Lean Analytics
Data & AnalyticsLeadership

Lean Analytics

Benjamin Yoskovitz, Alistair Croll
Read 25 November 2017

Review

Lean Analytics is dense, but it's dense with the right things: a genuinely practical approach to using data to build a startup, not just admire dashboards. Yoskovitz and Croll's real contribution is teaching you which questions to ask of your data at your current stage, not just how to read a metric.

Honestly, this could have been two books. Six business models across five stages over 400-plus pages is a lot, and a fair amount of it won't apply to whatever you're actually building. Read it while you're inside a company where you can test the ideas directly and skip what doesn't fit, rather than cover to cover in one sitting.

Key Takeaways

The parts worth keeping:

Why analytics matters

Entrepreneurs run on relentless optimism, a reality distortion field that pushes them through odds that would stop anyone else. Analytics is the necessary counterweight of realism. Good metrics aren't just numbers you report, they're the ones that change what you do next, and ratios or rates tend to beat raw counts because they're comparative and easier to act on.

The Lean Analytics Cycle

  1. Choose the one metric that matters most right now.
  2. Set a realistic target, using industry benchmarks where you don't have your own baseline yet.
  3. Work out how to move it: study what your best users already do and make that easier, then brainstorm, test, and ship the strongest ideas.
  4. Measure the impact. If it worked, pick the next metric. If it didn't, pivot, try a different approach, or go back to discovery.

One Metric That Matters

Pick the single metric that matters most for your current stage and optimize it with real discipline. You'll usually hit diminishing returns eventually, a local maximum, and that's the signal it's time to move to the next metric rather than a failure. Set explicit target numbers rather than adjusting the target down to match wherever you already are, and know your benchmarks before you panic (or celebrate) about a number: what looks alarming in isolation can turn out to be the best in your industry, and vice versa. Most startups fail, which is a reminder that average performance usually isn't good enough.

Five stages, one gate each

StageThe gate you need to pass
EmpathyFound a real, poorly met need in a market you can reach
StickinessFigured out how to solve it in a way people accept and pay for
ViralityUsers and features fuel growth on their own, organically and artificially
RevenueFound a sustainable, scalable business with the right margins in a healthy market
ScaleCan achieve a successful exit on the right terms

The Lean Canvas, one line each

ProblemWhat real problem do people already know they have?
Customer segmentsWho's the target market, and what messaging reaches them?
Unique value propositionThe one clear, memorable reason you're different or better
SolutionCan you actually solve the problem?
ChannelsHow you deliver the product and get paid
Revenue streamsOne-time or recurring, direct or indirect
Cost structureDirect, variable, and indirect costs
MetricsThe numbers that tell you if you're actually progressing
Unfair advantageThe force multiplier competitors can't easily copy

Common data traps

Sharper individual insights