How to Price a Product

Pricing is the highest-impact decision most founders touch least. A 10% price increase, all else equal, routes directly to the bottom line on every future sale. A 10% acquisition improvement saves you the CAC but leaves margin unchanged. Yet pricing conversations tend to happen once at launch and then never again.

This guide walks the three standard pricing approaches, their honest trade-offs, and the minimum evidence each requires. The framework follows the standard Nagle treatment^[1].

1. Pick a pricing model that fits the evidence

Three decision points narrow the choice quickly:

• Do I have direct-variable-cost visibility? If yes, cost-plus gives you a useful floor. If no (SaaS with mostly fixed costs), cost-plus is close to useless.
• Are there three or more direct competitors I can benchmark? If yes, competitive pricing gives you a ceiling signal. If no, you are probably in a nascent category and benchmarks will mislead.
• Can I measure the quantified outcome my product produces for the buyer? If yes, value-based pricing is both defensible and profitable. If no, it degenerates into storytelling.

Most products should use at least two of the three — cost-plus as a floor, competitive as a sanity check, value-based as the aspiration. Using only one is the common mistake.

2. Cost-plus: the floor, not the answer

Cost-plus pricing is Total Unit Cost × (1 + markup). For a product with $30 variable cost and a target 60% gross margin, price = $30 / (1 − 0.60) = $75, a 150% markup.

This gets you a price. It does not get you the right price. Cost-plus ignores what the buyer will pay, which in most markets exceeds what the seller calculated from the cost stack. Nagle's objection is that cost-plus treats cost as the anchor when the buyer's willingness to pay should be, and that it is circular: unit cost depends on volume, and volume depends on the price you are trying to set^[1].

Use cost-plus to answer one question: what is the minimum price below which I am operating at a loss? Never set the selling price there. Add the willingness-to-pay signal and competitive benchmark on top.

3. Competitive: benchmark the ceiling

Competitive pricing means setting price relative to direct substitutes. The method is simple: identify 3–5 genuinely comparable products, tabulate their public pricing, and decide whether you are premium, parity, or discount. Each position requires supporting evidence.

Two failure modes:

• Picking the wrong comparison set. If your product is meaningfully differentiated, pricing against the nearest competitor strips out the premium your differentiation should command.
• Public pricing is not actual pricing. Especially in B2B, the published list price and the average realised price diverge once discounts, negotiated terms, and enterprise concessions are included. Ask buyers and lost prospects what they were actually quoted, and factor the gap in when benchmarking.

Competitive pricing as the primary model works for commoditised categories. For differentiated products, it should constrain your pricing but not determine it.

4. Value-based: what it's worth to the buyer

Value-based pricing anchors on the economic value created for the buyer, minus a fair share retained. If your product saves a buyer $100k annually, pricing it at $20k captures 20% of the created value and leaves the buyer $80k better off, a split that is easy to defend in a sales conversation.

The challenge is measurement. You need (a) a quantified outcome model, (b) evidence that your product causes the outcome, and (c) a way to communicate the value to buyers. Without all three, value-based pricing becomes storytelling.

The Van Westendorp Price Sensitivity Meter is the cheapest way to get willingness-to-pay data from a small sample^[2]. Four questions to 30–100 prospects:

1. At what price would this be so expensive you would not consider buying?
2. At what price would this be expensive but you might still consider it?
3. At what price would this be a bargain?
4. At what price would this be so cheap you would question the quality?

Plot the share of respondents giving each answer at each price, with "too cheap" and "bargain" falling as price rises and "expensive" and "too expensive" rising. The crossings give four readings^[2]: the Optimal Price Point where "too cheap" meets "too expensive"; the Indifference Price Point where "bargain" meets "expensive"; and the acceptable price range, from the Point of Marginal Cheapness ("too cheap" meets "expensive") to the Point of Marginal Expensiveness ("too expensive" meets "bargain"). More responses give steadier crossings; treat a few dozen as a directional signal.

5. Test, price, revisit

For SaaS, packaging matters as much as price. A value metric (seats, API calls, revenue processed) lets revenue expand as the customer grows without a renegotiation; a flat fee only expands when the customer upgrades. The packaging decision is: what metric does the customer scale on, and how do I tie price to that metric.

Three packaging principles that hold up:

• Three tiers, not five. Every extra tier is another comparison the buyer has to make. Three — typically Starter, Growth, Enterprise — map to buyer segments cleanly.
• Anchor high, not low. Show the premium tier first visually. A higher-priced option makes the middle tier look more reasonable by comparison, the anchoring and decoy effects Ariely popularised^[4].
• One clear upgrade trigger per tier. Each tier should have one feature or limit the buyer crosses that forces the next tier. Vague value differences cost expansion revenue.

Revisit pricing annually at minimum. US consumer prices rose 3.7% in the 12 months to September 2026 and about 24% from 2020 to 2025 on annual averages^[3]; holding prices flat means a real-terms price cut. Grandfathering existing customers is common and defensible; raising new-customer pricing is not a breach of trust.

The worst pricing decision is the one that never gets revisited. Changing a price feels riskier than leaving it alone, so founders rarely test a higher one; the Van Westendorp survey below is a cheap way to find out whether there is room.

6. Price by segment when the data supports it

Different buyer segments often have different willingness-to-pay. Segment-based pricing captures more value if segmentation can be operationalised:

• Volume discounts. Standard in B2B. Larger customers pay lower unit prices in exchange for higher total commitment. Structure tiers by seat count, usage bands, or revenue bands.
• Geographic pricing. Common in international software, where lower-income markets are often offered parity pricing at a discount. Works when markets can be separated (e.g., separate purchase channels); breaks when arbitrage is easy.
• Industry-specific pricing. Regulated industries (healthcare, finance, government) often pay premiums for compliance features. Unregulated segments get base pricing.
• Academic / non-profit pricing. Common discount tier. Builds goodwill and early-career exposure to the product.

Segment-based pricing risks: arbitrage (customers buy through the cheaper channel), transparency concerns (buyers discovering peers pay less), and administrative overhead (validating segment eligibility). Run the cost-benefit before introducing multi-tier segment pricing.

7. Discounts and promotional discipline

Discounting is widely over-used. Four rules that help:

• Discounts should cost less than they save. A 20% discount that accelerates a deal close by 30 days is usually positive-value. A 20% discount that closes a deal that would have closed anyway is pure margin erosion.
• Prefer time-bound discounts to permanent ones. "25% off if you close by end of quarter" creates urgency and ends. Permanent discounts become the baseline price.
• Multi-year prepayment trades margin for cash. Treat the discount as the price of that cash. Against the monthly payments you would otherwise collect, 10–15% off for a year paid upfront costs the same as borrowing at about 27–44% a year, and 25% off for three years upfront about 23% a year. That is expensive money unless the cash funds something that returns more, or the commitment keeps customers who would otherwise churn mid-term.
• Authority to discount is a cost. Discount authority that exists gets used. Structure approval levels to match the margin at stake — small discounts at rep level, larger at manager, material at executive level.

In the typical case, sustained discounting of 20%+ off list price is a signal that list price is aspirational rather than realistic. Raise the list price less, or discount less — mixing both is the pattern that produces the worst price realisation.

8. Numeric worked example — Van Westendorp on a micro-SaaS

A bootstrapped SaaS with 300 active trial users runs the four Van Westendorp questions through in-app surveys with an 18% response rate (54 respondents). The raw curves intersect at roughly:

Question                              Median  P25    P75
──────────────────────────────────────────────────────────
Too expensive (won't consider)        $89     $65    $120
Expensive but still consider          $59     $42     $75
Bargain                               $29     $22     $39
Too cheap (quality doubt)             $15     $10     $19

Share of respondents at each price
Price   Too cheap   Bargain   Expensive   Too expensive
$29         8%        50%         5%            1%
$35         4%        33%        11%            2%
$39         2%        24%        17%            3%
$45         1%        15%        26%            7%
$50         1%        10%        35%           10%

Point of Marginal Cheapness    (too cheap × expensive)      ≈ $31
Optimal Price Point            (too cheap × too expensive)  ≈ $37
Indifference Price Point       (bargain × expensive)        ≈ $41
Point of Marginal Expensiveness (too expensive × bargain)   ≈ $50
Acceptable price range                                      ≈ $31–$50

Current list price: $29, just below the acceptable range: at $29 more respondents call the product too cheap (8%) than expensive (5%). Phase the change: new customers at $39 immediately (+34%), which sits between the optimal ($37) and indifference ($41) points; treat $49, at the top of the acceptable range, as a later test only if trial-to-paid conversion holds; grandfather existing customers for 12 months at the current rate.

9. Failure modes worth naming

• Reading Van Westendorp with fewer than 30 responses. The four curves are noisy below a meaningful sample: with 12 responses, each answer is more than 8% of a curve, so one or two respondents can move a crossing. Resample your own responses to see how far the crossings move before acting, and don't make a repricing decision off a 12-person survey.
• Pricing off competitor list, ignoring discount reality. Enterprise software is often sold below list. If competitors discount and you don't, pricing at their list makes your realised price higher than theirs, an unintentional premium position that gets flagged in every buyer comparison.
• Grandfather period too generous. Three-year grandfathering on a 50% price increase is effectively a permanent two-cohort pricing system with operational complexity for years. Cap grandfathering at 12 months unless the customer relationship genuinely warrants longer.

Companies that held nominal prices flat from 2021 to 2024 absorbed about 16% of US inflation on CPI annual averages^[3], a real-terms price cut of about 14%; catching up requires a deliberate price event, not a gradual creep.

Frequently asked questions

What is the best pricing model for a bootstrapped product?

There is no single best model; use at least two of the three standard approaches. Cost-plus (total unit cost times one plus markup) gives you a price floor below which you operate at a loss — never set your selling price there. Competitive pricing benchmarks 3 to 5 genuinely comparable products to signal a ceiling. Value-based pricing anchors on the economic value created for the buyer and is the most profitable when you can measure that outcome. Nagle's objection to cost-plus is that it treats cost, not willingness to pay, as the anchor, and that it is circular: unit cost depends on volume, and volume depends on the price you are trying to set. Use cost-plus as a floor, competitive as a sanity check, and value-based as the aspiration.

How do I find willingness to pay with a small customer base?

The Van Westendorp Price Sensitivity Meter is the cheapest way to get a credible willingness-to-pay range from a small sample. Ask 30 to 100 prospects four questions: at what price is it too expensive to consider, expensive but still worth considering, a bargain, and so cheap you doubt the quality. Plotting the four curves gives an optimal price point (too cheap crosses too expensive), an indifference price point (bargain crosses expensive), and an acceptable price range bounded by too cheap crossing expensive and too expensive crossing bargain. More responses give steadier crossovers; treat a few dozen as a directional signal, and with a dozen, one or two answers can move the result, so do not reprice off a 12-person survey.

How many pricing tiers should I offer?

Three is a sound default. Three tiers — typically Starter, Growth, and Enterprise — map to buyer segments cleanly, and each extra tier is another comparison the buyer has to make. Anchor high by showing the premium tier first: a higher-priced option makes the middle tier look more reasonable by comparison. Give each tier one clear upgrade trigger — a feature or limit the buyer crosses that forces the next tier — because vague value differences between tiers cost you expansion revenue.

How often should I revisit pricing?

Revisit pricing at least annually. US consumer prices rose 3.7 percent in the 12 months to September 2026 and about 24 percent from 2020 to 2025, so holding prices flat is a real-terms price cut. Pricing on a value metric (seats, usage, or outcomes) lets revenue grow as a customer's use grows, which a flat fee cannot. Grandfathering existing customers for up to 12 months is common and defensible; raising new-customer pricing is not a breach of trust.