Key Takeaways

• Good-better-best splits one product into three rungs so buyers with different willingness to pay all find a price.

• The middle rung wins on comparison, not merit. In Ariely's Economist study, 84% picked the $125 print and web bundle next to a $125 print-only option, and 32% picked it once that option disappeared.

• Vercel runs a live three-rung ladder: Hobby at "$0/mo.", Pro at "$20/mo.", Enterprise at "Custom", with limits separating the bottom two and security plus SLA separating the top.

• Each rung should withhold something a real buyer segment can name. A rung that only adds a longer feature list gets ignored.

• The ladder breaks when cost of delivery moves per customer, which is why AI and infrastructure products meter on top of the rungs instead of adding a fourth.

How do you build a good-better-best ladder?

You build it by picking one axis buyers already argue about, then deciding what each rung withholds rather than what it adds. Withholding is the harder discipline, and it's what makes the rungs legible.

The separator changes at every step:

Those figures come off Vercel's published pricing page, checked 18 September 2026. The shape repeats: a limit, then headroom, then a promise someone in legal cares about.

The middle rung also gets help from the rungs beside it. Dan Ariely's Economist study, recounted in the decoy effect literature, offered online-only at $59.00, print-only at $125.00, and print plus web at $125.00. With all three shown, 16% chose online-only, nobody chose print-only, and 84% chose the bundle. Drop print-only and the bundle fell to 32%. The dominated option sold nothing and changed everything, which is why a top rung nobody buys still anchors the middle.

Two mistakes do most of the damage. Pricing the rungs on the unit the product already charges by, usually seats, collapses the ladder into plain per-seat pricing with cosmetic labels. The other is treating packaging as a rate structure: what a unit costs at each quantity band is tiered vs volume pricing, a separate question.

When does good-better-best stop working?

It stops working when the value a customer gets varies continuously and the price only moves in three steps. Three rungs assume buyers cluster into three groups, and many products have no such clusters.

The signals that a ladder has stopped fitting show up in the pipeline before they show up in revenue:

• Sales discounts the top rung on most deals, so it's priced for a segment that doesn't exist at that size.

• The middle rung absorbs accounts whose usage spans two orders of magnitude, so one price covers a $200 customer and a $20,000 one.

• Every enterprise deal adds a custom clause, and the "Best" rung becomes a negotiation starting point rather than a product.

• Gross margin on the top rung falls as usage grows, which happens fast on AI products where inference cost lands per call.

• The tier list grows a fourth and fifth rung, and the pricing page stops being readable.

Most teams I've seen fix the margin case without abandoning the ladder. They keep three rungs for access and support, then meter on top of each one, the structure covered in hybrid pricing model and consumption-based pricing. Vercel's Pro rung does that, charging $2 per additional million edge requests past the included 10 million.

The quieter failure is enforcement. A rung is only real if the product blocks what the page says it blocks, and that lives in feature gating, not in the pricing table. When the two drift apart, entry-rung customers keep using what they'd have upgraded to get.

Related terms

Packaging sits next to several pricing decisions worth reading separately:

• Per-seat pricing charges by user count, the unit most good-better-best ladders sit on top of.

• Tiered vs volume pricing covers the rate-structure question that gets confused with tier packaging.

• Feature gating is the enforcement layer that makes a rung's boundaries real in the product.

• Hybrid pricing model combines a fixed rung with metered charges when usage varies inside a tier.

• Consumption-based pricing prices by what a customer actually uses, the alternative when rungs stop fitting.

• Subscription pricing models surveys the wider set of structures a recurring product can charge on.

FAQ

How many tiers should a good-better-best model have?

Three is the default because it gives buyers a cheap option, a safe option, and a premium option without forcing a spreadsheet. Go to four only when a real segment sits between two rungs and names itself in sales calls. Five or more usually means the team is packaging around individual deals.

Is good-better-best the same as tiered pricing?

No. Good-better-best is a packaging decision about which capabilities sit behind which price. Tiered pricing is a rate structure setting what a unit costs at each quantity band. One product can run both: three named plans, each with graduated rates above its allowance.

Which tier should most customers buy?

The middle one, and by a wide margin. Healthy ladders put the bulk of accounts on "Better", with the entry rung feeding trials and the top rung carrying deals that need security review. If most revenue lands on the entry rung, the middle rung isn't withholding anything buyers care about.

What should go in the top tier?

Guarantees rather than features. The items that earn a premium are the ones a buyer's security, legal, or finance team asks for:

• Single sign-on, audit logs, and directory sync

• Uptime SLAs with credits attached

• Data residency or on-premise deployment

• Named support with a response-time commitment