Many founders’ first pricing move is copying competitors.

They bill monthly because rivals do; add annual because rivals do; bolt on credits because AI products do.

Then launch reveals: users never needed you that long.

A developer may use a Next.js boilerplate for a few hours.
An AI image user incurs cost on every generation.
A project tool may open every day.

All three are “software.” The billing logic is not.

The first pricing question is not “how much,” and not even “monthly vs annual.”

It is:

When does the user believe they are done with you?

That answer largely decides one-time, metered, or recurring. Add a second axis—does another user or another use cost you real money?—and the shape is mostly set.

This guide decides that shape only. Not the dollar amount. Not Stripe setup.

1. Keep this table first

Usage ↓ / variable cost →	Low marginal cost	High marginal cost
Episodic / one-shot	Buyout (or short quarterly)	Per use / credits, or very short sub
Ongoing use	Subscription	Sub + caps / overage, or pure usage

Judge two variables, not industry habit: usage cycle and your marginal cost. Wrong cycle kills conversion; wrong cost kills margin.

Three lines to remember:

• One-shot delivery → buyout
• Ongoing use → subscription
• More use = more loss for you → usage / sub + credits

2. Two extremes: why templates buy out, AI search often can’t

One-shot delivery: ShipFast (ShipAny same cell)

ShipFast is a strong counterexample to “software must subscribe.”

A reflexive $29/mo SaaS frame can add friction. Buyers want code today and their own product tonight. They are not opening ShipFast daily for ongoing service—they treat it as finished infrastructure.

What sells is:

A one-time code asset + time not spent rebuilding plumbing.

Once the project runs, core value is delivered. Buyout fits because delivery is episodic and post-delivery marginal cost is low—not because “templates always buy out.”

idoubi’s ShipAny sits in the same cell: an AI SaaS scaffold sold once, not a hosted service you live in → high one-time tickets feel natural. No second full story needed: same cell as ShipFast.

Continuous burn: ThinkAny-style AI search

Flip to ThinkAny-class AI search.

Imagine 100 searches a day, each burning model and retrieval cost. A “$99 lifetime unlimited” plan looks great on paper and absurd in the ledger:

Your biggest fans can be your biggest losses.

The issue is not “buyout too cheap”—it is model vs cost structure. Hence sub + credits / caps / overage is common: not “AI must subscribe,” but more use means more cost.

Middle: ongoing workflows (and short-window subs)

CRM, monitoring, collab, accounting: value keeps happening—data stays, people open daily. Default subscription. If variable costs spike (SMS, AI summaries), add caps or overage inside the sub.

Another mismatch: episodic need forced into long commitments. EasyCV dropping annual for quarterly aligned billing to the job-hunt window—still subscription, but short-window fit, not “SaaS must be annual.”

3. Core formula: two questions

Question 1: When does the user feel “done”?

Not “will they like it?” Ask:

At what moment have they already gotten the main value?

Product	Main value delivered when	Closer to
Resume tool	Resume finished	Buyout / per use / short quarterly
Image gen	Image out	Credits / per use
Next.js template	Project running	Buyout
Uptime monitor	Continuous watch	Subscription
CRM	Ongoing customer ops	Subscription

• Value delivered once → buyout (or usage packs)
• Value keeps accruing → subscription
• Value scales with volume → usage / sub + credits

Quarterly and annual are not “more advanced”—they only match billing to the real usage window (e.g. hiring season).

Question 2: Does another user / another heavy use cost you much?

• Almost nothing → buyout or sub both possible; then check cycle.
• Yes, and heavier use loses money → put cost into usage, credits, or price—don’t fake unlimited.

Two practical checks:

• Can buyers see ROI in one glance? (If not, fix the value story before more tiers.)
• Can you honor the promise? (Buyout + vague “lifetime updates” = long debt.)

Shape first; dollar amount and second tier later.

4. Wrong shape hurts more than “too expensive”

• Subscription = advanced by default: episodic tools on annual plans get rejected by “will I renew a year?”
• Buyout = easy win: high variable cost + lifetime = prepaid unlimited loss.
• Usage always converts better: unpredictable bills are a buyer cost too.
• Discounts hide wrong shape: 50% off annual still dead → suspect cycle first.

Signals to change shape (qualitative is enough)

Signal	Try
Lots of trials, almost no long subs	Shorter terms; or buyout / project fees
Buyouts hot, support/updates crush you	Narrow delivery; move updates to a sub add-on
Free usage maxed, costs up, pay rate low	Caps + credits / overage
10× usage spread on one flat monthly	Tiers or usage layer

Test on a slice of traffic; spell out old vs new; avoid overnight whole-site flips.

When this applies

Best for solo/small-team tools, templates, light SaaS, AI apps. Weaker for sales-led enterprise contracts and complex seat matrices. This is not legal, tax, or a substitute for price experiments.

30-second pricing shape check

① When does the user feel done?

• Once → buyout / per use
• Ongoing → subscription

② Does more use make you lose more?

• Yes → credits / usage / sub + overage
• No → buyout or sub, then pick by cycle

③ Can they see value vs price immediately?

• No → fix the value story before stacking tiers

If those three lines are fuzzy, don’t argue about annual discount percentages yet.

Related cases

• ShipFast: why a boilerplate sold ~$250k
• idoubi: ThinkAny traffic vs ShipAny cash
• EasyCV: drop annual, charge quarterly