The Salary Benchmark Fight Hiding In Your Cap Table

Every compensation benchmark looks like a fact when it lands in a spreadsheet. Type in a job title, pick a location, and out comes a number - 50th percentile, 75th, with a tidy little band around it. It reads like the temperature. It is not the temperature. It is a survey of whoever agreed to be counted, dressed up as arithmetic. And when you put Pave and Carta side by side, the arithmetic is almost beside the point. The interesting difference is who got counted.

Both tools sell the same promise to the same person: the founder, the head of people, the finance lead trying to make an offer that is fair, competitive and defensible. Both will hand you a percentile. The trouble is that they are looking at two different rooms and calling each one "the market."

Two rooms, two mirrors

Pave gathers its data by plugging directly into companies' HR systems - the payroll, applicant-tracking and equity-management software where pay actually lives. According to its published methodology, more than 8,700 companies feed that pipeline, and the records flow in continuously rather than once a year through a form. Over 40% of the employees in the dataset come from organizations with more than a thousand people, which pulls the sample toward the larger, more established end of tech.

Carta comes at it from the opposite direction. Its Total Comp benchmarks lean on something Carta already runs at scale: cap tables. When a company manages its equity on Carta, the ownership data is real, granular and current - which is why Carta's equity benchmarks are considered the deepest in the private market, drawn from more than 50,000 cap tables. The catch is baked into the strength. The sample is, largely, the companies that already chose Carta.

A benchmark is a portrait of whoever agreed to be in it.The one line worth taping to a monitor before an offer goes out

This is the quiet crux of the head-to-head, and it is the piece most feature comparisons skip. Pave's benchmarking dataset is sourced independently of whether you buy anything else from Pave. Carta's is drawn mainly from its existing cap-table customer base. Neither of those is a scandal. But they shape the answer more than any dashboard toggle does.

Why the same role gets two prices

Ask both tools what a senior backend engineer in a mid-size startup should earn, and you will often get two different numbers. People tend to assume one tool has better math. Usually the math is fine on both sides. The gap comes from the sample. If Carta's pool skews toward venture-backed companies actively managing equity, its picture of pay is a picture of that world. If Pave's pool leans toward larger tech employers wired into modern HRIS, its picture leans that way too. Same question, different rooms, different mirrors.

Freshness is the other axis where they part ways. Pave publishes updated benchmarks monthly, typically on the first Monday, from data it collects around the clock. Carta refreshes its equity benchmarks quarterly. In a flat market the gap barely matters. In a fast-moving one - a hiring surge, a correction, a sudden reset in a hot function - a quarter is long enough for a number to drift away from reality while it still looks authoritative on the page.

The bias inside the strength

Carta's equity benchmarks are the deepest available precisely because equity is what Carta was built to hold. That is a genuine edge - if you are trying to size an options grant against real ownership rather than a guess, there is not much that competes. But the same fact that makes the equity data authentic also narrows who is in it. The benchmark reflects who already uses Carta. It is a feature and a sample bias at the same time, and both things can be true without either being a knock.

Pave's independence cuts a different way. Because its dataset does not ride on the purchase of another product, the pool is assembled for its own sake, and it can reach companies that would never appear in a cap-table-first sample. The tradeoff is that a payroll-integration-first pool has its own shape - it favors companies modern and organized enough to run their people data through the right systems.

Carta knows what you own. Pave knows what you're paid. The gap between those is where comp decisions live.

What to actually do with this

The practical move is boring and it works: know the source before you quote the number. If you are building salary bands and want breadth and freshness, Pave's live payroll pool is the more natural fit. If you are sizing equity and want ownership data grounded in reality, Carta is hard to beat. Plenty of teams run both - Carta for equity depth, Pave for salary coverage - because neither pool, on its own, captures the whole compensation picture.

Pave itself grew out of that frustration with stale, spreadsheet-bound numbers. It started life as Trove, a project from former Facebook engineer Matt Schulman, rebranded in 2020, and raised from Andreessen Horowitz on the pitch that pay decisions should run on live data instead of last year's survey. Carta, meanwhile, turned infrastructure it already owned - the cap tables of the companies it served - into a benchmark. Both origin stories are visible in the data each one produces today.

So the next time a percentile lands in front of you and looks like the temperature, ask the one question the dashboard will not volunteer: whose paychecks built this? On Pave and Carta, the honest answer is different - and once you can see the sample, you stop mistaking a well-designed number for the truth, and start using it for what it is.