The layer that makes every Mark possible
Benchmarks
Every Mark is read against a benchmark — a working range curated for your stage, your model, and your category. That's what separates fine from actually working.
A benchmark isn't a single number. It's a working range — and where your number lands inside it is what becomes a Mark.
The numbers, briefly.
A snapshot across B2B SaaS and D2C. Medians, not averages — because averages flatter the unhealthy.
Working ranges drawn from Benchmarkit, FirstPageSage, HubSpot, Databox, and ProfitWell, sharpened against the Marks our engines generate.
Every Mark is read against a benchmark.
When you run an engine, your numbers don't float in space. They're compared against a healthy range. That's what lets an engine separate fine from actually working.
Benchmarks here aren't industry averages from a research firm. They're working ranges we've validated against real engines, real Marks, and real marketing teams — and we sharpen them as more Marks are generated.
A read across the working ranges.
Nine sample bands across SaaS and D2C. The marker shows a healthy median; your number, when an engine runs, lands wherever it lands.
Geography moves the numbers.
Search CPCs swing wide across regions. A benchmark valid in New York can be misleading in Munich, and irrelevant in Mumbai.
What's coming on this page.
The benchmark library is in active development. The working set behind it already spans every live engine — acquisition, revenue, conversion, retention, audience and brand — routed by country, stage and sales motion. When the library ships, you'll be able to browse the working range for any metric an engine reads, compare across stages and categories, and see the cohorts behind each number. We'll publish the methodology alongside, so the numbers carry their own audit trail.
For now, every live engine reads its benchmarks inline as part of the Mark — so you never have to leave the engine to know what "healthy" means for your stage.