TheGP’s Guide to Building an Early-Stage Compensation Philosophy

One of the questions TheGP’s talent team gets asked most often by founders is: How should we think about compensation?

And it’s usually quickly followed by: How do we build a compensation philosophy?

So we put together a guide to answer both, specifically focusing on pre-product-market fit companies compensating employees with base salary plus stock options.

Note: This framework can eventually be adapted to RSUs and later-stage compensation structures.

Step 1: Start with market data

The first step in this process is to get access to a credible compensation dataset, from places like Pave, Carta Total Comp, OpenComp, and Radford.

Once you have that data, compare it to companies that 1) resemble your startup and 2) are where you’d like to poach talent from. If there’s a large gap in company stage or product maturity, use that delta to inform your comp and recruiting strategy (especially your pitch and value prop).

At a minimum, look at comparable companies in terms of:

• Location

• Capital raised

• Company size/headcount

• Post-money valuation

• Stage

There’s some judgment involved here. Play with those variables until you find a peer set that feels representative of the market you’re actually competing in.

From there, break compensation into percentile bands: 25th, 50th, 75th, and 90th percentile, for both cash and equity. Here’s a template you can use.

For a seed or pre-seed company, keep the initial framework simple. You probably aren’t paying meaningful annual bonuses yet, so instead of modeling total cash compensation, focus on base salary + a new-hire equity grant.

For equity, express the grant as basis points of the company, rather than simply the number of options.

Note: market data isn’t perfect.

It’s worth noting here that compensation data is useful, but also inherently backward looking (the market moves faster than surveys). Importantly, what that means is that the compensation required to actually close a highly qualified candidate may be above what the data suggests. As a working assumption, I’d expect particularly competitive candidates to command a 10–15% premium over your 75th-percentile benchmark, especially in highly competitive fields like applied AI engineering in the current market.

It’s worth noting that real-time candidate behavior can also become compensation data. As you develop your compensation philosophy, we recommend starting a live database to gather intel on competing offers candidates receive, existing compensation numbers that candidates volunteer, and candidates’ “target compensation” or “comp expectations” for the given role.

Step 2: Triangulate for emerging roles that don’t map cleanly to compensation surveys

Not every role will have enough market data to produce a clean benchmark. When there isn’t enough reliable data for the role itself, you can triangulate compensation using adjacent jobs.

Take design engineers, for example. To figure out the right comp for a design engineer, you might look at both product design and frontend/software engineering compensation and determine where the role should sit between them.

We’re seeing this become increasingly important as the boundaries between engineering, product, and design continue to blur. Tools like Codex and Claude are making it possible for folks to operate beyond the traditional boundaries of their disciplines, and over time, we expect engineering, product, and design to converge, which will potentially affect how we compensate those folks.

But for now, it’s already changing what makes exceptional EPD talent stand out.

Increasingly, one marker of top talent is independent shipping ability. Can this person take an ambiguous product or technical problem, get to an MVP, put something in front of users or generate meaningful user feedback, and iterate without being completely dependent on adjacent disciplines?

That should factor into how you think about the upper end of your compensation bands.

Step 3: Establish a “most of the time” compensation philosophy

One of the most important things a founder can do while hiring is establish their compensation philosophy before they’re negotiating with a candidate.

We think of this as a “most of the time” rule.

Historically, earlier-stage companies have been somewhat more conservative on cash and more generous on equity. Exactly where you land is ultimately a founder decision, but if you’re trying to compete for top talent in markets like San Francisco and New York, be very cautious about consistently paying below the 75th percentile on base salary.

From there, decide where you want to sit on equity.

A reasonable philosophy might be: Most of the time, we target the 75th percentile for base salary and offer equity at the upper end of the market.

The exact percentile matters less than having an intentional philosophy that can be applied consistently.

Step 4: Create rules for when you’re willing to break the rules

A compensation philosophy shouldn’t prevent you from making exceptional hires.

Because there will be candidates where you need to ask a different question: Can I afford not to hire this person?

In other words—if this candidate walks away, does an important business line, product, or company objective materially suffer? If the answer is yes, you should have a predefined mechanism for breaking your normal compensation rules.

For example:

• Default hire: Target roughly 75th-percentile base salary and your standard equity band.

• Critical hire: Be willing to move toward the 90th percentile on equity

• “Can’t afford to lose” hire: Be willing to move toward the 90th percentile on both cash and equity, with the understanding that truly competitive candidates may still require a premium above the published benchmark.

• Note! One easy hack here is also using a sign-on bonus to bridge the gap between cash needs. If a candidate is incredibly expensive, make it a two-year sign-on, split evenly across those years, with an explicit clawback clause.

Remember that the specific numbers can of course change depending on your philosophy. What’s important is creating the escalation framework before you’re emotionally invested in closing someone.

Speaking of your philosophy, you should also create rules around how exceptions affect existing employees. A company that repeatedly stretches for new hires without adjusting incumbents can create pay compression and internal-equity problems. In other words, if you bring somebody in widely above your comp band, the news will eventually spread. Assume that employees will speak to each other about their compensation. To accommodate for that, get into the habit of auditing external comp bands and internal salaries yearly (based on performance). It would be awful to lose a top engineer because they’re comped $25K under market

Step 5: Leveling has to sit underneath the framework

Finally, none of this works without leveling.

Compensation bands need to correspond to clearly defined levels so that you’re comparing like with like and maintaining consistency across the organization.

Engineering organizations often use frameworks that begin around an L3-equivalent level for early-career engineers and progress through increasingly senior individual-contributor and management levels. The exact nomenclature varies significantly company to company, so founders shouldn’t become overly attached to the level labels themselves.

If it’s helpful, Levels.fyi is a common tool used by candidates. The tool relies on self reported comp data, which likely skews the data set a bit. Keep in mind that candidates will likely come with biased comp expectations and should be course corrected quickly. As a general rule of thumb, you should explicitly ask, “What are your compensation expectations for this specific role?” Another way to think about this is “what is your take home cash minimum?” Some candidates may be hesitant to answer this question, and in states like California, employers are legally required to quote a cash (but not total) compensation band. In cases like this, quote up to your standard 75th percentile band and ask a candidate, “How does that compare with your needs and expectations.?” Never explicitly ask about current compensation. This is illegal and could get you in murky waters.

Remember, what really matters here is defining what each level means at your company and then mapping compensation bands against those expectations.

The ultimate goal is a system where a founder can answer three questions quickly:

1. What level is this person?

2. What does the market pay for that level and role?

3. How strategically important is this particular hire?

That gives you a compensation system that is grounded in market data without becoming beholden to it.

Josh Hernández is Head of Talent at TheGP. Previously, he was the first recruiter at Robinhood where he helped scale the company from 30 to 4,000 people.