Innovation Accounting and Portfolio Management

A Horizon 1 bet — extending an established core business — can reasonably be judged against revenue, profit margin, and market share, because the underlying business model is proven and the relevant question is one of execution and optimization. A Horizon 3 bet — a genuinely new, exploratory initiative — cannot reasonably be judged against these same metrics in its earliest stages, not because the team is executing poorly, but because the entire premise of an early-stage exploratory bet is that the business model itself has not yet been validated, and demanding revenue-scale proof before that validation has occurred is asking the bet to demonstrate something it is structurally too early to demonstrate. Applying Horizon 1 metrics to a Horizon 3 bet doesn't produce a more rigorous evaluation; it produces a category error that reliably kills promising early bets before they've had a chance to answer the actual questions they were designed to test.

This lesson introduces the Portfolio Health Grid, plotting each bet in a portfolio along two axes: its Three Horizons classification from Lesson 71, and its actual stage of validated progress.

The Grid's core discipline is that the appropriate metric for any given bet depends on its position on both axes simultaneously — a Horizon 3 bet at the Concept stage should be evaluated on whether its core hypothesis has been clearly articulated and an initial test designed, not on revenue; a Horizon 3 bet that has progressed to Pilot stage should be evaluated on whether a real, if limited, customer segment shows the validated behavior the hypothesis predicted, a meaningfully different and more demanding bar than the Concept stage, but still not the same bar as a mature Horizon 1 business. Placing every bet somewhere on this Grid, rather than evaluating all bets against a single organizational-standard metric, is what makes stage-appropriate evaluation possible at all.

Validated learning, a concept from lean startup methodology, refers to evidence that a specific, falsifiable hypothesis about customer behavior or business viability has actually been tested and either confirmed or disconfirmed — directly connecting to the falsifiable Strategic Bet discipline from Lesson 71. Vanity metrics, by contrast, are numbers that look impressive and tend to always increase over time (total signups, cumulative downloads, total page views) without actually testing whether the bet's underlying hypothesis is correct. A Horizon 3 bet can generate an impressive-looking vanity metric — a large number of free trial signups, for instance — while providing no validated learning at all about whether those users would actually pay, retain, or behave in the way the bet's underlying hypothesis predicted. Innovation accounting, done well, insists on validated learning metrics specific to the bet's stated hypothesis, rather than accepting vanity metrics as a substitute simply because they are easier to produce and more comfortable to report.

Metric mismatch produces two distinct, opposite failure modes. Premature cancellation occurs when a genuinely promising early-stage bet is killed because it hasn't yet produced Horizon 1-scale results it was never structurally positioned to produce this early — the specific risk Lesson 71 flagged for Horizon 3 bets judged by near-term metrics. Prolonged survival occurs when a genuinely failing bet continues to receive resources because it generates comfortable-looking vanity metrics that mask the absence of any real validated learning supporting its underlying hypothesis — a bet can look active and growing by activity metrics while its actual, falsifiable hypothesis has already been quietly disconfirmed by the available evidence, with no one having checked because the vanity metrics provided a comfortable alternative narrative.