TL;DR: Feed an AI your audience size, conversion rate, and three real buyer personas, then make it produce three tiers built on genuine capability gaps — not a decoy middle nobody buys. Use the prompt below, then check every tier against one question: would a rational buyer pick it on its own merits, ignoring the other two? Anchor the page on the tier you want read as the default purchase (usually the middle), default to monthly billing with a two-months-free annual discount (~16.7%), and forecast ARPU across three mix scenarios before you lock prices.
The task
You are launching paid tiers — for a newsletter, a community, a SaaS, a Patreon — and you need three of them. Every pricing post tells you to insert a “decoy” middle tier nobody buys to push people up. That trick worked in 2014 and now reads as condescending: your buyers have seen it, they resent it, and the smart ones leave.
The data does not even support the lazy version. Three-tier pricing pages convert at roughly 1.4x the rate of two-tier pages, but pages with four or more tiers convert worse — so the win comes from three distinct tiers, not from padding the page with options. You want a memo that designs three tiers around real value differences, picks an anchor that signals where you charge, and forecasts ARPU before you commit. AI is a fast first draft of the memo, not the pricing decision itself.
Use a reasoning-grade model for this, not a quick chat. As of June 2026, GPT-5.5 (Thinking mode) and Claude Opus 4.7 both hold the full memo — personas, three tiers, the ARPU grid, and the decoy audit — in one context and reason across them. A default instant model will give you three feature lists with one checkbox moved.
When this is the right job for AI
- You can name three meaningfully different things you do for three different buyer types — not three feature lists with one moved checkbox.
- You have an audience size, a conversion rate guess, and a target monthly revenue.
- You will pressure-test the memo against five real customers (not your group chat) before locking the prices.
What to feed the AI
- Product type and what the buyer actually gets (access, deliverables, software seats, community)
- Three distinct buyer personas — name, what they do, what makes each one open the wallet
- Audience size you are pricing against and a realistic conversion rate (see the benchmark table below)
- Your target monthly recurring revenue and the headcount you need to support delivery
- The pricing the closest 3 competitors charge and what they do at each tier
- Any features that legally or operationally must live at a specific tier (e.g. SSO at top, 1:1 calls at top)
Pick a conversion rate the math can survive
Do not guess “1-3%” for everything — the right number depends entirely on your funnel. These are median free-to-paid conversion benchmarks as of early-to-mid 2026:
| Funnel type | Typical median | ”Good” range |
|---|---|---|
| Freemium (free tier lives alongside paid) | ~4.5% | 2-8% |
| Opt-in free trial (no credit card) | ~14% | 8-22% |
| Reverse trial (full features, then downgrade) | ~24% | 18-32% |
| Opt-out free trial (credit card required) | ~44% | 35-55% |
The single biggest lever is the credit-card requirement: trials that ask for a card convert at roughly 5x the rate of those that do not. Feed the AI the number that matches your funnel, not a blended average.
Copy-ready prompt
Draft a 3-tier subscription memo for my launch.
Product type: [newsletter / community / SaaS / Patreon / hybrid]
What the buyer gets: [tangible deliverables]
Three buyer personas:
Persona A: [one sentence — who they are, what triggers purchase]
Persona B: [one sentence]
Persona C: [one sentence]
Audience size right now: [N]
Conversion rate (match my funnel type): [%]
Target MRR: [USD]
3 competitors and their pricing: [paste]
Constraints — must-live-at-top features: [list]
Output 5 sections:
1) The 3 tiers. For each: name, monthly price, annual price (two months
free), the persona it is for in one line, the 3 things that meaningfully
differ from the tier below, and the one thing that intentionally does NOT
scale up (so the tier is bought for itself, not as a step ladder).
2) Anchor logic — which tier is the anchor, why, and what signal it sends.
The anchor is NOT "the most expensive one" by default; it is the one I
want the buyer to perceive as the typical purchase.
3) ARPU forecast — for [audience size] at [conversion], model three mix
scenarios (40/40/20, 25/50/25, 10/60/30 across low/mid/high) and show
MRR for each. Show the arithmetic. Do not round MRR up.
4) Decoy check — for each tier, one sentence: "Would a rational buyer in the
target persona pick this tier on its own merits, ignoring the other two?"
If no, the tier is a decoy and must be redesigned.
5) Five pressure-test questions to ask real prospects before I lock prices.
Banned moves: pricing ending in 7 or 9 unless I asked; "Unlimited" without a
soft cap; "Custom" tier with no price range listed.What good output looks like
A useful tier line: “Tier 2 — Builder — $79/mo or $790/yr — for the working PM who wants the templates and the monthly office hours but does not need the 1:1 review. Differs from Tier 1 by: full template library, monthly group office hours, private community. Does not include: 1:1 review (that is Tier 3’s reason to exist).”
A useful anchor note: “Anchor is Tier 2. Tier 1 reads as ‘try before you commit,’ Tier 3 reads as ‘I want concierge.’ The buyer’s default mental purchase is Tier 2 — design the page for that.”
A useful pressure-test question: “If we removed the 1:1 review from Tier 3 entirely, would you still buy Tier 3 at $189, or drop to Tier 2 at $79?”
Read the ARPU grid before you celebrate
The ARPU section is where most memos quietly fall apart. The same three prices produce wildly different revenue depending on which tier people actually pick. Make the AI show the grid so you can see how fragile the plan is.
Worked example: 5,000-person audience, 3% paid conversion = 150 subscribers, at $29 / $79 / $189.
| Mix (low / mid / high) | Subscribers | Blended ARPU | MRR |
|---|---|---|---|
| 40 / 40 / 20 | 60 / 60 / 30 | $81.00 | $12,150 |
| 25 / 50 / 25 | 38 / 75 / 38 | $94.73 | $14,209 |
| 10 / 60 / 30 | 15 / 90 / 45 | $107.00 | $16,050 |
The point is not the exact dollars — it is the spread. The same launch swings from ~$12k to ~$16k MRR purely on tier mix, which is why a strong middle tier matters far more than a clever top tier. If your plan only works in the 10/60/30 world, you are betting the launch on an outcome you cannot control.
Common mistakes
- Three tiers separated only by quantity (10 / 50 / 100 of something) — buyers read it as “pay more for the same product.”
- A decoy middle nobody buys — short-term it nudges revenue, long-term it bleeds trust and gets called out on social. Real value differences beat asymmetric-dominance tricks because they survive a second look at the page.
- Custom enterprise tier with no anchor — without a starting number, prospects assume the worst and never reach out.
- Annual discount above 20% — the standard is two months free (~16.7%); 15-25% is the defensible range. Go deeper and the LTV math collapses for anyone who would have stayed monthly.
- Grandfathering everyone forever — fine for the first 50 buyers, ruinous past 500. Write the grandfather window into the memo.
- Hidden fees (transaction, processing, “platform”) — fold them in or commit to none. Buried fees roughly double churn in the first 90 days.
If you are on a creator platform, price the platform cut in
Creator platforms take a real bite, and as of June 2026 those cuts changed enough to break old napkin math. Bake the net into your ARPU before you set tier prices:
| Platform | Headline cut | Realistic net loss (with processing) | Note |
|---|---|---|---|
| Patreon (new creators) | 10% flat | ~12-15% of gross | Flat 10% since Aug 4, 2025; pages created earlier keep older tiered rates while published |
| Substack | 10% | ~13-16% of gross | 10% + Stripe ~3.6% per transaction |
| iOS in-app (either) | Apple 30% (15% after year 1) | Platforms auto-raise iOS prices ~43% to offset | New memberships bought in-app are pricier than web |
If a meaningful share of your buyers will subscribe on iOS, your real take-home on a $20 web price is closer to a $20 web sale minus the platform cut, while the in-app shopper pays ~43% more for the same tier. Tell the AI your platform so the ARPU grid uses net revenue, not gross.
How to refine the draft
- All three tiers look the same: “Each tier must have one capability the tier below cannot get at any price. State that capability in one sentence.”
- The middle is a decoy: “Redesign Tier 2 so a buyer chooses it on its own merits, not as a comparison to Tier 1 or 3. If you cannot, kill the tier and ship two.”
- ARPU math is hand-wavy: “Use the exact conversion and audience numbers I gave. Show the arithmetic. Do not round MRR up.”
- Pricing reads anchoring-heavy: “Drop any price ending in 9 unless I asked. Round prices read as honest.”
- The ‘Unlimited’ tier is dangerous: “Replace ‘Unlimited’ with a generous soft cap (e.g. 100/month, fair use). Open-ended commitments destroy unit economics within 6 months.”
FAQ
- How many tiers should I ship? Three is the workable maximum for most creator and small-SaaS launches; three-tier pages convert ~1.4x better than two-tier, but four or more convert worse. Add a fourth only when there is a real plan-level capability gap (seats, infra limits).
- Monthly or annual default? Show monthly first. Buyers who want annual will switch on their own; pushing annual at first purchase increases sticker shock and lowers signup conversion.
- What discount on annual? Two months free (~16.7%) is the industry standard and the easiest to explain (“12 months for the price of 10”). The defensible range is 15-25%; beyond that you erode LTV.
- What conversion rate should I plug in? Match your funnel: ~4.5% median for freemium, ~14% for a no-card free trial, ~44% for a card-required trial. A blended “1-3%” understates trial funnels and overstates freemium.
- What about pay-what-you-want? Works for one-time products with high goodwill (album, ebook). A disaster for subscriptions — the recurring decision reopens every month.
- Should I A/B test prices at launch? No. Pick prices the memo defends, ship, watch for 60 days, then test one variable. Early traffic is too thin for a clean read.
Related
- Digital Product Sales Page AI
- Affiliate Program Launch AI
- Newsletter Lead Magnet AI
- Brand-Deal Outreach AI
- AI Product Copy
- Budget Narrative AI
Tags: #AI writing #Subscription #Pricing #creator-monetization