Pricing Strategy and Business Model Questions
Setting and evolving prices and the monetization model they sit in. Covers pricing metrics and structure (per-seat, tiered, usage-based, hybrid, outcome-based), packaging and bundling, freemium versus paid versus ad-supported monetization, and migrating between pricing models, including from perpetual licences to subscription and consumption billing. Includes estimating willingness to pay and price elasticity (van Westendorp, Gabor-Granger, conjoint, observational and causal methods), designing and analysing price tests and reporting their results, discount and deal-desk policy, regional and personalized pricing, and quantifying the revenue, churn, LTV and margin impact of a price or packaging change. Also covers fee design and surge pricing on two-sided marketplaces, responding to competitor price cuts, and rolling out algorithmic pricing safely. Assesses commercial judgment and quantitative rigor on pricing decisions from product, sales, finance and analytics angles. Business-model archetype selection, unit-economics definitions and go-to-market execution are adjacent topics, not the focus.
Your company sells perpetual licenses today and wants to move to subscription, and later to consumption pricing. Model the financial impact over five years: revenue recognition differences, cash flow timing, churn and valuation effects. Then propose a migration plan with milestones for the CFO.
Sample Answer
Direct answer
Moving from perpetual licences to subscription shrinks reported revenue and profit for a few years before it overtakes the old model, because a perpetual licence is recognised up front while a subscription is recognised over time. I would model both paths over five years, show the CFO the dip and the crossover (in the model below, annual revenue crosses in year 6 and cumulative revenue in year 11), and migrate in stages: new customers first, the installed base (existing customers) second, consumption pricing last and only once usage is metered (counted and recorded per customer) and billing is proven.
Terms
- Perpetual licence: customer pays once to use the software indefinitely, plus annual maintenance (support and updates).
- Subscription: customer pays a recurring fee for access during the term.
- Consumption pricing: the customer pays for measured usage.
- ARR (annual recurring revenue): yearly value of recurring contracts.
- Churn: customers or revenue lost each period.
- Bookings: the contract value signed in a period, whether or not it has been recognised as revenue yet.
- Win rate: the share of sales opportunities that close as a win.
- Committed minimum plus overage: the customer commits to a minimum monthly spend and pays extra for usage above it.
- Dual-pricing rules: rules for how sales reps are paid when both the old and the new price options are on sale.
- J-curve: revenue dips first, then climbs past where it started.
Revenue recognition and cash flow
Under revenue accounting rules (ASC 606 in the US, IFRS 15 internationally; the rules for when a sale counts as revenue), a licence to use software as it exists when delivered is generally recognised when control passes to the customer (when the customer can use the software and benefit from it), while a hosted service or a right to updates is recognised over the term. Which treatment applies to your product depends on its contract terms, so the finance team decides it. Planning consequence: a $100,000 perpetual sale lands mostly in one quarter; a $40,000 a year subscription lands $10,000 a quarter. Cash follows billing, not recognition: annual billing in advance gives cash early and a deferred-revenue balance (cash received for service not yet delivered); monthly billing delays cash and makes the early dip deeper. Example: 100 new customers sign on day one at $40,000 a year. At the end of quarter 1, annual billing has collected $4.0M while monthly billing has collected $1.0M, yet recognised revenue is $1.0M under both ($4.0M / 4). The $3.0M difference is deferred revenue in the annual case and cash not yet received in the monthly case.
Five-year model
Illustrative company: 100 new customers a year, perpetual price $100,000 with 20% annual maintenance ($20,000), subscription price $40,000 a year, 400 customers on maintenance at the start, 90% of customers renewing each year. The customer's own total cost breaks even at five years ($100,000 + $20,000 x 5 equals $40,000 x 5), so the price is not a discount in disguise (that holds if maintenance is paid in the year of sale; in the model below it starts the year after the sale, so over five years a perpetual customer pays $100,000 + $20,000 x 4 = $180,000 against $200,000, and the subscription costs the customer about 10% more over that horizon). Customers sign on day one of each year and pay annually in advance, so revenue and cash coincide within a year; this is a simplification.
# Illustrative company. All figures in $ millions unless stated.
NEW_PER_YEAR = 100 # new customers signed at the start of each year
LICENCE = 0.100 # perpetual licence, $100k
MAINT_RATE = 0.20 # maintenance = 20% of licence per year ($20k)
SUB_PRICE = 0.040 # subscription, $40k per year, billed annually in advance
RETAIN = 0.90 # share of customers renewing each year (maintenance and, in the base case, subscription)
LEGACY_BASE = 400 # customers already paying maintenance at the start of year 1
def run(transition, sub_retain=RETAIN):
maint_cust, sub_cust = LEGACY_BASE, 0.0
prev_new_perp = 0.0
rows = []
for year in range(1, 6):
maint_cust = RETAIN * maint_cust + prev_new_perp # last year's new licensees start paying maintenance
sub_cust = sub_retain * sub_cust
if transition:
sub_cust += NEW_PER_YEAR
licence_rev, prev_new_perp = 0.0, 0.0
else:
licence_rev, prev_new_perp = NEW_PER_YEAR * LICENCE, NEW_PER_YEAR
maint_rev = maint_cust * LICENCE * MAINT_RATE
sub_rev = sub_cust * SUB_PRICE
# Annual billing in advance and signing on day one: revenue and cash coincide per year here
rows.append((year, licence_rev, maint_rev, sub_rev, licence_rev + maint_rev + sub_rev, sub_rev + maint_rev))
return rows
for name, flag in [("Stay perpetual", False), ("Move new sales to subscription", True)]:
print(name)
total = 0
for y, lic, m, s, tot, recurring in run(flag):
total += tot
print(f" Y{y}: licence {lic:5.1f} maintenance {m:5.1f} subscription {s:5.1f} total {tot:5.1f} recurring {recurring:5.1f}")
print(f" 5-year total {total:.1f}")
for sub_retain in (0.85, 0.80):
rows = run(True, sub_retain)
print(f"Subscription retention {sub_retain:.0%}: Y5 total {rows[-1][4]:.1f}, 5-year total {sum(r[4] for r in rows):.1f}")
Output:
Stay perpetual
Y1: licence 10.0 maintenance 7.2 subscription 0.0 total 17.2 recurring 7.2
Y2: licence 10.0 maintenance 8.5 subscription 0.0 total 18.5 recurring 8.5
Y3: licence 10.0 maintenance 9.6 subscription 0.0 total 19.6 recurring 9.6
Y4: licence 10.0 maintenance 10.7 subscription 0.0 total 20.7 recurring 10.7
Y5: licence 10.0 maintenance 11.6 subscription 0.0 total 21.6 recurring 11.6
5-year total 97.6
Move new sales to subscription
Y1: licence 0.0 maintenance 7.2 subscription 4.0 total 11.2 recurring 11.2
Y2: licence 0.0 maintenance 6.5 subscription 7.6 total 14.1 recurring 14.1
Y3: licence 0.0 maintenance 5.8 subscription 10.8 total 16.7 recurring 16.7
Y4: licence 0.0 maintenance 5.2 subscription 13.8 total 19.0 recurring 19.0
Y5: licence 0.0 maintenance 4.7 subscription 16.4 total 21.1 recurring 21.1
5-year total 82.1
Subscription retention 85%: Y5 total 19.6, 5-year total 78.8
Subscription retention 80%: Y5 total 18.2, 5-year total 75.7
How the maintenance column is built: each year maintenance customers equal 90% of last year's plus last year's new perpetual licensees, each paying 20% of $100k = $20k. Staying perpetual: year 1 is 0.9 x 400 = 360 customers, x $20k = $7.2M; year 2 is 0.9 x 360 + 100 = 424, x $20k = $8.5M; year 3 is 0.9 x 424 + 100 = 481.6, x $20k = $9.6M. Moving new sales to subscription, nobody new joins maintenance, so it simply decays at 90% (360, 324, 291.6 customers: $7.2M, $6.5M, $5.8M), while subscription customers go 100, then 0.9 x 100 + 100 = 190 ($7.6M), and so on. The model shows the effect of moving new sales only; it does not include converting the existing maintenance customers, which is the quarters 3-4 step in the migration plan and would need a conversion rate and a credit for the licence already paid.
What it shows, in $ millions:
- Revenue dip: year 1 is 11.2 against 17.2 (6.0 lower, about 35%). Five-year total is 82.1 against 97.6, a gap of 15.5.
- Crossover: there is no crossover inside the five years. By year 5 the new model is at 21.1 against 21.6, still 0.5 below, although the gap has narrowed from 6.0 in year 1 and the new model is growing faster because each new customer keeps paying. Extending the same model, annual revenue crosses in year 6 (23.0 against 22.4), and the cumulative revenue gap of 15.5 at year 5 is not recovered until year 11 (cumulative 242.4 against 242.2; at year 10 it is still 212.4 against 216.6). Show the CFO both dates: annual crossover in year 6, cumulative payback in year 11.
- Quality of revenue: in year 5 all 21.1 is recurring; in the old model 11.6 of 21.6 is recurring (about 54%). Recurring revenue is more predictable and is what ARR-focused investors value most highly.
- Churn sensitivity: the base case assumes subscribers renew at 90%. Subscriptions give customers an exit at every renewal, so churn could be worse. At 85% retention the five-year total falls to 78.8 and the year-5 revenue to 19.6; at 80%, 75.7 and 18.2. Retention is the assumption the whole case is most sensitive to.
Valuation effects
Reported growth and profit fall in the transition years even though the business gets healthier, so lead with ARR, retention, billings (amounts invoiced in the period) and cash conversion, not only reported revenue under GAAP (generally accepted accounting principles). How the market rewards this is not something I would quote a multiple for; I would say plainly that the story depends on investors accepting recurring revenue and retention as the primary lens. A valuation multiple is the number an investor multiplies a metric by to value the company. Illustration with placeholder multiples, not market figures: if recurring revenue is valued at 6x and one-off licence revenue at 2x, year 1 is worth 7.2 x 6 + 10.0 x 2 = $63.2M under the old model and 11.2 x 6 = $67.2M under the new one, so the value can rise while reported revenue falls. If investors apply the same 4x to all revenue, year 1 is 17.2 x 4 = $68.8M against 11.2 x 4 = $44.8M, and the dip shows up in full.
Migration plan with milestones for the CFO
| When | Milestone | Gate to proceed |
|---|---|---|
| Quarter 0 | Pricing equivalence model, accounting review, board alignment, dual-pricing rules for sales compensation | Finance signs off treatment and the dip in the forecast |
| Quarters 1-2 | New customers only on subscription; perpetual still available for named strategic cases | Win rate and average price within agreed range of perpetual |
| Quarters 3-4 | Offer installed base conversion at renewal of maintenance, with credit for the licence already paid | Conversion take-up and no spike in churn |
| Year 2 | Stop selling new perpetual licences; instrument and meter usage | Billing accuracy and metering reconciled to customers' own records |
| Years 3-4 | Introduce consumption on a committed-minimum plus overage design for new deals and renewals | Usage forecast error within tolerance, invoice disputes low |
| Year 5 | Review: consumption as default for qualifying products | Revenue predictability and net revenue retention (revenue kept from existing customers including upsell) at or above subscription |
For consumption, the revenue forecast must model per-customer usage distribution and invoice timing, which is why the committed minimum is the bridge: it keeps some predictability while customers learn the new bill.
Trade-offs and pitfalls
- Sales compensation is the biggest hidden risk: reps paid on bookings will resist a smaller first-year deal, so redesign commission before launch.
- Moving the installed base too early invites customers to reopen the whole contract.
- Do not state the dip as a surprise to the board; show it as a planned J-curve with a measured exit.
- What would change my call: if churn risk after conversion is high, delay base conversion and rely on new-customer subscription only.
A recurring theme from customer feedback is that pricing complexity prevents expansion discussions. Propose a simplification approach for pricing and packaging that preserves revenue leverage for expansion, including analysis steps, stakeholder alignment, communication plan to existing customers, and metrics to monitor post-change.
Sample Answer
Direct answer
Simplify the structure, not the revenue logic: collapse the price list to a few packages with one clear expansion path, keep customers' existing prices unchanged at their first renewal after the switch, and measure whether expansion conversations get easier and expansion revenue holds. Complexity (many plans, add-ons and one-off exceptions) hurts expansion when a customer cannot tell what the next step costs or a rep needs approval to quote it. Revenue leverage for expansion comes from a ladder (an ordered set of packages, each a clear step up in price and capability) with clear steps (seats or usage bands, tier-fenced features, meaning features available only in higher tiers, and add-ons), not from the number of SKUs (stock-keeping units, the individual items on the price list).
Analysis steps
- Inventory: list every plan, add-on and active custom price, with customer count and ARR (annual recurring revenue) for each. Typically a few items hold most of the ARR and a long tail holds the confusion.
- Quantify the pain: quote turnaround time, share of deals needing deal-desk approval (the group that approves non-standard prices), discount depth, expansion opportunity-to-close rate and cycle time split by how many SKUs the customer holds, plus billing-related support tickets and win/loss notes mentioning price.
- Find what expands: which plans, add-ons and thresholds actually precede upgrades (usage hitting a limit, seat growth), so the new ladder preserves those triggers.
- Design: three packages and one add-on set, one value metric (the unit the price scales with, for example seats or usage), published step prices, and a rule for exceptions. An illustrative ladder: Team $3,000 a year for up to 10 seats; Business $7,200 for up to 30 seats with reporting; Scale $17,000 for up to 100 seats with advanced analytics included; each extra block of 10 seats on any tier is priced at a published step (illustratively $1,500 per block, the figure used in the customer wording below), so a rep can quote the next rung without approval.
- Simulate the migration customer by customer: old price against new price, flag who pays more or less, and total the ARR change.
Example artifact: mapping legacy plans (illustrative)
| Legacy plan | Customers | Average ARR | ARR total | New package | Price at first renewal |
|---|---|---|---|---|---|
| Basic | 400 | $3,000 | $1,200,000 | Team | $3,000 (unchanged) |
| Standard | 350 | $7,200 | $2,520,000 | Business | $7,200 (unchanged) |
| Pro | 250 | $15,000 | $3,750,000 | Scale | $15,000 (unchanged) |
| Pro plus Analytics bundle | 120 | $19,000 | $2,280,000 | Scale with Analytics included | $19,000 (unchanged) |
| Enterprise custom | 80 | $48,000 | $3,840,000 | Enterprise (custom, kept) | Unchanged |
| Total | 1,200 | $13,590,000 | $13,590,000 |
New list prices for new customers: Team $3,000, Business $7,200, Scale $17,000 (Analytics included), Enterprise by quote. The rule for existing customers is hold-harmless at first renewal: nobody pays more at migration and nobody is moved to a lower price, so the migration is ARR-neutral by construction ($13,590,000 before and after) and the change is purely structural. The new list prices apply to new customers and to any expansion after migration. Anyone whose legacy price sits below the new list price closes the gap only through added value at a later renewal, capped (for example at 10% per renewal). Worked example: a legacy Pro customer, now on Scale at $15,000, has a $2,000 gap to the $17,000 list and closes it only when it expands; the 10% cap allows $1,500 at the next renewal ($16,500) and the remaining $500 at the one after, reaching $17,000. The 120 customers on Pro plus Analytics pay $19,000, above the $17,000 list that now includes Analytics. They stay at $19,000 because there are no automatic decreases, and they are the group to watch for renewal risk. Any customer asking for a decrease goes through the deal desk.
Preserving expansion leverage
- Make the next step visible: a published step price for each seat band or tier so a rep can quote in minutes.
- Keep fences (limits that reserve a feature or capacity for a higher tier) on the features that drive upgrades; fewer SKUs must not mean unfenced features.
- Keep a usage or seat metric that grows with the customer, so expansion happens without a negotiation.
- Keep one exception path, owned by the deal desk, with a monthly review of exceptions so they do not rebuild the complexity.
Stakeholder alignment
| Group | Concern | Action |
|---|---|---|
| Sales | Compensation and quote speed | Check quota and commission rules against the new structure; training and a quoting guide |
| Customer success | Renewal risk, customer questions | Account-by-account migration list, talk track, FAQ |
| Finance | Revenue recognition (the rules for when billed amounts count as revenue), billing changes | Confirm the billing system can express the new packages; sign off ARR neutrality |
| Product | Fences and entitlements (the features and limits each package switches on for a customer) | Align feature gating with package definitions |
| Legal | Contract terms | Check that existing contracts allow the change |
| Support | Ticket volume | Pre-brief on likely questions |
Customer communication
Notice at least 60 to 90 days before the customer's renewal, sent by email and the account manager: what changes, that their price is protected through renewal, what the new options are, and who to contact. Illustrative wording: "We are simplifying our plans from five to four. Your current plan maps to Scale, and your price of $15,000 stays the same at your renewal on 1 March. Nothing changes in what you can use. If you later want more seats, the next step is $1,500 for each additional 10 seats, and your account manager can walk you through it." Offer a one-to-one call to the largest accounts. Never announce a simplification that is a price rise in disguise: customers see through it.
Metrics to monitor after the change
| Metric | Why | Compare against |
|---|---|---|
| Expansion rate and net revenue retention (NRR, revenue from the same customers one year later, including their expansion and losses) | Is expansion easier? | Pre-change cohort (a group of customers that shared the same starting point, such as the same renewal quarter) |
| Quote-to-close time and deal-desk request share | Is complexity falling? | Pre-change baseline |
| Average discount depth | Did discounting rise to compensate? | Baseline |
| Billing and pricing support tickets | Confusion | Baseline |
| Renewal and churn in migrated accounts | Customer reaction | Accounts migrated later (staggered rollout: moving accounts in waves, not all at once) |
| ARR bridge: new, expansion, contraction (customers paying less), churn | Revenue neutrality: the bridge walks from opening to closing ARR through these four movements | Forecast |
Rollout in waves, so the unmigrated accounts act as a comparison group, and a problem in the first wave is caught before everyone is affected. Stop condition: pause if churn or contraction in migrated accounts exceeds a threshold set before launch.
Pitfall: deleting add-ons that customers actually use, which shows up as contraction. Check usage before removing anything.
You run a freemium product and must decide how generous the free tier should be. How do you trade rapid user growth against near-term revenue, where do you put the limits, and how do you test a change without harming existing users?
Sample Answer
Direct answer
Put the free-tier limit on a value metric (the unit the limit is counted in, such as projects or seats) that bites only after the user has experienced the product's core value (the point where leaving would hurt), and bite hardest on the segments most able to pay, such as teams and commercial users, rather than on individual hobbyists who generate referrals. Model it as a funnel with a referral loop, treat each stage's response to the limit as an elasticity assumption (how much an outcome moves when the limit moves) to be measured, and test any change on new signups while grandfathering existing users (keeping them on their current terms).
Funnel model (illustrative inputs, 100,000 new signups)
For example, with projects as the value metric, "tight" might be 3 free projects, "medium" 5 and "generous" 10 (illustrative): a tight cap blocks some people before they reach the core value, so activation is lower, but more of those who do activate hit the cap. Stages: signup to activated (reached the core value, a), activated users who hit the limit (h), hitters who convert to paid (c), and a referral loop (a viral loop, where existing users invite new ones) in which each activated user brings r new signups. Paid price $8 per month, kept for 12 months (ignoring churn, for simplicity). Referral loop multiplier: 1 / (1 - a x r), summing all generations.
| Limit | a | h | c | r | Payers per 100k signups | 12-month revenue per signup (no referrals) | With referral loop |
|---|---|---|---|---|---|---|---|
| Tight | 0.30 | 0.45 | 0.10 | 0.05 | 1,350 | $1.296 | $1.316 (x1.0152) |
| Medium | 0.40 | 0.30 | 0.08 | 0.10 | 960 | $0.922 | $0.960 (x1.0417) |
| Generous | 0.50 | 0.15 | 0.07 | 0.18 | 525 | $0.504 | $0.554 (x1.0989) |
Worked line, tight: 100,000 x 0.30 x 0.45 x 0.10 = 1,350 payers; 1,350 x $8 x 12 = $129,600 = $1.296 per signup; multiplier 1 / (1 - 0.30 x 0.05) = 1.0152; $1.296 x 1.0152 = $1.316. Why the multiplier works: 100 signups produce 100 x 0.30 = 30 activated users, who bring 30 x 0.05 = 1.5 new signups; those 1.5 bring 1.5 x 0.30 x 0.05 = 0.0225 more, and so on. The total is 100 x (1 + 0.015 + 0.015^2 + ...) = 100 x 1 / (1 - 0.015) = 101.52 signups, a factor of 1.0152.
On these inputs the tight limit wins. For the generous tier to equal it, each activated user would need to refer about 1.2 new signups, a very strong loop. The working: matching the tight row's $1.296 per signup needs a multiplier of 1.296 / 0.504 = 2.571 on the generous row, so 1 - 0.50 x r = 0.389 and r = 1.22 (1.23 if the tight tier's own small loop is also counted). That is the central trade-off: generosity is only worth it if free users are a growth asset (viral loop, network effect, meaning the product gets more useful as more people join, or useful data) and cheap to serve; otherwise it gives value away.
Where the elasticity assumptions come from
The parameters a, h, c and r are the stage rates you must estimate at each candidate limit, and their elasticities describe how each rate moves when the limit moves (an elasticity is the percentage change in an outcome for a percentage change in its driver; in the table, h falls from 0.45 to 0.15 as the limit goes from tight to generous, and from tight (3 projects) to medium (5 projects) it falls 33% while the limit rises 67%, an elasticity of about -0.5). Estimate them from:
- the usage distribution from logs (what share of free users reach 5, 10, 20 uses a month tells you h at each candidate limit);
- past limit changes or accidental variation (an outage, a regional pilot);
- an experiment on new signups.
Segmentation
Do not use one limit for everyone. A hobbyist and a ten-person team hit the same cap at different value to them. Prefer limits on features that scale with business use (seats, shared workspaces, history length, exports, integrations) over limits on the core action that every user needs to experience.
Testing without harming existing users
- Randomise new signups only at the user level (by a hashed ID, a scrambled but repeatable code derived from the user ID so the same user always lands in the same arm), keeping existing users on the current terms. Never lower an existing user's free entitlements as a test arm. Loosening limits for a new cohort is low risk.
- Conversion lags, so read activation and limit-hit rate in weeks 1 to 4 and conversion at 60 to 90 days; do not stop at the first dip.
- If you must tighten for existing users, announce, grandfather for a period, and stage by cohort.
Post-change KPIs
Activation rate, share hitting the limit, free-to-paid conversion by cohort age (cohort: users who signed up in the same period), revenue per signup at 12 months, free-user retention, referrals per active user, paid churn (people forced to pay may leave sooner), support tickets, and cost to serve per free user (what it costs to run the product for one free user) compared with budget.
Pitfalls
- Optimising near-term revenue by tightening, then discovering the referral loop was the growth engine.
- Ignoring that free users cost money (compute, storage, support).
- Reading a one-month result for a decision whose payoff arrives later.
You plan to change your pricing tiers and want evidence on willingness to pay first. How would you choose research methods, turn the findings into a rollout, and limit churn risk after launch?
Sample Answer
Direct answer
I would combine two or three methods that answer different questions, never rely on one survey, and treat the research as a way to pick a hypothesis that a controlled rollout then proves. Method choice follows what I already know: for a price range and relative value, a price-sensitivity survey; for trade-offs between tiers and features, conjoint; for behaviour, a price test on new customers and analysis of what existing customers already do. The rollout is staged (new customers first, existing customers later, with a holdout: a randomly chosen group kept on the old price as the comparison), and churn risk is limited with guardrail metrics (measures that must not get worse, such as cancellations) agreed before launch, grandfathering (existing customers keep their old price) and notice periods.
Step 1: choose methods by the question
| Question | Method | What it gives | Main weakness |
|---|---|---|---|
| What price range feels acceptable? | Van Westendorp price sensitivity meter: four questions (price so high you would not buy, getting expensive, a bargain, so low you doubt quality) | Plotted curves give an acceptable range and an optimal point | Ignores competitors, no demand curve |
| How many would buy at each price? | Gabor-Granger: ask "would you buy at $X?" at a sequence of prices | A demand curve and a revenue-maximising price | Stated intent overstates buying |
| How do they trade off price against features and tiers? | Choice-based conjoint (survey where people choose between bundles) | Utilities, WTP per feature, share simulation | Needs design and sample size |
| What do customers actually do? | Live test on new signups, plus history: discounts, win/loss, upgrade and downgrade patterns | Revealed behaviour | Slow, limited price range |
| Which segments differ? | Usage data cut by segment (usage of the value metric, the unit you charge per, plus seats and industry) | Where value is captured | Correlation, not price response |
Willingness to pay (WTP) is the most a customer will pay for something. My default for a tier change: conjoint or Gabor-Granger for the shape, a live test for the truth, and usage data to decide which segments to treat differently.
Two survey methods in numbers (illustrative)
Van Westendorp. Five respondents each name four prices (too cheap, bargain, expensive, too expensive): (10, 15, 25, 35), (12, 18, 28, 40), (14, 20, 30, 42), (16, 22, 32, 45), (18, 25, 35, 50). At $18, 20% call $18 or more too cheap and 20% do not call it a bargain, so the lower curves meet and the lower edge of the acceptable range is $18. The share calling a price too expensive passes the share who do not call it expensive between $34 and $36 (0% against 20%, then 20% against 0%), so the upper edge is about $35.
Gabor-Granger. Ask each person whether they would buy at a sequence of prices and read off the share saying yes. Say 50% at $30, 46% at $35, 38% at $40 and 26% at $45. Price x share, which is revenue per person shown the price (multiply by 100 for per 100 people), gives $15.00, $16.10, $15.20 and $11.70, so $35 maximises revenue. Stated intent overstates real buying, so use the shape (where revenue peaks and how steeply it falls) and confirm the level with a live test.
Step 2: turn findings into a rollout
- Pick the decision first. Write down which of price level, tier boundaries, or the value metric (what you charge per, for example seats) is changing, because a study is easiest to read when it targets one of those at a time (a conjoint can estimate price and tier trade-offs together, but the rollout should still change one lever at a time so the result can be attributed).
- Choose candidate packages (two or three) from the research and the simulator, each with a hypothesis ("moving advanced reporting to the top tier lifts top-tier share without hurting conversion").
- Test on new customers first. Randomise new signups to the old or new pricing page and measure conversion, plan mix, and 90-day retention. New customers have no expectations to break and no contract to renegotiate.
- Roll out to existing customers in waves, lowest-risk segments first, with a randomly chosen holdout that stays on old pricing so you can measure churn caused by the change (not by the season).
- Give customers a reason and a runway: advance notice, a locked-in price for those who renew annually, and a clear explanation of what they get.
Step 3: limit churn risk after launch
- Define guardrails before launch: cancellation rate, downgrade rate, support tickets about price, and net revenue retention (revenue from existing customers after churn and expansion) in treated vs holdout accounts. State the stop rule in advance (for example, pause the wave if monthly cancellations in treated accounts, the ones moved to the new price, are more than 1 percentage point above the holdout after one billing cycle, say 3.0% against 2.0%; set the margin from the normal month-to-month noise in the baseline).
- Grandfather (keep on old terms) the accounts where the loss risk exceeds the gain.
- Offer a downgrade path so price-sensitive customers shrink instead of leaving.
- Have the account teams contact the largest accounts personally before notice goes out.
Worked example: how much churn can a price rise absorb
Take a plan priced at $100 per month with $30 per month cost to serve (so 70% gross margin). Raise it 15% to $115. Revenue is flat only if retention after the rise is 1 / 1.15 = 87.0% (13.0% can leave). Gross profit is flat when 85 x r = 70, so r = 82.4% (17.6% can leave), because the extra $15 is pure margin while cost per account is unchanged. So the break-even churn tolerance is looser in profit terms than in revenue terms. For a 10% rise the same calculation gives 90.9% (revenue) and 87.5% (gross profit). I would use this table to set the guardrail: if the test shows incremental churn near those break-evens, stop. This assumes the cost to serve a departing customer disappears; if it is fixed cost the profit tolerance tightens.
Trade-offs and pitfalls
- A survey says what people say; a test says what they do. When they disagree, believe the test, but check whether the test lasted long enough to show renewals.
- Raising prices on everyone at once removes your control group.
- Existing customers react to the change in price relative to what they had (loss aversion: a loss feels larger than an equal gain), not to the absolute price, so findings from prospects do not transfer fully.
- Looking only at churn misses downgrade and contraction (customers buying fewer seats or a lower plan); track net revenue retention.
- Beware a price test that exposes the same buyer to two prices (they can see both), which creates trust damage; randomise (assign each account or visitor to one price by chance and keep it there) at the account or visitor level and keep the page consistent.
Write the pricing section of a PR/FAQ for a new B2B analytics add-on. Offer at least three options, say who each is for and why the tiers exist, and list the metrics you would track to validate the pricing after launch.
Sample Answer
Direct answer
This is the pricing section of the press release and FAQ for a new B2B analytics add-on, "Analytics Plus", sold on top of an existing workflow product. It offers three plans, says who each is for and why the tiers exist, and closes with the eight metrics we would watch after launch. All prices are illustrative.
Pricing section (excerpt)
Pricing, in the customer's words. "Analytics Plus is priced per workspace (one team's shared account), per month, so you pay for the team that uses it, not for every person who views a dashboard. Start with the plan that matches your team today and move up when your reporting grows. No setup fee."
| Starter | Growth | Enterprise | |
|---|---|---|---|
| Price (illustrative) | $300 per workspace per month | $900 per workspace per month | From $2,500 per workspace per month, annual contract |
| Who it is for | Teams of up to 25 that want standard dashboards without hiring an analyst | Teams that define their own metrics and share reports across departments | Companies with security review, many teams and a data team that builds on the product |
| What is included | Prebuilt dashboards, 3 data sources, 12 months of history, unlimited viewers | Custom metrics, scheduled reports, 10 data sources, 36 months of history | Single sign-on (SSO, one company login for all tools), row-level access rules (controls on which rows of data each user may see), API access, dedicated support, unlimited history |
| Why the plan exists | Low-risk entry so attach (the share of eligible accounts that buy) is not blocked by price | The plan most teams grow into; the fences (the limits that separate one plan from the next, such as data sources or history) are things that need more compute (server processing capacity) and support | Features that buyers with compliance needs require and will pay for, priced by contract |
Why three plans. Each step up is gated by something that costs us more to serve (more data sources, longer history, support) or that matters more to a larger buyer (access control, API). We avoid fences customers would see as arbitrary, such as hiding basic chart types.
FAQ: Which plan should I choose? If your team uses standard views and has under 25 people, Starter. If you build your own metrics or send reports outside the team, Growth. If a security team has to approve the tool, Enterprise.
FAQ: What happens if I outgrow a plan? You are prompted to upgrade; nothing is switched off mid-month, and the upgrade is prorated.
FAQ: Can I try before buying? Yes: 30 days on Growth, on your own data, then you choose a plan.
How we will validate the pricing after launch
Eight metrics, each with the decision it informs:
- Attach rate: share of eligible accounts that buy within 90 days. Low attach with healthy trial use points at price or packaging, not at the product.
- Trial-to-paid conversion: by plan and segment.
- Plan mix: share of buyers on each plan. If nearly all land on Starter and few upgrade, the Growth fences are too weak or too strong.
- Upgrade and downgrade rate: movement between plans in the first two quarters.
- Discount depth: how far below list price deals close, as an average discount and the share of deals discounted. Heavy discounting on Enterprise means the list price is out of line with what buyers will pay.
- Gross margin per plan: price minus compute and support cost per workspace. If Growth heavy users cost more than they pay, the data-source or history limit needs tightening. Worked check (illustrative): a typical Growth workspace costs $180 of compute and $120 of support, so $300 against a $900 price leaves $600 (66.7%). A heavy one refreshing all 10 sources hourly costs $620 of compute and $200 of support, $820, leaving $80 (8.9%). If a tenth of Growth workspaces look like that, tighten the refresh limit.
- Add-on revenue retention: how much of month-one add-on revenue is still there after 90 days and 12 months.
- Time to first shared dashboard: a leading indicator of whether a trial turns into a lasting habit.
Before launch I would write down the thresholds that trigger a change, for example "if fewer than 1 in 20 Starter accounts move up to Growth within two quarters, revisit the Starter limits", so that the review is a decision and not a debate. Example reading: after two quarters, 14 of 400 Starter accounts have moved to Growth, which is 3.5%, below the 1 in 20 (5%) trigger, so we first check whether the Starter limits (3 sources, 12 months) are too generous, and only then touch prices. The thresholds are ours to set from early data, not industry figures.
Trade-offs and pitfalls
- Per-workspace pricing is simple but a very large workspace pays the same as a small one; if margin per plan shows that, add a usage ceiling to Growth rather than a new plan.
- Three plans is the minimum to create a visible middle option; more plans raise the cost of choosing and of supporting them.
- Packaging that gives away what the sales team uses to justify Enterprise (SSO, access control) weakens the top plan, so those are held back deliberately.
Unlock Full Question Bank
Get access to all 32 Pricing Strategy and Business Model interview questions and detailed answers.
Sign in to ContinueJoin thousands of developers preparing for their dream job.