Your best customer pays you $49 a month. Your worst pays $49 a month too. That's the flaw hiding inside most subscription businesses, and it doesn't show up in your MRR chart. It shows up 14 months later, when the heavy users have quietly eaten your gross margin and the light users have churned because they felt overcharged for what they actually used.
A pricing strategy for a subscription business is the decision about who pays what, for which value, on which cadence, and what happens when their usage changes. Get it right and pricing becomes your strongest growth lever. Get it wrong and every other metric—CAC payback, LTV, net revenue retention—gets distorted by a number you set once and never revisited.
I'll be blunt about my own history here: I left a flat-rate plan in place for 19 months because churn looked stable. When I finally segmented usage, roughly a third of accounts were consuming four times the average and paying the same. Fixing it took one quarter and cost me two customers who screamed on the way out. It was still the right call.
Key Takeaways
- Price against value delivered, not against your cost structure—cost-plus pricing caps your ceiling at your own inefficiency.
- Flat-rate plans are the easiest to sell and the fastest way to subsidize your heaviest users.
- Usage-based and per-seat models both work; the deciding factor is whether usage tracks value or just tracks activity.
- Grandfather existing customers when you raise prices, and give a dated migration window. Surprise price changes are the most reliable churn generator I know.
- Test on cohorts, not on your whole base. A price test run across everyone is not a test, it's a gamble.
- Free plans only pay off when the conversion path is short and the ceiling is genuinely felt.
What are the 5 C's of pricing, and why do they matter for subscriptions?
The five C's are cost, customers, competitors, channel, and constraints. They're a checklist, not a formula, and for subscriptions the order matters more than the list.
Cost tells you your floor—what you must clear to keep gross margin healthy. Customers tell you the ceiling: what the buyer believes the outcome is worth. Competitors give you a reference point, but only a reference point; anchoring your price to a competitor's is how whole categories end up commoditized. Channel covers the cost of acquiring and serving through each route—a self-serve signup and a sales-assisted deal have wildly different economics and should not share a price. Constraints are the boring ones that bite: payment processing, tax handling across jurisdictions, contract terms, and whatever your billing system can actually meter.
Where the framework breaks down
Here's the part the checklists gloss over. For a subscription, cost is not a single number—it's a curve that bends with usage. Your cost to serve one account at 100 API calls is not the same as at 100,000. If your pricing model assumes a flat cost, your margin quietly becomes a function of customer behavior you never modeled.
In my experience the framework is most useful as a constraint solver: competitors and constraints eliminate options, customers set the target, cost validates that the target is survivable. Run it in that order and you waste less time arguing about numbers that were never available to you anyway.
What are the 7 main types of pricing strategies?
The seven you'll encounter most often are cost-plus, value-based, competitive, penetration, skimming, freemium, and dynamic pricing. Below is how each behaves once you put it inside a recurring revenue model, which is where the differences actually show.
| Strategy | How it works | Best fit | Main risk |
|---|---|---|---|
| Cost-plus | Price = cost + fixed margin | Commodity infrastructure, low differentiation | Ceiling set by your own cost base |
| Value-based | Price anchored to customer outcome | B2B tools with measurable ROI | Requires real willingness-to-pay research |
| Competitive | Price near market reference | Crowded categories, replacement purchases | Race to the bottom |
| Penetration | Deliberately low entry price to win share | Two-sided platforms, network effects | Hard to raise later without churn |
| Skimming | High launch price, lowered over time | New capability with few substitutes | Invites fast undercutting |
| Freemium | Free tier, paid upgrade | Product-led acquisition, short time-to-value | Most free users never convert |
| Dynamic | Price varies with demand or load | Capacity-constrained services | Unpredictability scares buyers |
Two of these I'd avoid by default in a subscription context. Penetration pricing works once and poisons the well afterward—raising from a deliberately low anchor produces churn disproportionate to the increase, because the anchor was never the real value. Cost-plus is fine if you sell undifferentiated capacity, and a slow death if you sell outcomes.
Value-based pricing is where I'd put the effort, and I'll defend that position. The difficulty isn't philosophical, it's methodological: you have to go find out what the outcome is worth, and most teams skip that research because it feels slower than picking a number off a competitor's page. It isn't slower. It's just less comfortable.
How do you price your subscription service in practice?
Start with the unit of value, not the number. Before you name a price, decide what the customer is buying more of when they pay more. Seats, API calls, projects, contacts, transactions, storage, active users—the choice of unit is the pricing decision. The dollar figure is downstream of it.
Step 1: define the value metric
A good value metric passes three tests. It grows as the customer succeeds. It's predictable enough that the customer can forecast their bill. And it's cheap for you to meter accurately. Seats pass all three for collaboration tools. Seats fail badly for tools where value comes from automation—as headcount falls, your revenue falls with the customer's efficiency, which is exactly backwards.
Step 2: research willingness to pay
The Van Westendorp price sensitivity meter is the cheapest tool that produces usable answers. You ask four questions: at what price is this so expensive you'd never consider it, at what price is it expensive but you'd still consider it, at what price is it a bargain, and at what price is it so cheap you'd doubt the quality. Plot the cumulative curves and the intersections give you an acceptable range and an indifference point.
Run it on 30 to 40 real prospects, not on your friends. I made that mistake early: I surveyed people who liked me, got an inflated range, launched high, and spent two quarters discounting. The data was worthless because the sample was polite.
Step 3: set tiers that separate, not decorate
Most tier structures I audit have three plans where the middle one is a decoy and the top one exists to make the middle look reasonable. That's legitimate—anchoring works. What doesn't work is a top tier that nobody buys because it bundles features nobody asked for.
- Entry tier: one clear outcome, hard limits, self-serve only. Its job is qualification, not revenue.
- Middle tier: where most revenue should land. Price it against the value of the primary job, and check that your usage data says the majority of active accounts fit inside its limits.
- Top tier: usage headroom plus the operational things larger buyers need—audit logs, SSO, higher rate limits, a support commitment.
- Enterprise: no public price. Custom contracts only exist because procurement wants them, not because your product changed.
One number worth watching: net revenue retention. If it sits below 100%, your existing customers are shrinking their spend faster than you're expanding it, and that is a pricing-model problem far more often than a product problem.
Step 4: test on cohorts
Price tests need clean separation. Split by acquisition cohort or by region, hold the test for a full billing cycle at minimum, and measure conversion, expansion, and gross margin per account—not just signups. A price change that lifts revenue per customer by 20% while doubling month-two churn is a loss dressed as a win.
How do you value a subscription-based business?
Recurring revenue businesses are valued primarily on recurring revenue quality, not on a simple multiple of last year's sales. The practical inputs are ARR, growth rate, gross margin, churn, and how concentrated the customer base is.
Three metrics do most of the work. Annual recurring revenue sets the base. Gross margin tells you how much of each dollar survives delivery—a business at 85% margin and one at 55% can have identical ARR and wildly different values. Net revenue retention tells you whether the base grows on its own; a business expanding above 100% is compounding without new sales, and buyers pay for that. Customer concentration cuts the other way: if your largest account is 30% of revenue, expect a discount on the whole.
Free cash flow matters more than headline growth at scale. And payback period on acquisition cost—how many months of gross profit it takes to recover what you spent to win a customer—is the number that quietly determines whether your growth is funded or financed.
What buyers actually probe is durability. Does the pricing model hold when customers grow? Does it hold when they shrink? A flat-rate business looks clean on paper until a diligence team models usage growth and finds the margin curve bending the wrong way. That's a valuation problem created by a pricing decision, and I've seen it shave real money off a deal.
How do you handle a price change without losing your base?
Grandfather the existing base, then migrate them on a schedule you announce. That's the whole answer, and most companies skip the second half.
Concretely: new customers get the new price immediately. Existing customers keep the old price for a defined window—six to twelve months is common—then move to the new structure with a written notice well in advance. Offer an annual prepay option at the old rate for a limited period; a meaningful share of customers will take it, and you convert churn risk into cash up front.
What kills you is the surprise. A price increase delivered with 30 days' notice and no grandfathering turns your most loyal accounts into your loudest critics, and in a market with visible alternatives they leave on principle, not on economics.
One thing I'd do differently than most: I now communicate price changes to the accounts most affected personally, before the email goes out. The two customers I lost during my own repricing were both on the flat plan and both heard about it from an automated notice. Neither conversation would have saved the contract, but both would have saved the referral, and referrals were worth more than either account.
Does freemium ever make sense?
Yes—when time-to-value is measured in minutes and the free tier creates a ceiling the user can feel. Collaboration tools work because the free user invites others, and the invite is the growth mechanism. Developer tools work because a free key gets embedded in a codebase and the upgrade happens when production load arrives.
What doesn't work is freemium as a marketing tactic for a product with a long sales cycle. If your buyer needs a demo, an implementation, and a procurement review, a free tier mostly generates support cost. I've watched a team run free accounts for a year at roughly 2% conversion with a support load that ate the margin on the paid base. The free tier wasn't the problem. The product's time-to-value was, and freemium just made it visible.
The part nobody puts in the framework
Pricing is a decision you make with incomplete information, and the most useful thing you can do is shorten the loop between setting a price and learning whether it was right. Teams that revisit pricing twice a year consistently outperform teams that set it once and defend it. Not because they guess better, but because they find out sooner.
The question I'd leave you with: if your heaviest users paid what their usage actually costs you to serve, would your current price list survive the comparison? If the answer is uncomfortable, that's not a reason to wait for the next planning cycle. It's the reason to run the segmentation this month.