Pricing is the highest-leverage decision most teams make once and forget. These 15 prompts force you to write the hypothesis behind each price — who pays, what they compare against, who downgrades if you raise it — so you can falsify it cheaply instead of shipping a guess. They cover free vs freemium vs trial, 2-tier vs 3-tier vs custom, usage-based vs seat-based, anchor and decoy patterns, packaging vs pricing, and the most-skipped exercise: writing why a tier exists at all.
Run them in any current chat model — GPT-5.5, Claude Opus 4.7, or Gemini 3.1 Pro (all as of June 2026). For pricing work, paste real numbers (your competitors’ tiers, your churn, your ACV) and use a thinking/reasoning mode so the model shows its trade-off math instead of a generic checklist.
TL;DR
- AI is a sparring partner for pricing logic, not a substitute for talking to 10 customers. It is fastest at structuring tiers, drafting interview scripts, and running pre-mortems.
- The four-tier decision order: model (free/freemium/trial) → tier count → metric (seat vs usage) → psychology (anchor/decoy). Get model wrong and everything downstream fights uphill.
- 2026 benchmarks to anchor your hypothesis: opt-in free trial converts ~14% median, credit-card (opt-out) trial ~44%, freemium ~4.5% (ChartMogul, 2026). Don’t promise yourself numbers your model can’t support.
- Always pair a price hypothesis with the metric that would kill it (e.g., monthly churn over 8% on the top tier).
Who this is for
Founders pricing a v1, PMs leading a pricing rework, growth leads testing packaging changes, and operators preparing to raise prices on existing customers.
When not to use these prompts
Skip them for one-time consumer purchases (single-SKU e-commerce) — those need different anchor and decoy framing. Skip them too if you have fewer than 50 paying customers — at that stage, direct conversations beat any model’s guess.
2026 pricing-model benchmarks
Feed these into the prompts so the model reasons against real conversion rates instead of inventing them. Figures are medians from ChartMogul’s 2026 study of ~200 SaaS products and widely cited tier-psychology research; treat them as starting anchors, not guarantees.
| Model | Median free-to-paid | Best fit | Main risk |
|---|---|---|---|
| Freemium | ~4.5% (top quartile 8%+) | High-frequency, network-effect, content products | Free tier with no upgrade trigger just burns money |
| Opt-in free trial (no card) | ~14% | Self-serve with fast time-to-value | Tire-kickers; needs in-app activation nudges |
| Opt-out trial (card required) | ~44% | Higher-ACV, lower-volume buyers | Smaller top of funnel; sign-up friction |
| Reverse trial | Varies | Products that prove value fast, then downgrade | Confusing if the free fallback is too weak |
About 57% of SaaS products lead with a free trial versus 26% with freemium (ChartMogul, 2026). On three-tier pages, roughly 60–70% of buyers pick the middle option (the compromise effect), and good-better-best structures lift average order value 15–25%. Price your middle tier at your target margin.
The decision order
These prompts work best run in sequence, because each answer constrains the next:
- Model — free / freemium / trial / paid only (prompts 1, 6).
- Tier count — 2 vs 3 vs custom (prompts 2, 3, 9).
- Metric — seat vs usage vs hybrid (prompts 4, 11).
- Psychology — anchor, decoy, packaging (prompts 5, 10, 13).
- Change management — increases, grandfathering, tests (prompts 7, 8, 14).
- Copy — the public pricing page (prompts 12, 15).
15 copy-ready prompt templates
Placeholders in [brackets] are yours to fill. Each prompt asks the model to commit to one recommendation plus the data that would change it — that “what would falsify this” clause is what turns a guess into a testable hypothesis.
1. Free vs freemium vs trial decider
The first decision. Get this wrong and the rest of pricing fights uphill.
You are a SaaS pricing strategist. For [product], recommend one of:
free / freemium / 14-day trial / 30-day trial / reverse trial / paid only.
Output: the recommendation, 3 reasons it fits, the 1 biggest risk, and
what data would change the answer. Reference the 2026 medians I give you
(opt-in trial ~14%, card-required trial ~44%, freemium ~4.5%) and say
which my product realistically lands near.
Context: [product, segment, sales motion, willingness-to-pay signal]Variables to swap: product, segment, sales motion, WTP signal
Optimization: If the recommendation is too generic, add: “Justify by naming a comparable named product and what they tried — what worked, what failed.”
2. Tier structure builder (3-tier)
Design a 3-tier pricing structure for [product]: Starter, Pro, Business.
For each: target persona, price point, top 5 features, what is excluded,
the upgrade trigger from the tier below. Price the middle tier at target
margin (most buyers pick it). End with which tier you expect to over-sell
and which to under-sell, and why.
Context: [paste]3. 2-tier minimalist alternative
Argue the case for cutting our pricing to 2 tiers instead of [current N].
What would each tier include, what gets killed, what moves to add-ons.
Predict revenue impact over 1 quarter. End with the strongest objection
to going simpler.4. Usage-based vs seat-based decision
For [product], recommend usage-based vs seat-based vs hybrid pricing.
For each option: ideal customer profile, expected ACV impact, churn-risk
profile, billing complexity. Pick one and name the metric or threshold
that should make us revisit it.5. Anchor + decoy structure
Design a 3-tier pricing page with a deliberate anchor (high-price tier)
and decoy (a tier that makes the target look obvious). For each tier:
price, features, intended psychological role (anchor / decoy / target).
Keep the price gaps meaningful but not extreme so the compromise effect
holds. Show how a buyer scans the page in 8 seconds.6. Willingness-to-pay interview script
Generate a 30-minute customer interview script to surface
willingness-to-pay for [product]. Use the Van Westendorp 4-question
framing (too expensive / expensive but acceptable / a bargain / so cheap
I'd doubt quality) plus 4 open-ended follow-ups about anchors. End with a
checklist of red flags (interviewer-leading questions) to avoid.7. Pre-mortem of a price increase
We plan to raise prices by [X%] on [date]. Run a pre-mortem: 5 ways this
could fail (mass churn, NPS drop, negative press, sales-team revolt,
competitor weaponization), and the smallest mitigation for each. End with
a kill-switch trigger ("if X happens, roll back").8. Grandfathering policy
Design a grandfathering policy for an upcoming price increase. Options:
full grandfather, 12-month grace, partial discount, no grandfather.
For each: revenue impact, churn risk, brand-trust impact. Recommend one
with reasoning.9. Tier explainer rewrite
Below is our current pricing page. For each tier, write a 1-line
"this tier exists for X" statement. If you cannot, that tier is redundant.
Output: tier, current copy, proposed 1-liner, kill / keep / merge.
[paste pricing page]10. Packaging vs pricing audit
Audit our pricing problem: is it packaging (wrong bundle) or pricing
(wrong number)? Score 1-5 on each of 6 dimensions: feature-tier fit,
upgrade-path obviousness, decoy effectiveness, price-feature ratio,
competitor parity, willingness-to-pay alignment. Recommend the smallest
fix that moves the most.11. Per-seat vs flat-rate trade-off
For a [team-size] customer, calculate effective cost under per-seat vs
flat-rate vs hybrid. Show the breakeven seat count and where each model
creates buyer friction. Recommend which to lead with and the team size
at which the recommendation flips.12. Add-on vs core feature decision
Below are 8 features we are considering. For each, decide: include in
core, make a paid add-on, or push to a higher tier. Decision criteria:
usage frequency, dev cost, perceived value, willingness-to-pay signal.
Features: [paste]13. Competitor pricing reverse-engineering
For each competitor in [list], infer the pricing logic from their public
page: who they expect to buy each tier, what they signal with anchor
pricing, where they hide cost. End with one move we could make that none
of them are doing.14. Price-test design
Design an A/B price test for [product]: variants (control vs +20% vs
+50%), sample size, success metric, guardrails (CAC, churn, support
load), duration, kill criteria. Mark which decisions cannot be A/B-tested
ethically and must be run sequentially by cohort instead.15. Pricing-page copy from structure
Given this finalized tier structure, write the pricing-page copy. For each
tier: a value statement under 12 words, 4 feature bullets, 1 social-proof
line. Then write the FAQ block (5 Qs covering refunds, billing cadence,
upgrade, downgrade, custom plans).
Structure: [paste]Common mistakes
- Pricing by gut without writing the hypothesis — you cannot test what you did not articulate.
- Copying a competitor’s price without copying their cost structure or sales motion.
- Running 3 tiers when 2 would do, or 4 when 3 is the ceiling for most buyers.
- A free tier with no upgrade trigger — it just costs you money and muddies positioning.
- Raising prices without a grandfathering policy — a predictable churn spike.
- A pricing page that is pure feature lists — buyers scan for value, not feature counts.
- Skipping willingness-to-pay interviews and relying only on AI guesses.
How to push results further
- Write the “this tier exists for X” line for every tier before publishing. If you cannot, the tier is dead weight.
- Pair every price hypothesis with the metric that would falsify it (e.g., churn over 8% monthly on the top tier).
- Anchor pricing only works when the anchor tier is plausible. Fake anchors get caught.
- Test by cohort and date, not blanket A/B — pricing tests are sensitive and have ethics issues (charging two users different prices for the same thing).
- Talk to 10 customers about pricing before any change. AI cannot replace this.
- When in doubt, raise prices to acquire fewer, better customers. Rarely the other way.
- Refresh the pricing model once a year. Product changes faster than the pricing page does.
FAQ
- Should I always start with freemium?: No. Freemium fits high-frequency, network-effect, or content products and converts around 4.5% free-to-paid in 2026. For high-touch B2B, a card-required trial (median ~44%) usually beats it.
- How many tiers is right?: Most SaaS land at 3 plus custom enterprise. Going below 3 forces fewer decisions; going above 3 confuses buyers. On 3-tier pages, 60–70% of buyers pick the middle option, so price it at your target margin.
- When can I raise prices?: When new-customer NPS is healthy, monthly churn is under 3%, and at least 3 recent customers said “we would have paid more.”
- Should I show prices publicly?: For self-serve, yes. For enterprise, you can hide the custom tier but always show starter prices — hiding all pricing damages trust.
- How do I avoid grandfather-rage on a price increase?: Notify 60 days ahead, offer a 12-month lock-in at the old price, and personally email top accounts before the public announcement.
- Which model should I run these in?: Any current chat model works. Use a thinking/reasoning mode (GPT-5.5 Thinking, Claude Opus 4.7, Gemini 3.1 Pro) so it shows the trade-off math behind each recommendation.
Related
- Generating pricing hypotheses with AI before launch
- Startup idea evaluation prompts
- Feature prioritization prompts
- Growth experiment prompts
- Churn analysis prompts
- Product, App & Startup Prompts hub
External references: ChartMogul SaaS Conversion Report and the Van Westendorp Price Sensitivity Meter.
Tags: #Prompt #Product startup #Pricing