Lore

SMT Model (Strategy, Metric, Tactics)

Overview

SMT pairs each product strategy with exactly one fast-moving proxy metric and a supporting set of tactics. Top-line metrics (e.g., monthly retention) move too slowly to guide week-to-week decisions, so a strategy needs its own leading indicator: for "better watching experience," Netflix used % of customers watching ≥40 hours/month as the proxy, with tactics like ultra-HD and faster localization meant to move it. Proxy metrics are phrased as "% of customers who do at least X," not as an average, so a few outliers can't mask an unmoved majority.

Proxy metric selection criteria

Choosing a proxy metric is a genuine debate — Biddle says these discussions ran 3–9 months at Netflix — tested against four questions:

  1. Findable — can the data actually be captured?
  2. Movable — can the team's tactics plausibly move it in the near term?
  3. Correlated — does it correlate with the true north-star/top-line metric?
  4. Smallest meaningful increment — what is the smallest unit of customer value the metric should register?

Getting this right is what Biddle credits with making a pod's job "no longer subjective."

Organizational execution

Each pod/squad/swim lane owns one SMT triple. See Highly Aligned, Loosely Coupled for how leadership allocates investment share across lanes top-down while leaving tactics to each team, reconciled at a recurring quarterly product-strategy meeting.

Related

An SMT triple operationalizes a single DHM Model (Delight, Hard-to-Copy, Margin-Enhancing) hypothesis; GEM Model (Growth, Engagement, Monetization) sets which force (growth/engagement/monetization) a given SMT triple should prioritize when they conflict.