A market data subscription gives you numbers. It doesn't tell you which numbers matter for your business.
Pick the wrong peer group and you end up chasing a market that isn't yours – benchmarking against companies three funding rounds ahead, or a geography where the talent pool, cost of living and expectations look nothing like your own. The data looks precise. The decisions built on it aren't.
Andreea Lupu and Juan Novoa have both built and defended benchmarking approaches in practice, and they'll bring that experience to the conversation.
This session they covered:
Expect practical frameworks and honest conversation – and plenty of time to raise your own questions or experiences too.
If you’re more of a reader than a watcher, here are a few of the most interesting insights from Andreea and Juan’s discussion on benchmarking beyond market data.
A benchmarking exercise for setting pay ranges is a fundamentally different exercise to benchmarking in response to a specific retention risk.
In the first case, you’re looking at your closest competitors – companies with similar operating models, funding structures, and recruiting pools.
In the second, you might reasonably reach for a narrower sample, because the question you’re answering might be specific to one company poaching your talent.
A data provider gives you numbers, but it can’t tell you which numbers are the right ones for the question in front of you. Defining your compensation philosophy, and being clear on the specific use case for the benchmarks, is the way to combat this.
When a benchmarking presentation goes badly, it can feel like the data was the problem. But in Andreea and Juan’s experience, it’s often the peer group that’s causing misalignment.
If your leadership team sees you competing for talent with one set of companies, and your peer group is built from another, they’ll question the numbers.
It sounds like pushback on the salaries, but when you dig deeper the disagreement is about who you’re benchmarking against.
Raising that for the first time in the room makes it much harder to resolve. Again, this comes back to having a clearly defined compensation philosophy before you reach for the data, and getting alignment from stakeholders on the approach upfront.
💡 Practical application: Before your next benchmarking cycle, agree the criteria that define your peer group (company size, sector, funding stage, geography, or whatever matters most for you) and get sign-off from the right stakeholders before you look at any numbers.
Benchmarking platforms give you a large number of market filters – split the benchmarks by location, by company size, by industry, by revenue stage, and so on.
The temptation is to use all of them – to narrow the sample until it feels like an exact match for your company and your roles.
The problem is that over-filtering can leave you with a sample size too small to be reliable, and small samples can create volatility, depending on the methodology of your chosen benchmarking provider.
If you’re using a compensation survey provider, the data becomes sensitive to which companies happened to participate in the survey that year, rather than reflecting a real market signal.
The key is to apply the filters that actually move the data – typically sector, geography, and company size – and resist the ones that feel precise but don’t significantly change the result. A broader, more stable sample is almost always more useful than a narrow one that looks specific but behaves erratically year on year.
Benchmarking without a compensation philosophy means the data will make the decisions for you. Where do you want to sit in the market? What’s the balance between base pay and variable? Are there roles where you need to be more aggressive than others?
Without answers to those questions, every benchmarking conversation becomes a negotiation from scratch, and different parts of the business will reach different conclusions from the same data.
The philosophy doesn’t need to be perfect before you start, but it does need to exist.
A document that says we target the 50th percentile against this peer group for most roles, with flexibility to 75th for hard-to-fill positions, gives everyone a shared reference point for every conversation that follows.
And remember it’s a live document – it will need updating as the business changes – but something imperfect and written down is more useful than something perfect and still in someone’s head.
💡 Practical application: If you don’t have a documented compensation philosophy, start with four questions: what market are we competing in, where do we want to sit in that market, what do we prioritise between base and variable, and are there roles or functions where we’d flex that position?
Your monthly dose of market insights and expert perspectives
Product Manager salary in 2026: £69,000 on average in the UK, £122,100 at senior manager level, £214,100 for a CPO. Plus pay by country, stage, and hiring trends.
What should you pay at seed vs Series A vs Series C? Startup salary benchmarks for 2026 by funding stage, plus equity, variable pay, and pay strategy advice.
AI Engineer salaries in 2026, how much more AI companies pay (9.7%), and how 27% of companies are paying a premium for AI talent – from Ravio's live data.
