filter those data points to narrow down to the right market dataset. It is worth playing with those benchmarks to see how different the data is until you find the one that feels right, based on the people you have already recruited and the salaries that you already pay.
If your company is in a niche industry and you’re recruiting a lot of people who are not in traditional “tech” roles, check whether your vendor has a specialized database for you. Sometimes you need to buy a database if you’re recruiting for a lot of roles where people have backgrounds in specific industries like education, healthcare, etc.
The spreadsheet will show the salary, bonus, and equity amounts that were paid to different roles and levels, divided into percentiles (10th, 20th, 50th, 75th, 90th, 100th). You’ll see percentile numbers for:
Base salary (cash)
Total cash [Base cash salary plus cash bonus] – this number includes base salary and cash bonus
Equity – How much equity is provided over 4 years. You’ll see
The % awarded
The dollar amounts
Pave has some examples on their website so you can start to visualize what I mean. Companies like Facebook, Google, and Apple pay at 100th percentile. The percentile you choose the most expensive decision you'll make. Startups that need to be careful with cash flow usually pay 50th to 75th percentile for salary and a higher percentile for equity (e.g., 90th).
(4) Decide how many “functional groupings” you are going to have.
Technically, every single role has its own market. Finance people have a different market than lawyers. Designers have a different market than product people. Front-end engineers have a different market than back-end engineers.
When your company gets bigger, each role will likely have its own salary bands. But when you are small, I highly recommend not taking on that level of complexity.
For a software startup that doesn’t have a lot of specialized roles, I typically end up with 3 “functional groupings”:
Engineering, Product, Design (EPD)
“Business” which includes Legal, People, Finance, Marketing, etc.
Customer Support
To create the bands, I anchor each group’s band roughly to the most expensive role in the group (e.g. engineering or legal). This can overpay certain roles, and I’m fine with that. You could also find the average or the median if you prefer. You just may need more space in your bands for the more expensive folks.
Oddball roles that might be grouped into business or EPD include things like Developer Relations or Sales Engineering. Their salary group depends on how technical the people are. If they are technical enough to be able to get an engineering job at a different company, then they probably belong on your EPD band.
I’m honestly the wrong person to ask about Sales comp. Go read Jason Lemkin. Every time I have designed Sales comp for startups, I have used one of his frameworks.
(5) Design compensation bands
Once you have designed your levels and your compensation philosophy, bought your market data, and decided on your functional groupings, you are ready to design your compensation bands. There are basically 3 steps to designing bands:
a) Pull base salary numbers for each level
In your market data spreadsheet or tool, filter by a role type like “Software Engineer” and find the column that reflects the cash base salary percentile you chose in your philosophy (let’s say it’s 50th); that column will hold all the base salary numbers for your levels.
Grab each number and put it in your own spreadsheet. For software engineering at companies of a certain size / stage, the market data might spit out numbers that look like this (fake data, don’t use it):
Depending on what you choose in the next step, those numbers can either act as the point salaries for each level or as the rough mid-points for your bands.
b) Decide whether you’ll use points or bands
Deciding on points or bands is really a philosophical point and I talked about it a bit in the compensation philosophy article. But here’s a tiny bit more detail:
Bands: You provide a range for each level; the company can negotiate within that range for new hires and raise salaries within the band (e.g. $190-$225k range for an IC5)
Points: Each role has an exact salary amount that someone starts at, and there’s no negotiation (e.g. IC5 makes exactly $210k). You can still raise salaries based on performance or tenure, but those raises should be formulaic.
I prefer points because there’s less room for bias and discretion, but bands are actually easier for startups to implement, for a couple of reasons.
First, to do points well, you’d basically need a salary point for every job and every level of every job. That’s a lot of points to keep track of. Bands let you simplify by creating the functional groupings that I mentioned above.
The second reason is that to do points well you also need to say no to negotiating … and it takes a lot of maturity and confidence to say no to candidates you’ve been recruiting when they ask for more money; to say no to employees when they have a higher offer from another company but want to stay, etc. To implement points, you have to be willing to defend your system and your points all the time. I’ve made the mistake of implementing a point-based system and then ended up making a ton of exceptions so that we basically ended up with no system at all. Don’t do that.
If you decide to implement points and end up giving $10,000 more dollars to someone, technically you should give it to everyone at the same point. Colleen McCreary’s role-based pay system is points-based, and she has some good thoughts about how to scale that system effectively.
In general, I would say that it is probably simplest to just design bands unless you feel very very strongly about no negotiation and believe you can get the team to enforce it.
c) Decide on the size/overlap of bands
If you use points, then the data that you pulled out of the spreadsheet can basically be the points that you offer each level. You can tweak them a bit based on your existing employees and any other market signals you have, but you are basically done :)
If you decide to move forward with bands then you need to decide the size of your bands and whether they overlap.
The size of bands: I wouldn’t recommend large bands for the lower levels – about a $20-30k spread between the min and the max is standard. But it’s common to make bands wider for higher levels because peoples’ experience varies more and also because senior ICs/Directors/VPs/CXOs stay at one level for much longer, so large bands give them room for salary/equity growth over the years.
I will just note that the wider the band is, the more room there is for bias. Recruiters/hiring managers can randomly decide that this level 4 engineer gets $10,000 more dollars as a starting salary than that level 4 engineer. You can try to mitigate that through your hiring process and also your annual comp review, but it does tend to creep in.
Overlapping bands: In 2008, at Facebook, we designed the comp system to have overlapping bands and high multipliers on performance, so that a high-performing IC3 could be paid more than an average-performing IC4. The purpose of that design was to focus incentives on performance rather than promotion. That approach is a little advanced and can get complicated, but it’s something to consider if you want to aggressively reward performance. By overlapping bands, I mean something that looks like this:
After you are done with salary, do the same exercise for equity
Now that salary bands are done, go back to your market data, find your equity percentile, and pull in the benchmark equity at each level. It will end up looking something like this (still fake data, don’t use it):
Vendors like Pave also typically provide dollar values for equity in addition to basis points (what’s above). As with the salary numbers, these equity numbers can either be used as points or the mid-points for your equity bands.
The biggest complexity in deciding equity bands is figuring out how to value your company, particularly these days. Typically, companies use the last post-money valuation to determine the size of equity grants. With so many overvalued rounds from 2021, right now, post-money valuation can be tricky from a recruiting perspective – sophisticated candidates are going to be very very skeptical of accepting an equity grant based on a post-money valuation that is 100x revenue.
What I’ve been seeing somewhat regularly is startups using a discounted post-money valuation to shape their equity approach. Whatever approach you take to the valuation you use to calculate equity grants, it’s important to think about how to handle it with candidates and consult your lawyers to ensure you are not hurting yourself in any way.
As I said in the philosophy post, a lot has been written about designing equity compensation systems so rather than try to go in to more detail, take a look at the post on Carta’s equity program or even Emily Kramer’s equity compensation explanation (though it is focused more on the candidate point of view).
(6) Map your current employees to your new levels and comp system
You’ve designed your new comp system, congrats! But you have 50 employees who’ve been paid somewhat randomly up to this point… How do you move them onto the new system and make sure they’re at the right level and paid the right salary/equity?
Mapping your existing employees to your new bands is a super healthy process to ensure that your new bands and levels are correct and also to start to iron out the mess you have created by not having a system. As I mentioned in Step 1, I recommend scheduling this during a performance review cycle.
The process is essentially like two sides of a sandwich (yum).
Side 1: Based on your new leveling framework, what level do your managers think their employees are?
Side 2: Based on your new comp bands, what level is implied for each employee based on their salary?
Whatever process you use (I’ll explain mine below), you need to bring those two sides of the sandwich together because the discrepancies will tell you things like who needs to be promoted, who might need to be fired, or where there might be problems with your comp/leveling system.
Here are the steps:
First, get a list of everyone’s current salary. If you’ve been using geo multipliers (meaning a multiplier that increases or decreases the compensation for folks based on their location – a bit more on this below), you’ll want to remove them. Essentially, you want to bring everyone back to a Tier 1 (SF, NY) salary equivalent. Apples-to-apples comparison is essential for leveling.
Compare each employee’s salary (Tier 1 equivalent) to your new framework/bands. Ask, based on what they’re currently making (without geo multiplier), what level should they be, assuming your framework is correct? This spits out a level number for each engineer, marketer, etc. (e.g., If they’re making $200k as an IC engineer, they’re theoretically a level IC5, etc). This exercise creates one side of the sandwich – a list of employees and their theoretical new levels based on what they’re currently being paid.
Then, create the other side. You have two options for how to do this. You can bring the list of people and implied levels that you created in step 2 to your managers and say “Does this look correct?” OR you can bring them the leveling matrix and ask them to independently level each employee. This creates the other side of the sandwich – a list of employees and their levels based on what their managers think.
When you put the two sides of the sandwich together, you will see where there are discrepancies. Those discrepancies typically come from a couple places:
it’s possible that the band is wrong in some way (but before you assume that, consider the next two options)
the easy one! Someone’s manager thinks they are a higher level than their salary indicates – maybe that person needs to be promoted.
the hard one :-/ Someone’s manager thinks they are a lower level than their salary indicates. Are they underperforming? Did we overcompensate them when they joined because they negotiated heavily? Did we make some weird special agreement with them when they joined (trade equity for cash, etc)?
You basically need to handle each of the discrepancies individually and figure out if it is specific to the person or a problem with the system. You can also use the process of calibration (the next step) to discuss the individuals and ensure that they’re not leveled incorrectly..
Finally, you need to calibrate. This is one of the reasons I like doing this compensation work in tandem with a performance cycle. You can use the calibration step of the performance process to confirm the levels. I’ll go over this in more detail in the performance review post, but the basic way calibration works is a large meeting with managers who manage similar employees (like all EPD managers in one calibration) and you go through employees level by level. You have discussions that basically look like: “In theory, these three people are all level five. Here are the expectations of a level five. Are they all fives?” The main point here is that it’s healthy and valuable to get all managers together to ensure that folks are applying the same standards to their employees as they think about levels. Ideally, you do this before you communicate levels to employees.
Those are the steps to map your current employees to your new salary bands, bringing everyone up to speed with the comp system.
Before we move on, another thought about what to do if you find a big discrepancy — you’re paying someone $250k when they should truly be at $190k in your new system. Do you take it away? To be honest, taking away compensation is equivalent to telling someone you are fine if they quit. I would typically recommend a different conversation about holding that person to the expectations of a $250k employee. In all honesty, though, this is when you should be asking yourself what $250k can “buy” you in the market. If that person left, and you could spend $250k on a new employee, what could you get? Usually, it means this person isn’t a fit in your organization. This is what I mean about this process exposing performance issues. For better or for worse, this process can put real dollar numbers on those choices.
(7) Add in geographic multipliers
When you are doing the initial stages of implementation, you need to look at employees' apples-to-apples, so as I said above, you need to take geographic multipliers out. However, if you plan to use geographic multipliers – meaning paying people in different locations different amounts – at some point, you need to put them back in.
The basic process is to design your most expensive “Tier 1” salary band first — the most expensive city like SF or NYC. Use that salary band to level everyone and do the apples-to-apples comparison. Then, for people who are in a different geo, you’ll use a multiplier on those salary band numbers to adjust downward (e.g., “people in XYZ city will make 0.9x what people in SF make”). Pave has a pretty good summary of this and part of what you can buy from them is geographic differentials.
(8) Start making new candidate offers using the new bands
As soon as you feel confident in your new bands, start using the new salary bands to make offers. I wrote more about this below, but you need to make sure to train your recruiting team on the new system and philosophy before they can start making offers.
That said, the longer you wait to start using the new bands to make offers, the more people you’ll bring into your old system, and the more work you’ll have to do to adjust them and move them onto your new system. Sometimes it feels odd to offer new employees more than what current employees in the same role are making, but at some point, it is usually the right call so you don’t accrue more “debt” in terms of employees you have to adjust.
How to maintain/defend your new compensation system
As I mentioned at the top, it is as important to figure out how to maintain and defend your compensation system as it is to implement it. Implementation without proper thought to how it will be supported is just setting yourself up for having to do another major redesign. But if you have a good process to defend the integrity of the system, this system can last for many years.
There is a kind of process that I call “happy friction” – in my mind, that is a process that makes it more difficult to do the things that should be difficult. In the case of maintaining a compensation system, I like adding “happy friction” around ANYTHING that breaks the system.
There are 4 things I would do to maintain the integrity of the system in an ongoing way:
A) Make guidelines and train your recruiters and your managers
The most important thing to maintain and defend the integrity of your system is to realize that your work isn’t done after you implement the system. You have to create clear training and communication about the new system for your managers and your recruiters so they understand it and can help maintain it. The day-to-day reality of this new system will be carried out through their interactions with candidates and with existing employees. As an example, if a recruiter or a manager chooses to make an offer to a candidate at the top of the band (or above it) and that person now has no room to grow inside their band, it starts to affect the whole system. If your recruiters and managers don’t understand the principles of the system and their role in maintaining it, it is much more likely to fall apart.
B) Make an exceptions process
You should try to make zero exceptions to your comp policy because it undermines your systems. And yet … exceptions will almost always come up. The best thing to do in advance (now) is to determine who has the authority to approve exceptions and design what that process looks like.
I am a fan of having it be HARD to make exceptions. That could look like creating a process where the CEO has to sign off on exceptions. This puts weight on exceptions and makes them scary – the more weight, the less often you’ll do it. You can also make it so that the Head of People or the Head of Finance owns exceptions (or both).
Whatever process you chose, I would make absolutely sure that people like recruiters or hiring managers definitely can’t approve exceptions. Those folks care more about getting the employee they want in the door than they do about maintaining the integrity of the system. You want exceptions to be approved by the people who feel ownership and responsibility for the integrity of the system.
C) Create an offer approval process
In the spirit of both simplicity and “happy friction”, I would typically say that any offer that follows your guidelines – titles, etc. – and is within your bands does not need to be approved. Anything that is out of band, or in any way bends the guidelines or the bands, should go through an approval process. If you want things to be as simple as possible, have just one process to handle these types of approvals and any other comp-related exceptions.
That said, when the system is new, you may want to have someone review and approve all offers for a 3- to 6-month period to ensure that recruiters and hiring managers understand and are following the new system.
D) Create an annual review process
After your system is rolled out, I suggest looking at compensation for the whole company once a year. This is pretty standard practice. Whatever timing you decide, communicate it to the team and really discourage discussions about compensation outside of this time. It is key to maintaining the integrity of your system to not promote people or adjust people’s compensation outside of the scheduled times.
When I design these systems, I typically have two performance cycles per year, and one of them has a full compensation review attached to it. The other one only has promotion-related changes.
A 1x per year full comp review includes changes for the whole company:
Adjusting all salaries relative to market (“market adjustments/raises”)
Performance-based raises
Additional equity grants for performance or promotion, and refreshers based on tenure
Promotion-related changes
I typically tie this to a Dec-Jan perf cycle that culminates in compensation adjustments for everyone by mid-Feb.
A 1x per year promotion-only cycle only adjusts levels, salary, and equity for people who were promoted. I typically tie this to a June-July perf cycle, and I adjust compensation for people who were promoted during that cycle by mid-August. (To clarify: everyone is reviewed and considered for promotion 2x per year in both Jan and July, but there are no compensation changes outside of promotions in the July cycle.)
While the dates I’ve shared are common, you can pick your own. I go into more detail about scheduling these large company “moments” in my post about company annual calendars.
Communication: Your comp philosophy document
The last step once you have made all your decisions and implemented your compensation system is to communicate your compensation philosophy to employees and candidates.
Before you share ANYTHING with employees, make sure you prep your managers and answer their own questions. All employee questions will go to managers first, so they need to be able to explain the system.
In the compensation philosophy article, we shared a template compensation philosophy document that you can publish on your wiki – feel free to borrow and customize it. It can be a really helpful resource to maintain the system in an ongoing way.
For current employees that are being transitioned into this new comp system (usually at the end of a performance cycle), design a series of moments when you’ll communicate about it – maybe a waterfall that starts with a message from the CEO, follow-up team meetings led by your directors and managers, and 1:1 individual conversations where managers deliver updated compensation letters to their direct reports. We have templates for compensation letters in our DIY package.
It’s rare to make it through a compensation cycle without getting a bunch of hard questions or having a few people that are upset. Expect that and try to prepare for it. But most importantly, make sure you’ve designed a system that you’re prepared to defend.
When you’re recruiting candidates, the earlier you can communicate about your compensation philosophy and help set their expectations about the compensation associated with their role, the less time you will waste for everyone. Nothing is worse than getting to the late stages of recruiting with a candidate you like and realizing that their expectations on compensation don’t match what you can afford. Talk about it early and often, and you can avoid spending time on candidates that will never join due to misalignment about compensation.
Wahoo!
There are so many more things we could talk about around compensation from how to handle hard conversations to the pros and cons of transparency, etc., but for now, I’m going to put the pen down.
Compensation is definitely an overwhelming thing to implement, which is why a lot of startups use consultants to do it. At a minimum, this post should help you get a sense of whether you have the resources to do it yourself.
But primarily, as I said above, the purpose of this post is to help you see what it can look like to design a compensation system around fairness and consistency. And if you have a system that you believe is designed to be fair, then it gives you the benefit of being able to be clear and direct with employees about how it works, which can let them go back to focusing on building a company that will make the world better in some way. And, hopefully, all that hard work will turn into valuable equity for everyone.
This article is part of a series on compensation for startups. It covers implementing and defending your compensation system. The four parts are:
If you are actively designing and implementing any of these systems, we also have some Do-It-Yourself packages you can buy that include a recording of us explaining all of this in more depth as well as dozens of templates.