Pricing decisions usually fall out of a Slack thread and a gut feeling. A memo forces the assumptions to the surface: who the buyer is, what the anchor is for, why the middle tier is where 60-70% of revenue should land, and how low you will go before walking away. AI is good at drafting the memo; you are good at deciding whether to ship it.
TL;DR
- Use AI to structure the pricing memo and name the role of each tier, not to invent willingness-to-pay (WTP) numbers. Feed it your interview data; let it shape the argument.
- Three tiers is the workhorse: roughly 48% of SaaS companies ship three plans, and three-tier pages tend to convert better than two- or four-tier ones because the structure lets you anchor.
- Design all three tiers, not just the hero. A credible decoy/anchor tier can lift mid-tier conversion by 25-60% in published case data.
- Set one discount floor tied to unit economics (keep LTV:CAC above the 3:1 line) and write it into the memo before sales talks you out of it.
- Recommended models (June 2026): Claude Opus 4.7 or GPT-5.5 for the reasoning-heavy first draft; Gemini 3.1 Pro when you are pasting many competitor screenshots and want a long-context read.
The task
You need to set or reset pricing. You have a product, a rough sense of three buyer segments, and competitor screenshots open in another tab. You want a memo your team can fight over before Friday: three tiers with named differentiators (not “more seats”), the role of each tier (anchor, hero, ceiling), a single discount floor you will not cross in sales calls, and one experiment to run in 30 days to test the call.
When this is the right job for AI
AI is strong at structuring the memo, naming the role each tier plays, and pulling pricing-psychology language (“anchor,” “decoy,” “tier compression”) into a working draft. It is weak at telling you what the market will actually pay. That is a function of your buyer interviews, win/loss data, and competitor benchmarks. Hand the model your data; let it shape the argument. Do not let it invent the WTP numbers.
If you want a defensible WTP signal before you draft, run a short Van Westendorp price-sensitivity survey: four questions (too cheap, a bargain, getting expensive, too expensive) across 30-50 buyers per segment gives you a price band, not a guess. Paste the band into the memo and let the model reason from it.
What to feed the AI
- The product and the unit of value you charge for (seats, usage, outcome, hybrid)
- Three buyer segments with one-line descriptions and their current alternative
- Competitor pricing (at least three, with their tier names and price points)
- Your current pricing if you have one, and what you know is broken
- One WTP data point per segment (from interviews or a Van Westendorp band, not guesses)
- Your gross margin so the discount floor is grounded in unit economics
- The decision deadline and who has to approve
Recommended model and a note on cost
As of June 2026, the reasoning quality on this kind of structured-argument task is highest on Claude Opus 4.7 (Pro $20/mo) and GPT-5.5 (ChatGPT Plus $20/mo). Both will hold a seven-section memo together without losing the thread. If your input is mostly pasted screenshots and competitor tables and you want one long read, Gemini 3.1 Pro (Google AI Pro $19.99/mo) ships a 1M-token context window. For a one-off memo any of the three $20 plans is plenty; you do not need a $100-200 tier for this.
Copy-ready prompt
You are drafting a pricing strategy memo for an internal pricing review.
Product: [one line]
Unit of value we charge for: [seats / usage / outcome / hybrid]
Buyer segments (3):
Segment A: [description] | current alternative: [X] | WTP signal: [Y]
Segment B: ...
Segment C: ...
Competitor pricing (at least 3): [name | tiers | prices]
Current pricing (or "none"): [paste]
Known issues with current pricing: [list]
Gross margin: [%]
Decision deadline: [date]
Return a memo with these sections, exactly in this order:
1. The pricing question we are trying to answer (one sentence).
2. Three proposed tiers, each with:
- Tier name (no generic "Pro / Business / Enterprise" unless it fits)
- Headline price + billing unit
- Role: anchor / hero / ceiling
- Two named differentiators that a buyer can verify in 10 seconds
- The segment this tier is built for
3. Anchor logic: why the top tier price is credible and not laughed at.
4. Hero tier logic: why 60-70% of revenue should land here.
5. Discount floor: the single percentage off the hero tier we will not
cross. Justify against gross margin and segment WTP.
6. Three objections the team will raise, with the counter for each.
7. One 30-day experiment to test the riskiest assumption: sample,
measurement, decision rule.
Do not invent WTP numbers. If a segment is missing a WTP signal,
flag it as [needs interview] in the memo.
Tone: argumentative but specific. No "value-based pricing" without
naming the value.For the second pass: “Now write the one-paragraph version a CEO would forward to the board.”
Sample output structure
Question: Should we collapse from five tiers to three and raise the hero tier from $49 to $79?
Tier 1, Maker, $19/seat/mo (anchor): for solo creators, role is to make the hero tier feel obviously priced. Differentiators: 1 workspace, community support.
Tier 2, Studio, $79/seat/mo (hero): for 3-15 person teams. Differentiators: shared workspaces, audit log, priority support. This is where 65% of revenue should land.
Tier 3, Agency, $249/seat/mo (ceiling): for client-services teams. Differentiators: white-label, SSO, dedicated CSM.
Discount floor: 15% off the Studio annual list. Below that we are inside the LTV:CAC danger zone for the segment.
The three roles, and why each tier earns its place
| Tier | Role | What it is for | Common mistake |
|---|---|---|---|
| Anchor (cheapest) | Decoy / reference | Makes the hero look obviously priced; absorbs price-sensitive buyers | Built as the only “real” tier, so it cannibalizes the hero |
| Hero (middle) | Revenue center | The plan you want 60-70% of new revenue on | Under-differentiated, so buyers default down |
| Ceiling (top) | Anchor + expansion | Sets the high reference and houses enterprise asks | Priced as a fantasy nobody would buy, so it reads as a gimmick |
The middle tier doing the heavy lifting is not folklore. The asymmetric-dominance (decoy) effect has produced 25-60% lifts in mid-tier conversion in published pricing case studies, and most pricing teams target 60-70% of customers on the hero plan. The anchor tier only works if at least one real buyer would genuinely choose it; a pure decoy leaks in sales calls.
How to refine
- If tier differentiators are generic: “Replace ‘more storage’ and ‘priority support’ with a named capability a buyer can verify. ‘Audit log’ is verifiable; ‘better support’ is not.”
- If the anchor reads cynical: “The anchor tier needs at least one buyer who would genuinely choose it. If you cannot name one, the anchor is decoy-only and that will leak in sales calls.”
- If the discount floor is missing a justification: “Tie the floor to gross margin or segment WTP. ‘We feel uncomfortable below 20% off’ is not a floor.”
- If objections are softballs: “Surface the real objection from sales (‘we will lose deals at $79’) and write the counter that gives sales a script, not a brush-off.”
Common mistakes
- Letting AI invent willingness-to-pay numbers without source data
- Naming three tiers but designing only the hero. Anchor and ceiling matter equally
- A discount floor with no tie to unit economics. Sales will breach it the first quarter
- “Enterprise: call us” with no anchor price. Buyers self-select out before the call
- Forgetting the experiment. A memo without a 30-day test is a manifesto, not a decision
FAQ
Three tiers or four? Three unless you have data showing a real fourth segment. Around 48% of SaaS companies ship three plans, and three-tier pages generally convert better than four-plus because the choice stays legible and you can still anchor. Four tiers usually means two are doing the same job.
Per-seat or usage-based? Match the unit to the value the buyer feels grow. Per-seat is fine for tools used daily by a defined team; usage works when value scales with throughput. AI-native products increasingly mix the two (a seat floor plus a usage meter), which the memo should name explicitly so sales can quote it.
Where should the discount floor sit? Tie it to unit economics, not comfort. The standard healthy SaaS LTV:CAC line is 3:1 (median B2B is around 3.2:1, as of 2025), so set the floor at the discount that still keeps the hero tier above that ratio. Anything deeper than your annual-plan discount (commonly 15-20%) should require named approval, written into the memo.
How do I know the hero tier is right? 60-70% of new revenue lands there within two quarters. If it is below 40%, the differentiation is unclear: the anchor is too generous or the hero is missing a verifiable capability that justifies the step up.
When should I re-price? Annually, or after any roughly 20% shift in CAC, churn, or the competitive landscape. Re-pricing more often than that erodes trust and trains buyers to wait for the next change.
Related
- Pricing experiment AI — design the 30-day test
- Pricing hypothesis — earlier-stage pricing reasoning
- Pricing page copy — translate the memo into landing-page copy
- Unit economics AI — the numbers your discount floor is anchored to
- Board deck narrative AI — pricing usually shows up as a slide
- AB test summary AI — read out the experiment results