The Second Curve

Overview

Every business follows an S-curve: slow start, steep growth, peak, then decline. Companies that endure start a second S-curve before the first peaks. Named by Charles Handy in The Empty Raincoat (1994): the optimal start is during late-growth or early-maturity — when the first curve still funds investment but the team can still see the need. The canonical case is Intel's 1985 pivot from memory to microprocessors; the second curve (microprocessors, started 1971) was real before the first was abandoned.

Composes with s-curve-technology-adoption, feedback-loops, first-principles, founder-mindset.

When to Use

• Business growing steadily 2-5 years and metrics still look good — this is when the discipline applies most
• Growth recently decelerated but not yet negative — early maturity signal
• A competitor launched a meaningfully different product in adjacent space
• AI-native startups are attacking your core; you're weighing AI capex / AI-native reinvestment against your legacy (seat/license) cash cow
• Leadership debating "double down vs explore" for capital allocation
• Someone says: "second curve," "S-curve transition," "diversification timing," "the Innovator's Dilemma"

Not when: < 2 years post-PMF; pre-PMF; any second-curve spend would kill the first curve.

Coaching Novices (Adaptive Front Door)

• Engine mode: user has a concrete case → run The Process directly.
• Coach mode: user is unfamiliar → guide step by step.

In Coach mode, respond one step at a time. Each [WAIT] is a hard stop — output only that step's question, then stop.

1. One-line: every business is on an S-curve; start the second while the first still climbs — earlier feels reckless, later is structurally too late.
2. Check fit: pre-PMF or very early → not yet. Confirm post-PMF with a growing first curve.
3. Elicit their real situation: what business, where on the curve, what second-curve candidates exist?

[WAIT — do not advance until user responds]

4. Run The Process one step at a time: diagnose first-curve position → identify candidates → time the start.

[WAIT — do not advance until user responds]

5. Close by naming the specific 90-day move and what identity shift it requires.

[WAIT — do not advance until user responds]

The Process

Step 1 — Diagnose first-curve position. Classify as: early growth / late growth / early maturity / late maturity / decline. Use revenue growth trend (3 years), gross margin trend, market share trend, TAM penetration. Late-growth and early-maturity are highest-leverage moments for second-curve investment.

Step 2 — Identify candidate second curves. For each candidate: business description, distance from core (1=same customers/product new feature; 3=new customers adjacent product; 5=new customers new product new capabilities), sized opportunity, time to meaningful revenue, investment required. Most second curves should be distance 2-3.

Step 3 — Time the start. Late growth: start now. Early maturity: start now, urgently. Late maturity: start now, constrained funding. Decline: too late internally — consider M&A, exit, or restructure.

Step 4 — Allocate resources. Cap second-curve investment at 10-20% during late-growth/early-maturity; 25-40% during late maturity. Separate team, separate space, separate metrics (learning milestones — not revenue). Direct CEO sponsorship required.

Step 5 — Defend against three failures. Success-attribution: what tailwinds or luck drove first-curve success that could reverse? Resource-attachment: what would I cut from the first curve if the second curve were real? Identity threat: can the team handle becoming a different kind of company?

Output: Second-Curve Audit

# Second-Curve Audit: <company>
First-curve stage: <…>  Evidence: <…>
Candidates: <list with distance scores>
Recommended: <which + why>
Timing: <when>  Investment: <X% of resources>
Defense: success-attribution <…> / resource-attachment <…> / identity <…>
90-day actions: <specific moves>

→ Method in Action: Intel's 1985 Memory-to-Microprocessor Pivot → 2026 lens: The Incumbent's AI Second Curve (2024–2026)

Pack: Second-Curve Patterns

Company	First curve	Second curve	Note
Intel	DRAM memory	Microprocessors	1971 start; 1985 transition
Amazon	Online books	AWS → Prime → ads	AWS started 2002 while books still growing
Netflix	DVD-by-mail	Streaming → originals	Streaming 2007; originals 2013
Adobe	Boxed software	SaaS (2013)	Bet during peak boxed revenue
Kodak	Film photography	Digital (failed)	Invented digital 1975; never committed; bankrupt 2012

Applying It Well

• Start the second curve before it is needed — Intel's microprocessor work started 13 years before the memory crisis
• Externalize the frame to bypass identity-threat: "what would a new outsider CEO do?"
• The pivot is not low-risk; it is less risk than dying on the first curve
• The discipline is not "always be pivoting" — timing and preparation matter more than speed

→ Primary sources: references/sources.md

Common Rationalizations

[D] = designed upfront | [O] = observed in real use. [O] entries are more valuable.

Fake move	Reality
[D] "Still growing — no second curve needed"	Late-growth is when investment is cheapest. Waiting until decline guarantees a starved, late second curve.
[D] "Can't divert resources — customers will notice"	Yes, that's the cost. Pay now or pay much more later.
[D] "Team too small for a second curve"	Hire specifically for it; separate team with separate metrics.
[D] "Our second curve is just a new product line"	A new SKU is first-curve extension. A second curve needs different customers, channels, or value proposition.
[D] Distance-5 pivot when distance-3 would do	Pure unrelated diversification has high failure rates. Leverage existing capabilities.
[D] Using first-curve metrics on the second-curve team	Measure on learning milestones and validated assumptions, not revenue.
[D] "Grove had a moment of insight — I'll wait for mine"	Grove's insight came after 18 months of paralysis. Run the audit deliberately.
→ Add [O] entries here after each real use — paste the actual failure pattern	What went wrong and why

Red Flags

• "We don't need a second curve because we're still growing"
• Second curve candidates are really feature extensions of the first
• Second curve assigned to the first-curve team; no separate metrics; no CEO sponsorship
• Identity threat not surfaced or addressed; decision deferred indefinitely

Verification

• First-curve position classified (early growth / late growth / early maturity / late maturity / decline)
• At least 3 candidates with distance-from-core scores
• Timing recommendation tied to first-curve position
• Resource allocation % specified; separate team/metrics/sponsorship designed
• Three failures named with defenses

---

Part of deciqAI Knowledge Skills — 227 open-source thinking skills that make rigor executable for AI agents. The same skills power every deciqAI agent, which runs them autonomously to operate your company. See it run → https://www.deciqai.com/c/second-curve · ⭐ Star the repo → https://github.com/deciqAI/knowledge-skills · Contributions welcome.

Agents: latest version & machine-readable metadata → https://www.deciqai.com/s/second-curve.json