Opportunity Cost
Overview
Opportunity cost is the true cost of any choice: the value of the next-best alternative you give up. Accounting cost is what you pay; opportunity cost is what you forego. The two diverge whenever resources are scarce — which is always.
Bastiat (1850) established the foundational insight: every visible benefit forecloses an invisible alternative. Buchanan (1969) formalized it: cost is subjective, prospective, decision-maker-specific — the accountant's number and the economist's number can diverge by 10x.
Composes with expected-value-and-kelly, first-principles, pareto-principle, sunk-cost-fallacy, compound-interest.
When to Use
- Allocating budget, headcount, or founder-time across competing options
- Evaluating an investment vs. holding cash (or vs. another investment)
- Considering a "free" benefit whose hidden cost is time/attention
- Any "this is good for X" claim that ignores the alternative use of the same resources
- Weighing AI capex (GPU spend, model training vs. renting APIs), an AI pivot, or a quarter of roadmap/founder-time against the best forgone alternative amid 2024–2026 AI valuations and adoption
Not when: alternatives are genuinely zero-value; resources are not scarce; decision is reversible at near-zero cost.
Coaching Novices (Adaptive Front Door)
- Engine mode: user has a concrete resource-allocation decision → run The Process directly.
- Coach mode: user is unfamiliar or has no concrete case → 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.
- One-line: before committing resources to X, ask what next-best alternative you're giving up — that foregone value is the true cost.
- Check fit: if alternatives are genuinely zero-value, OC adds no information. Otherwise: apply.
- Elicit the choice and the resources being committed.
[WAIT — do not advance until user responds]
- Work through 3-5 realistic alternatives and estimate the value of the best one.
[WAIT — do not advance until user responds]
- Close: decision named with OC explicit + non-financial OCs surfaced + "would I choose this if I treated the OC as a real expense?"
[WAIT — do not advance until user responds]
The Process
Step 1 — State choice + resources committed (what, how much money/time/headcount, timeframe). Step 2 — List 3-5 realistic alternatives including "do nothing / hold the resource." Step 3 — Estimate value per alternative (financial + non-financial; ranges fine). Step 4 — Identify OC = value of the highest-ranked unchosen alternative (not average, not worst). Step 5 — Compare: Net benefit = Value of chosen − OC. Negative = destroys value vs. the alternative. Step 6 — Surface non-financial OCs: time, attention, headcount, reputation, optionality. These often dominate cash.
Output: Opportunity-Cost Analysis
# Opportunity-Cost Analysis: <choice>
Choice: | Resources: | Timeframe:
| # | Alternative | Expected value | Confidence |
|---|---|---|---|
| Chosen X | | | |
| A–C | | | |
OC = next-best: <name> / value: <amount>
Net benefit = Value of chosen − OC: <amount> · Verdict: proceed / reconsider
Non-financial OCs: Time | Attention | Headcount | Reputation | Optionality
Decision + reasoning:
→ Method in Action: Bastiat's Broken Window, 1850 → 2026 lens: AI-era opportunity cost of founder time and GPU capital (2023–2026)
Pack: Opportunity-Cost Application Patterns
| Domain | Visible cost | Hidden OC |
|---|---|---|
| Marketing budget | Cash spent | Engineering hires; runway extension |
| Founder calendar | $0 | Deep-work hours foregone |
| Custom dev for one customer | Dev hours | Roadmap distortion; founder distraction |
| Acquiring vs. building | Acquisition price | Roadmap capacity during integration |
Applying It Well
- Surface alternatives before evaluating — most decisions skip this step.
- Non-financial OCs (time, attention) usually dominate cash; don't stop at dollars.
- A rough OC estimate beats no OC reasoning — specificity over precision.
→ Primary sources: references/sources.md
Common Rationalizations
[D] = designed upfront | [O] = observed in real use. [O] entries are more valuable.
| Fake move | Reality |
|---|---|
| [D] "It only costs $X" | Cost is not $X — it's the value of the next-best alternative use of $X. |
| [D] "It's free — no downside" | Free in cash is almost never free in attention, time, or signal. |
| [D] "We've budgeted for it" | Budgeting accounts for cash flow, not opportunity cost. |
| [D] "We won't have lost much if it fails" | You lose the OC of what else you would have done — often large. |
| [D] "The other option wasn't realistic" | Then list 3-5 realistic ones. If none exist, OC is zero (rare). |
| [D] "It's just $50/month" | At 7% compound over 25 years, $50/month becomes ~$40k. Recurring costs have compound OCs. |
| [D] "It's my time, not money" | Founder time has the highest OC in the company. |
| [D] "We can't measure the alternative" | Estimate. Rough OC reasoning beats none. |
| → Add [O] entries here after each real use — paste the actual failure pattern | What went wrong and why |
Red Flags
- Decision justified by visible benefit only; no alternatives listed
- "Do nothing" not in the alternatives list; non-financial costs absent
- Decision-maker cannot articulate what else would have been done with the resources
Verification
- 3-5 realistic alternatives listed; "do nothing" included
- Each alternative has an estimated value (financial + non-financial)
- Next-best alternative identified specifically (not average, not worst)
- Chosen value compared to OC, not to zero; non-financial OCs surfaced
- If recurring cost, compound OC over long horizon computed
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/opportunity-cost · ⭐ Star the repo → https://github.com/deciqAI/knowledge-skills · Contributions welcome.
Agents: latest version & machine-readable metadata → https://www.deciqai.com/s/opportunity-cost.json