Pricing, Packaging & Willingness-to-Pay — and Synthetic Price Research

For factory operators. Pricing is the highest-leverage, least-experimented lever in SaaS — a few points of price beat most funnel optimization, yet pricing studies are slow, expensive, and small-N. This document covers how to measure willingness-to-pay properly, how to structure packaging, and how synthetic-consumer-panels can pre-screen price points so the factory tests the right few against real buyers instead of fielding everything.

The thesis: pricing is three separate decisions, not one — the value metric you charge on, the packaging that structures it, and the price points that hold up. Conflating them ("we use usage-based pricing") describes a licensing model while silently leaving the two higher-leverage decisions unmade.

1. The Three Decisions

  1. Value metric — the unit you charge on (seats, API calls, transactions, workflow runs). The single highest-leverage pricing choice, because it determines whether the customer's bill scales with the value they get.
  2. Packaging — how features are bundled into tiers (Good-Better-Best) and what gates each tier.
  3. Price points — the actual numbers that survive real deals.

Get these in order: metric → packaging → price. A great price point on the wrong value metric still misaligns price and value.

2. Measuring Willingness-to-Pay

Van Westendorp Price Sensitivity Meter (Peter van Westendorp, 1976) — four questions yielding a range, not a point:
1. Too cheap — "so low you'd doubt the quality?"
2. Cheap/bargain — "a great buy for the money?"
3. Expensive — "getting expensive but you'd still consider it?"
4. Too expensive — "so expensive you wouldn't consider it?"

Plotting cumulative curves yields four intersections (verified, Conjointly):
- OPP (Optimal Price Point): "too cheap" × "too expensive" — equal numbers reject on each side.
- IPP (Indifference Price Point): "cheap" × "expensive".
- PMC / PME (range bounds): "too cheap"×"expensive" and "cheap"×"too expensive" — the lower/upper bounds of the reasonable range.

Gabor-Granger — ask purchase intent (yes/no) at discrete price points; the cumulative "would buy" curve estimates the demand curve, price elasticity, and revenue-maximizing price. Use Van Westendorp to find the acceptable range when you don't know it; Gabor-Granger to find the revenue-max point within it. Many teams run both to triangulate. Conjoint adds feature/price trade-offs for packaging design.

3. Packaging

  • Good-Better-Best (GBB): the leading SaaS structure — each tier adds value so the buyer's choice is obvious; good packaging reduces evaluation burden.
  • Per-seat: best for collaboration tools; simple to forecast, but revenue is capped by headcount and under-monetizes small intensive teams.
  • Usage-based: aligns price with consumption and value, but enterprise buyers resist because they can't forecast usage to commit budget.
  • Hybrid (base + variable): the dominant trend — ~43% of SaaS now, projected ~61% by end of 2026 — combining budget predictability with value alignment.

No model is universally best; the right one matches how the customer gets value.

4. Synthetic Price Research (the SSR tie-in)

Real pricing studies are slow, costly, and small-N — exactly the constraint synthetic-consumer-panels address. SSR can run Van Westendorp / Gabor-Granger / conjoint synthetically: score price points and packages against anchor-statements using semantic-similarity-rating to rank options before fielding any to real buyers. The same two non-negotiables from the SSR ingest apply:

  • Comparative use only — synthetic panels predict relative preference (package B > A, price tolerance higher here), not absolute WTP in dollars. Use them to choose what to field, never to set the price.
  • Mandatory calibration-loop — every synthetic verdict that later gets a real outcome feeds back to measure agreement; uncalibrated panels drift (domain-transfer-risk). Pricing is a high-stakes, high-drift domain — calibration is not optional.

The workflow: SSR ranks many price/package variants → field the top 2–3 to a small real panel → decide with the bayesian-decision-rule under minimum-detectable-effect limits.

5. Factory Integration

Pricing research becomes a recurring, mostly-agent-run study: generate price/package variants, pre-screen synthetically, field survivors to a small panel, fit the WTP curves, and surface the OPP/elasticity with confidence bounds — human makes the price call. It links positioning to monetization (message-market-fit and the value proposition feed the value-metric choice) and extends the freemium-subscription-model into deliberate tier design.

Provenance Note

  • Primary-sourced, fetched & verified: the four Van Westendorp questions and the OPP/IPP/PMC/PME intersection definitions (Conjointly, fetched).
  • Secondary — search summary, not independently fetched: Gabor-Granger demand-curve framing; the value-metric/packaging taxonomy; and the hybrid-pricing adoption figures (~43%→~61%) — directionally cited, not promoted to verified.
  • Established method: Van Westendorp (1976); Gabor-Granger; conjoint analysis.

Sources

Concepts

Extracted from this source: willingness-to-pay · pricing-packaging · synthetic-pricing-research

Related concepts: synthetic-consumer-panels · calibration-loop · freemium-subscription-model · message-market-fit · bayesian-decision-rule