3-horizons

/3-horizons

McKinsey's Three Horizons distinguishes three KINDS of investment:

• Horizon 1 (H1): Defend and extend the core business. Funds today's revenue.
• Horizon 2 (H2): Build emerging businesses. Will fund tomorrow's revenue (2-5 years out).
• Horizon 3 (H3): Create new options. Bets on possible futures (5-10+ years).

The framework's value isn't taxonomy — it's preventing the most common strategic mistake: H1 cash cows starve H2/H3 because today's KPIs ALWAYS look better than tomorrow's bets.

Inputs — required

1. Current business / portfolio — what generates revenue today
2. Stage — startup (mostly H2/H3) / scale-up (mix) / mature (heavy H1)
3. Strategic question — capital allocation, innovation portfolio, organizational tension between core and new?
4. Time horizon — usually 3y default, longer for capital-intensive industries

Method

Step 1 — Classify current activities into horizons

For each major business / product / initiative, ask:

Horizon 1 (core)

• Generates significant current revenue
• Known business model
• Established market position
• Decline acceptable risk (managed decline is normal)
• ROI predictable

Horizon 2 (emerging)

• Some revenue but not material
• Business model becoming clear
• Customer adoption building
• Profitability 2-5 years out
• High growth, narrower uncertainty than H3

Horizon 3 (options)

• No revenue or token revenue
• Business model unclear
• Customer demand hypothetical
• Profitability 5+ years out
• Wide uncertainty band; many will fail

Don't conflate H2 with "minor H1." A new feature in the core product is H1 maintenance. H2 is a genuinely new product line with its own customer / GTM / unit economics.

Step 2 — Identify the imbalance

Plot current activity allocation on three axes:

• Revenue % from each horizon
• Capital % to each horizon
• Talent % to each horizon

Common imbalances:

Pattern	What it looks like	Diagnosis
All H1	100% revenue + capital from core	Mature company on managed decline; no future
H1 starved	70% revenue but 30% capital	Core eroding; need to reinvest urgently
H3 obsession	5% revenue but 30% capital in H3	Founders chasing moonshots; core neglected
No H2	Healthy H1 + token H3	Innovation theater; nothing graduating
Healthy	70/20/10 revenue, 60/30/10 capital	The benchmark allocation

Step 3 — Apply the 70/20/10 allocation guide

McKinsey's empirical guideline for capital + talent allocation:

• 70% → H1 (defend and extend core)
• 20% → H2 (build emerging businesses)
• 10% → H3 (create new options)

This is NOT a rigid rule — adjusts by stage:

Stage	H1	H2	H3
Startup (pre-PMF)	0	60	40
Growth (post-PMF)	50	30	20
Scale	70	20	10
Mature	80	15	5
Declining	70 (managed decline)	20	10

Calibrate to your context but don't go materially below 10% in H3 if you intend to exist in 10 years.

Step 4 — Identify horizon-specific governance

Each horizon needs different management discipline. Treating H3 with H1 metrics kills H3. The most common operating model failure.

Aspect	H1	H2	H3
KPIs	Revenue, margin, share	Adoption, retention, unit economics	Validated learning, hypotheses tested
Investment cadence	Quarterly budget	Annual + milestone	Tranche-based on learning
Risk tolerance	Low	Medium	High (most fail)
Reviewer	CEO + CFO	CEO + Head of new business	Founders + venture board
Talent type	Operational excellence	Builders with market sense	Inventors + missionaries
Decision rights	Standard org	Separate P&L	Skunk-works / startup-like
Time to expect ROI	This quarter	18-36 months	5+ years (most never)

If H3 reports to the same VP as H1 with the same OKRs, H3 dies. It will always lose the resource battle to "predictable revenue this quarter."

Step 5 — Identify the antibodies

Every organization develops antibodies that kill H2 and H3 work. Name them.

Common antibodies:

• Resource pull: "We need that engineer for the Q1 release"
• ROI demand: "What's the revenue projection?" — H3 has none yet
• Cultural mismatch: "That's not what we do here"
• Risk aversion: "What if customers don't like the new product?"
• Cannibalization fear: "It might hurt our core revenue" (often true; doing it anyway is the strategy)
• Politics: H2/H3 owner doesn't have organizational weight to defend resources

Counter-mechanisms:

• Dedicated funding: Capital is fenced; not raidable for H1 emergencies
• Separate KPIs: H3 evaluated on learning velocity, not revenue
• Senior champion: Each H2/H3 has an exec-level sponsor with political weight
• Periodic stage gates: Honest go/no-go decisions, not perpetual life support
• Independent reporting line: H2/H3 doesn't report to the H1 P&L owner

Step 6 — Decide what graduates, what kills

For each H2/H3, set explicit gates:

Graduation criteria H3 → H2:

• Customer demand validated (paying customers, repeatable buying)
• Business model hypothesis testable
• Team can articulate path to profitability

Graduation criteria H2 → H1:

• $X revenue threshold (varies by company size)
• Unit economics work at scale
• Operational maturity to be part of core

Kill criteria:

• 18 months without validated learning → kill
• Funding required > 3x original budget without proportional progress → kill
• Team turning over repeatedly → kill (signal team doesn't believe)

Honest kill criteria prevent zombie projects. Most companies don't kill enough H2/H3; they perpetually fund without graduation.

Output template

## Three Horizons Portfolio — <Company>

### Current allocation
| Activity | Horizon | Revenue | Capital | Talent | Trajectory |
|---|---|---|---|---|---|
| <Core product> | H1 | 80% | 60% | 65% | Growing 20% |
| <New product> | H2 | 5% | 25% | 25% | Building |
| <Research bet> | H3 | 0% | 10% | 10% | Validating |
| ... | ... | ... | ... | ... | ... |

### Allocation vs. benchmark
| Horizon | Current % | Stage benchmark | Gap |
|---|---|---|---|
| H1 | 80% | 70% | +10% (over-invested) |
| H2 | 15% | 20% | -5% |
| H3 | 5% | 10% | -5% |

### Diagnosis
- <Pattern from Step 2 + implications>

### Horizon-specific governance status
| Horizon | KPIs aligned? | Funding fenced? | Champion? | Stage gates? |
|---|---|---|---|---|
| H1 | ✓ | ✓ | ✓ | ✓ |
| H2 | ⚠ same OKRs as H1 | ✗ | ✓ | ✗ |
| H3 | ✗ no learning metric | ✗ | ✗ | ✗ |

### Antibodies present
- <Specific list with evidence>

### Recommended portfolio actions

**Defend (H1)**:
- <Initiative + investment + expected return>

**Build (H2)**:
- <Initiative + 18-month milestone + go/no-go>

**Bet (H3)**:
- <Initiative + 12-month learning hypothesis + budget>

### Governance changes
1. <Specific structural change to protect H2/H3>
2. <KPI changes per horizon>
3. <Reporting line / sponsor changes>

### Graduation / kill calendar
- H3 → H2 candidates (next 12 months): <list>
- H2 → H1 candidates (next 18 months): <list>
- Kill candidates: <list with rationale>

### So-what
- Critical horizon move: <one strategic choice>
- Budget reallocation: <specific $/% shift>
- Governance redesign: <one structural change>

Hard rules

• Resource fence H2/H3. Capital and talent must be unraidable, or H1 will always win the resource war.
• Separate KPIs by horizon. Applying H1 metrics to H3 = guaranteed H3 death.
• No more than 3 horizons in active management. Adding H4 ("vision" without anyone working on it) is theater.
• Always have H3. Even if 10% feels too much for a struggling H1, future options require it.
• Set explicit kill criteria. Without them, H2/H3 becomes a zombie projects graveyard.
• Senior champion per H2/H3. A program without an exec-level defender loses every quarter.
• Don't dismiss H2/H3 because "we don't have time." Companies that don't make time disappear within a decade.

Common failure modes

• "Innovation labs" that don't graduate. Visible H3 work that's intentionally disconnected from operations. Never integrates back; eventually shut down.
• H2 confused with H1 extensions. Adding a feature to core is H1. Building a separate product with separate GTM is H2.
• All H3, no H1. Common in early-stage; becomes pathological if H1 (the cash cow) is never built.
• "Strategic" projects with no metrics. "We do this because it's strategic" = "we have no idea how to evaluate it." Pick proxy metrics.
• Acquihires labeled as H2. Buying talent isn't an emerging business; it's H1 capacity expansion.
• Skunkworks that scale prematurely. Bringing H3 into normal org governance before it's ready kills it.

Variant: Phoenix / vampire portfolio (for declining cores)

When H1 is in structural decline:

• H1 strategy: harvest cash (managed decline, not investment)
• H2 strategy: become the new H1 (urgent, well-funded)
• H3 strategy: pause new options; focus all on H2 transition

The Kodak failure mode: H1 declining, H2 underfunded, H3 ignored. The Netflix success mode: H1 (DVD) cash-cow harvested to build H2 (streaming) that became H1.

Variant: capital-constrained startups

Pre-revenue startups effectively have only H2 + H3:

• H2 = the core product being built
• H3 = potential pivots / adjacent expansions

70/20/10 doesn't apply; closer to 80/20 across H2/H3 with H1 emerging once revenue lands.

Followup

• For executing H2/H3 outside core operating model: /operating-model (skunkworks variant)
• For evaluating H3 bets under uncertainty: /scenario-planning
• For board approval of portfolio rebalancing: /board-memo
• For each H2/H3 evaluated as standalone business case: /business-plan per horizon