Pricing tradeoff

Pricing Tradeoff

A pricing question in, structured options out — each one showing how the money actually works. A price without a mechanism is a vibe with a dollar sign.

Verification gates (defined first; output is blocked until all pass)

• G1 — Margin mechanism per option: every option states its mechanism — what the price covers, what absorbs the cost, and the margin consequence derived from input numbers (including the tail, not just the median). A naked price point fails the gate.
• G2 — Input-only figures: every dollar amount traces to the input or is [ESTIMATE: derivation]. No imported benchmarks ("industry attach rate is 20%") presented as fact.
• G3 — Symmetric tradeoffs: each option carries who it selects for/against and its failure mode. A strawman option (one obviously-broken alternative propping up a favorite) fails.

Steps

1. Bank the economics. Price, margin, unit costs — and the cost distribution, not just the average: the heavy-user tail is where pricing structures break. Median $4 and top-decile $15 are different problems; if only an average is provided, flag the missing distribution as the first data gap.
2. Identify the cost driver. What makes one customer cost more than another (usage, seats, storage)? Structures that price the cost driver are stable; structures that absorb it are bets on the distribution — say which each option is.
3. Build 3–4 genuine structures (not 3 numbers on one structure): e.g. included-in-base, flat add-on, usage-based/tiered. Include the stakeholders' asks as options and cost them honestly — "sales wants it free" becomes a real option with its margin give-up quantified and the condition that would justify it.
4. Run each mechanism. Per option: margin math at median AND at the tail, who it attracts/repels (usage caps repel the power users who'd champion the product; included-free selects for upgrade-lift), and the failure mode (negative-margin whales, adoption friction, billing complexity).
5. Recommend only with the deciding tradeoff named — and the data needed to confirm it (usually the real usage distribution). If the input can't support a recommendation, deliver the structured options and the missing-data list instead; that is a complete output.
6. Gate pass. Every option has its mechanism (G1), every figure its provenance (G2), tradeoffs symmetric (G3). Fix and re-run; maximum 2 repair loops, then report the failure instead of the output.

Output format

PRICING QUESTION: AI summary add-on (base $30/seat · 80% GM · LLM cost $4 median, $15 top-decile [input])
OPTION A — included in base (sales' ask)
  Mechanism: base absorbs $4/seat median → GM 80% → ~67%; top-decile users cost half the seat price
  Selects for: adoption, upgrade pressure. Fails if: upgrade lift < margin give-up (needs: lift data)
OPTION B — flat +$10/seat (finance's ask)
  Mechanism: $10 − $4 median = ~$6 contribution/seat; NEGATIVE ~$5 on top-decile; friction at point of value
  Selects for: committed teams. Fails if: attach rate collapses or whales concentrate
OPTION C — tiered: included to N summaries, usage-priced past cap
  Mechanism: prices the cost driver; tail pays for itself. Fails via: billing complexity, cap resentment
RECOMMENDATION: <option + the deciding tradeoff> — confirm with: full usage distribution beyond top-10% figure
GATE CHECK: G1 pass (n/n mechanisms) · G2 pass · G3 pass

Hard rules

1. No naked price points. Every number an option proposes shows what it covers, what absorbs the rest, and the resulting margin — median and tail.
2. Never invent benchmarks, attach rates, or willingness-to-pay figures. Missing load-bearing data is named as the confirm-before-deciding list, not filled in.
3. Stakeholder asks are costed, not dismissed and not adopted — "free to drive upgrades" gets its mechanism and its break-even condition like any other option.
4. The tail is mandatory. Any option whose mechanism only works at the median must say what the top decile does to it.

Limitations

• Mechanisms cover unit economics; full pricing strategy (competitive positioning, packaging psychology, price-testing design) is broader than this skill.
• Willingness-to-pay is not derivable from cost data — options are costed structures; validating what customers accept needs research (research-brief pairs with this).
• Margin math is static: no elasticity modeling, no volume-response curves — flip conditions are stated qualitatively with the data that would quantify them.
• Tax, billing-system, and contract constraints can eliminate options this skill scores as viable.