Field notes on metric definitions, KPI governance, benchmarking, and data quality — from the team building KPI Compass.
A great generalist — but even with memory turned on, it's personal to one account. Here's what that actually means for a team trying to run analytics on shared, governed data.
Data from companies like yours can be more relevant than a published report — but only past a certain sample size, and only with real privacy guarantees.
AI is genuinely useful for assembling a first-draft measurement framework fast. It's not a substitute for someone who understands your business deciding what actually matters.
A framework for picking KPIs that actually inform decisions, instead of just filling out a dashboard.
The right benchmark for a Series A company and a Series C company aren't the same number — they're not even measuring comparable situations.
The same word — churn, LTV, CAC — means something structurally different depending on whether you're selling a subscription or a product.
Most campaign naming conventions fall apart within a quarter. Here's how to design one that survives contact with a real team.
Signups and activation get most of the attention in PLG dashboards. The metrics that actually predict who sticks around live one layer deeper.
Why the numbers in your dashboard drift from what Google Ads or Meta Ads reports, and a repeatable workflow for catching it before a client does.
You don't need a data team to go from twelve conflicting spreadsheets to one governed set of definitions. You need an afternoon and a process.
Marketing reports one revenue number. Finance reports another. Both are technically correct — which is exactly the problem.
Governance has a reputation for adding process for its own sake. Done right, it's a twenty-minute recurring check, not a quarterly audit nobody looks forward to.
A practical, ship-it-this-week process for turning ad hoc campaign names into a schema your whole team can actually follow.
Every team defining "active user" or "qualified lead" slightly differently doesn't show up as an incident. It shows up as a thousand small disagreements nobody can quite trace.
Dashboards rarely lie on purpose. They drift quietly, one unnoticed mismatch at a time, until the number everyone trusts is wrong.
A benchmark without a visible formula and sample size isn't a benchmark — it's a number someone typed into a slide.
Both go up and to the right in a slide deck. Only one of them should change what you do on Monday morning.
A framework for organizing your KPIs into tiers, so a dashboard doesn't turn into an undifferentiated wall of numbers nobody actually reads.