How to Model Revenue and Margin from Ten Pricing Interviews

A demand curve answers one question: at price P, how many prospects in your pipeline would buy. With ten interviews you cannot draw a statistically robust curve, but you can draw a defensible one. This is the same logic as the Gabor-Granger method, which traces volume at each price point to find where revenue peaks, though those authors warn the output reflects hypothetical demand and should be read as a guardrail rather than a forecast (Conveo).

The simple approach: for each candidate price, count how many interviewees' upper bound sits at or above it. If you charge 75 dollars, everyone whose high bound is 75 or more is a "yes." From the table above, at 75 dollars, eight of ten say yes (interviews 1 and 2 have highs of 45 and 40, so they drop out). At 140 dollars, five of ten remain (interviews 5, 7, 8, 9, 10). At 200 dollars, only interview 8 remains.

Multiply that share by your pipeline to get potential accounts, then by price to get potential revenue:

• At 75 dollars: 8 of 10, times a pipeline of 40 similar prospects, is 32 accounts, or 2,400 dollars per month.
• At 140 dollars: 5 of 10, times 40, is 20 accounts, or 2,800 dollars per month.
• At 200 dollars: 1 of 10, times 40, is 4 accounts, or 800 dollars per month.

The middle price wins here, which is the common shape. Note what happened: dropping the price raised conversion but lowered revenue less than the loss of per-unit dollars, and the highest price left too many people out. This is the decision rule: compute revenue at three or four price points and pick the one that maximizes revenue, not the one with the most yeses.

One caveat on rigor: the aggregate of individual step-shaped willingness-to-pay functions only becomes smooth at large numbers of consumers, and with ten people you have ten steps, not a curve (Economics Stack Exchange). So treat your demand curve as a rough shape, not a precise forecast, and revisit it as interviews accumulate.

Now translate the demand curve into monthly numbers. You need three inputs: average price, new accounts per month, and monthly churn. The model has two regimes. In month one, revenue is simply new accounts times price. Over time, revenue approaches steady state where accounts in equals accounts out, and steady-state revenue is roughly monthly new revenue divided by the churn rate.

Run it with numbers. Assume you price at 100 dollars, add 10 new accounts per month, and lose 5 percent of customers each month. Monthly new revenue is 1,000 dollars. At a 5 percent churn rate, steady state is 1,000 divided by 0.05, or 20,000 dollars per month. That is the ceiling if nothing changes. Most early-stage teams find this number uncomfortable, because it shows how much churn caps growth regardless of how well sales does.

The month-by-month version is worth building because the ramp is not instant:

Two insights fall out of this table that a simple average hides. First, growth decelerates long before you hit the steady-state ceiling, so month 12 revenue is less than half of the eventual plateau. Second, the plateau is extremely sensitive to churn. If churn doubles to 10 percent, the steady-state ceiling halves to 10,000 dollars. This is the single most useful number in the model: churn rate is the denominator of everything.

If you want to keep the model honest, build it as a small spreadsheet with named inputs rather than hardcoded cells. A lightweight script works too:

price = 100
new_accounts = 10
churn = 0.05
accounts = 0
for month in range(1, 25):
    lost = accounts * churn
    accounts = accounts - lost + new_accounts
    print(month, round(accounts, 1), round(accounts * price))

Change one input at a time and watch which assumption moves the outcome most. In most early models, churn and new accounts per month dominate price.

Revenue is not margin. To get margin, subtract variable cost per account from price, then subtract fixed costs. Variable costs in a software business usually include hosting, support, and onboarding time. The US Chamber of Commerce's pricing guide walks through adding up variable costs per product first, then setting price against them, and separates variable costs like materials and per-sale fees from the ones that do not move with volume (US Chamber of Commerce).

Continue the example. At 100 dollars per month, assume variable costs are 25 dollars per account: 10 for hosting, 10 for support, and 5 for onboarding amortized over a year. Gross margin per account is 75 dollars, or 75 percent. At the 93-account month-12 level, gross margin is about 6,975 dollars per month. If fixed costs (salaries, tools, the founder's salary) total 6,000 dollars per month, you are barely profitable at month 12 and comfortably profitable near the 20,000-dollar steady state.

The model is only as good as the cost inputs, and both common pricing guides stress that underestimating fixed and variable costs undermines the whole plan (HubSpot). So keep fixed and variable costs in separate rows and update them monthly.

Margin per account also changes the price decision from earlier. If the 75-dollar price point has the same 25-dollar variable cost, margin per account is 50 dollars, and at 32 accounts that is 1,600 dollars of gross margin. At the 100-dollar price with 20 accounts (from the demand curve), margin is 1,500 dollars. The lower price is now slightly worse on margin, which is why you run the demand curve through the margin calculation, not just revenue.

A single-point forecast is a guess. Sensitivity analysis turns the model into a range. Change each input by a plausible amount, one at a time, and record the effect on steady-state revenue and margin.

Here is the same base case (100 dollars price, 10 new accounts, 5 percent churn) with each input flexed:

Read the top row first: doubling churn halves the business. Churn is the highest-leverage assumption you have, so if you are choosing where to spend effort, fixing retention beats raising price in almost every early-stage model. The third row shows that a 40 percent drop in new accounts cuts revenue 40 percent, which is linear and therefore easier to plan around than a churn surprise.

Discounting deserves its own line. The Chamber guide notes that business owners setting prices on guesswork can leave tens of percent of revenue on the table, and discounts compound that (US Chamber of Commerce). Model a 20 percent discount as a separate scenario. At 80 dollars per account with 25 dollars variable cost, margin per account falls from 75 to 55 dollars, a 27 percent margin loss for a 20 percent price cut. That asymmetry is why discounting to win deals is often worse than it looks in the sales conversation.

The fourth and fifth rows are identical on purpose: they show that a price cut only changes the margin outcome if variable cost per account also changes, which it usually does not. Price cuts hurt margin more than revenue, and the model makes that visible.

Grandfathering means letting early customers keep their old, lower price when you raise prices for new ones. It is a common retention tactic, and it is also a liability you should model before you promise it. The trade-off is simple: you keep early customers at a lower price and give up the difference in revenue, and the question is whether that retained revenue plus goodwill is worth more than the uplift you would get by raising prices on everyone.

Work it through with numbers. Say you have 30 early customers at 50 dollars per month, and you are moving new customers to 100 dollars. Grandfathering all 30 costs 30 times 50 dollars, or 1,500 dollars per month in foregone revenue. Over 12 months, that is 18,000 dollars. If raising prices on those 30 customers would cause a third of them (10 accounts) to churn, the comparison is: keep all 30 at 50 dollars (1,500 dollars) versus keep 20 at 100 dollars (2,000 dollars). In this case grandfathering costs you 500 dollars per month in revenue but retains 10 customers, so the question becomes whether those 10 are worth more than 500 dollars in future expansion and referrals. Often they are, but you should decide with the number in front of you, not by instinct.

Model grandfathering as a fourth set of rows in the sensitivity table: a scenario where a share of your base stays at the old price. Then set an explicit sunset. The cleanest grandfathering policy is a fixed window, such as 12 months, so the foregone revenue is capped and predictable. If you promise "locked in forever," model forever, because that is what you said.

The grandfathering decision is also a segmentation decision. If the early customers are in a different segment from your target, grandfathering them costs you little relative to the revenue you are protecting. If they are in the same segment as new prospects, the lower price becomes a visible reference point and can erode the new price. For a structured way to map this, the Business Model Canvas approach of separating customer segments and revenue streams is a useful scaffold (Wikipedia).

Once the model exists, keep it alive. Update new accounts per month, churn and variable cost monthly, and rerun the sensitivity table every time you get a new batch of pricing interviews. Ten interviews is a starting sample, and the model is a living document, not a one-time deliverable, which is the same reason business plans are treated as updatable working documents rather than static reports (Indeed).

If the arithmetic is the easy part and the hard part is building the product or app the model describes, that is the point where a conversation with Quanta (https://quantaintelligence.ai/) is useful: describe the site, app, document or plan you need, and a specialist picks it up and builds it with you.