How to Price and Package a New Product Before Launch
Your new product or module ships next quarter, sales wants a number for the deck, and the board wants a revenue line in the plan. Nobody has a value metric yet, and whatever number goes out first becomes the reference price your P&L defends for the next two years.
Price Before You Build, Not After
The first price a market sees becomes the price it judges every later price against. Kahneman's Thinking, Fast and Slow describes the mechanism: an initial number pulls every later estimate toward it. Raise your launch price later and buyers read a price increase. Cut it and they read an admission.
Ramanujam and Tacke built Monetizing Innovation around one instruction: design the product around the price, not the price around the product. They open with the Simon-Kucher research finding that most new products miss their revenue or profit targets or fail outright. The launch price is the only pricing decision you make without churn risk.
Harrowgate Compliance, $68M ARR, hospital vendor-risk software. It launched a continuous-monitoring module at $4,800 a year because sales needed a number for a pipeline review. The audit-prep hours it removed supported three times that. Eighteen months later Harrowgate moved list to $12,000: existing customers negotiated grandfathering or left, and new prospects had already heard $4,800 on reference calls. The repricing cost more in churn than the placeholder had collected.
Do the math for your own business. Take the annual value the module creates for one customer, multiply by the share you could reasonably charge and by the customers you expect in year one, then subtract what the placeholder price will collect. That gap is real money, your money. McKinsey's The Power of Pricing puts a 1 percent price improvement at roughly an 11 percent lift in operating profit, and a placeholder launch price can miss value by far more than 1 percent.
Pricing is a signal before it is a number: the launch price tells the market which category you are in and whether you believe the product's value. Set it in this order.
Segment
Which buyer does the launch serve, and what job does that buyer hire the product to do? Moesta's Demand-Side Sales 101 calls it the struggling moment.
Value metric
What does the customer get more of when they get more value? That unit is what you charge for.
Packaging
What goes in the base edition, what fences the upgrade, and what stays an add-on.
Price level
The number, from value-in-use math and a cohort test at launch, because the benchmark does not exist yet.
Offer
The trial, intro price, or bundle that lands the first customers, with a success metric written before launch.
Governance
The exception policy, discount authority, and deal desk rules that protect the price after launch.
Product and pricing decisions are one decision made by different people, the argument of Every Feature Decision Is a Pricing Decision. The launch is also the only window to test the price on real buyers without a repricing to sell afterward; how to run that test follows the price level.
Find the Value Metric
The value metric is the unit your price is denominated in: per seat, per record, per monitored asset, per dollar processed. Nagle and Müller, in The Strategy and Tactics of Pricing, treat the price metric as a structural choice that comes before the price level, because it decides who pays more and why. The test is one sentence: what does the customer get more of when they get more value? If that is not what you meter, price and value drift apart. Four tests.
Grows with value
When the customer gets more from the product, the metric goes up. Per-admin-seat pricing fails the moment one admin protects 500 locations.
Predictable for the buyer
The buyer can forecast the invoice before signing and at renewal. A metric the buyer cannot budget against is a procurement objection.
Measurable without a dispute
You can meter it, the customer can see the meter, and finance on both sides agrees.
Defensible
A competitor cannot undercut you with a metric that hides the same cost, and customers cannot game it by sharing logins or batching usage.
Seats are the default because they are easy to count and forecast, and when value comes from people using the product they are right. They are wrong when value scales with something other than headcount. The a16z rule of thumb is a useful check: subscription pricing when the end user is a human, usage-based pricing when the end user is software.
Usage-based pricing creates predictable revenue problems as often as it solves them. The buyer cannot forecast the invoice, so procurement asks for a cap, and the cap becomes the price. Revenue falls when the customer gets efficient. Any metric the customer cannot see on a dashboard becomes a dispute at renewal. a16z's own guidance says the same: customers want predictability, and overages are an overlooked problem.
The usual answer is a hybrid: a platform fee, included volume the buyer can plan around, and an overage rate that rarely fires. High Alpha and OpenView's SaaS Benchmarks report that 86 percent of SaaS companies above $100M ARR price on at least three dimensions. The worked example, including the per-truck plus take-rate structure that beat pure usage pricing, is Per-Truck + 2.4% Take Rate.
Ostrander Freight Systems, $110M ARR, launched a route optimization module at $1,200 per dispatcher seat per year. Depots ran three dispatchers whether they managed 40 trucks or 400, so every depot paid $3,600 a year. Ostrander moved to $9 per active truck per month with a $2,400 annual depot minimum. The 40-truck depot paid $4,320 a year. The 400-truck depot paid $43,200 a year.
Package Before You Price
Packaging beats pricing. Packaging decides which tier a buyer sees themselves in and whether they upgrade without a sales call. For a new product it decides whether the buyer understands what they are buying at all. Confusion is the enemy of willingness to pay, and a launch is when confusion is cheapest to create.
Good-better-best works for a new product when each tier is a complete answer for a different buyer. The base tier does the whole job for the narrowest segment you are launching to. The Better tier does the same job for the buyer with more at stake, separated by a fence. Nagle and Müller describe fences as conditions that let buyers sort themselves: a need that tracks willingness to pay and that the buyer already knows about, such as entities or regulated workflows. The Best tier is the executive outcome: audit-ready reporting, consolidation across business units, a contractual SLA. Anything only a minority values highly stays an add-on, because folding it into a tier raises the price for everyone else.
Ramanujam and Tacke give you the sorting tool: classify each capability as a leader, a filler, or a killer for each segment. Leaders are what the segment pays for. Fillers cost nothing to include. Killers lower willingness to pay for the whole package because they signal the product was built for someone else. Then group what survives by the job it does, not by how hard it was to build: capabilities that serve the same use case belong in the same tier, and a tier serving three use cases is three packages wearing one price. If the module ran as a beta, usage data shows where buyers already draw the lines: capabilities that co-occur in accounts that renew, and the cliff after which adoption stops. Those clusters are your tier boundaries. The SaaS packaging framework paper took one company from 14 SKUs to four editions this way.
Before the tier page goes live, run the 30-second test: put it in front of someone who has never seen the product and ask which tier they would buy. If they cannot say, or they answer in feature language instead of outcome language, rewrite it. Name tiers after what the buyer accomplishes; metals teach buyers to evaluate prestige, size labels teach them to evaluate headcount, and a name like Monitor, Remediate, Attest teaches them to evaluate the outcome. What does not work is renaming tiers and moving nothing between them. New labels on the same feature stacks change nothing, and buyers see through it on the first call.
Ashcombe Security, $35M ARR, raised its Professional tier from $18,000 to $22,000 and Enterprise from $45,000 to $52,000. Net revenue retention went from 98 to 97 percent and mid-tier expansion from 12 to 11 percent. The price had moved on the wrong package. Professional held 11 features serving three different use cases, so smaller buyers paid for two jobs they never touched and larger buyers could not tell it covered what they needed. Six months later Ashcombe split the use cases into two tiers with an outcome statement on each and left every price where it was. Mid-tier expansion reached 21 percent within two quarters. Same product. Same prices.
Over-packaging confuses a homogeneous market. If your launch buyers are alike, three tiers give one buyer three ways to hesitate. Schwartz's The Paradox of Choice describes the mechanism: more options raise the cost of choosing. If you cannot name a buyer attribute that predicts which tier a buyer will pick, you have one tier and two decoys. For a module sold into an existing base, one edition plus one add-on is often right; add tiers when usage shows the split. Treat every tier as a hypothesis with a number attached (the good-better-best pricing concept page covers the test and the buyer psychology), and see what skipping that costs in the $280K packaging project that reverted. If the same product serves a team edition and a solo edition, the packaging logic splits even when the codebase does not; that case is solo users converting at 14 percent, teams of five at 2.3.
Set the Price Level When There Is No Benchmark
Willingness-to-pay research fails in new categories because the benchmark does not exist yet. Survey methods ask buyers to place a known product on a price scale they already carry. A buyer who has never bought your category answers from the nearest one they have. Fitzpatrick's The Mom Test is blunt about it: hypothetical questions get hypothetical answers.
Four methods work when the benchmark does not. Run them in this order.
Value-in-use math
Nagle and Müller's economic value estimation: reference value plus differentiation value. Reference value is what the buyer's next best alternative costs, and there is always one: a spreadsheet, a headcount, a consultant, doing nothing. Differentiation value is what your product adds or removes.
Reference-price interviews
Dunford's Obviously Awesome starts positioning with the competitive alternative: what the buyer would do if you did not exist. Ask what they use today, what it costs, and which budget line pays for it. The budget line tells you who approves and what they compare against.
Sequential price tests
Launch in cohorts and change one variable per cohort: price level, then included volume, then the fence. The cohort design is in the next section.
Behavioral signals from existing usage
If the module ran as a beta or lived as a feature, you already have data: who used it, how deeply, what they would lose. Depth of use predicts willingness to pay better than any stated preference, and much of it is already in your CRM; the willingness to pay concept page covers how to read it.
Tallis Labs, $52M ARR, launched an audit-automation module for finance teams. The alternative was 1.5 analyst FTEs at $130,000 plus a $40,000 external review, about $235,000 a year. The module removed roughly 60 percent of the analyst hours and half the review scope, a differentiation value near $137,000. Tallis set list at $34,000 and tested it against $42,000 across two launch cohorts. Win rate held at the higher price, and it became list. See the value based pricing paper.
Van Westendorp asks four questions about a known product (too cheap to trust, a bargain, getting expensive, too expensive) and returns an acceptable range: cheap, fast, and worth running when your product extends a category buyers already buy. Conjoint asks buyers to choose between bundles at different prices and returns the value of each attribute: expensive, slow, and worth it when buyers trade several attributes against each other and a wrong tier structure would cost more than the study. The multi-instrument version is $148K of WTP Research Into $1.2M of Lift.
Run the Launch as a Pricing Experiment
A launch is the one moment buyers expect the price to be different. New product, new number, no anchor. A buyer who would fight a 6 percent increase on the core product will accept a metric they have never paid on, because there is nothing to compare it to. The window closes fast: once the first cohort talks on reference calls, the market has a norm.
Most launch prices get set in one meeting, checked with two friendly customers, and locked for 18 months. That forgoes the higher price the product could have carried and the learning about where willingness to pay sits, both cheapest to buy at launch. Four rules keep the test clean.
Split cohorts on a variable that is not price
Company size, acquisition channel, vertical, or region. Each cohort sees one price and no buyer sees two. If a buyer in the higher cohort finds the lower price, honor it and log it; the learning is worth more than the difference on one deal.
Change one variable per cohort
Price level first, then included volume on the value metric, then the fence between tiers. Two variables in one cohort give you a result you cannot act on.
Score on revenue per opportunity and retention
A cohort converting at 32 percent at $299 a month returns about $96 of monthly revenue per opportunity. A cohort converting at 24 percent at $450 returns $108, and it tends to hold the buyer who stays when the price moves. Track win rate, time to close, discount requests, and 60-day retention by cohort.
Write the decision criteria before the first deal
Which metric decides, what margin counts as a win, and how many deals or weeks the test holds. Then act on the result. A test that shows one price converting 12 percent better and then spends six months in debate was never a test.
The launch also makes two moves cheap that are expensive anywhere else: a new pricing unit, since nobody has paid for the module any other way, and premium capability as an add-on rather than a new top tier, since an add-on is additive and a tier makes the buyer re-shop the base contract.
Corvane Analytics, $47M ARR, launched a data visualization product at $199 a month, priced off a competitor's feature, with no test because the timeline was tight. Within six months, 400 accounts had adopted. Interviews at month six found enterprise accounts producing $300,000 to $500,000 a year of report output with it; $199 captured about 1 percent of that, roughly $1.2M a year left across the 400. The move to $450 cost 22 percent of the cohort, most of it the price-sensitive accounts the low price had attracted. Before: 400 accounts, $80K MRR. After: 312 accounts, $140K MRR. Two cohorts at $199 and $450 in month one would have found the same answer with no repricing to sell.
The failure that costs the most is timing. Pricing joins in the last week, when the deck is written and the first customer conversations have started, so there is no room to test. Bring pricing in at week eight of the checklist below, and put the 30, 60, and 90-day decision gates on the launch calendar with a named owner for each.
Design the Launch Offer With a Success Metric
Offer strategy without a success metric before launch is just cost. A trial with no target is free product. An introductory price with no expiry date is your list price. Write the metric first, then choose the offer that can hit it.
Free trial
Right when a buyer can reach the value moment inside the trial window. Bush's Product-Led Growth frames the choice around time to value: if the outcome takes longer than 14 days, a 14-day trial tests patience, not the product. Measure time to value from signup to first meaningful action, and set the trial at roughly twice it.
Freemium
Right when a free user costs you almost nothing and free usage produces an upgrade trigger the user hits on their own. Wrong for most B2B module launches into an existing base, because a free tier resets the reference price to zero. The design turns on who controls the budget, covered below.
Introductory pricing
Right when you need a launch cohort fast and can print the list price next to the intro price with the step-up date in the contract. Wrong when the intro price ships alone, because it becomes the reference and the step-up becomes a price increase you have to sell. Repeat the step-up date at order confirmation, in onboarding, and 30 days out, then honor it publicly.
Bundles
Right when the module's value depends on the core product and you want attach at renewal; a bundle also moves the comparison from a standalone competitor to an ecosystem. Keep the module's standalone list price on the order form, or the bundle hides its revenue from your tier mix analysis a year later.
Trial design is a conversion lever with no acquisition cost attached. Do the math for your own launch: 10,000 trial starts a year at 12 percent trial-to-paid is 1,200 customers; at 18 percent it is 1,800, and the extra 600 are $3M of ARR at a $5,000 contract value, from the trial rather than the product. If twice your time to value passes 30 days, a free trial is the wrong motion for a B2B module; a paid pilot with a defined scope, a price, and a conversion date will outperform it, because the buyer does not burn the window on setup and the pilot price sets the reference.
Gate on the second column: everything that solves the buyer's immediate problem goes in the trial, and the capabilities your retained users lean on at scale stay behind the paid tier, visible in the interface so the ceiling is felt. A trial with every feature open produces exploration, not urgency. Build onboarding to one meaningful action inside the first ten minutes. Trigger messages off behavior rather than the calendar; you have not run a second report yet outperforms a day-three tips email. Three or four days before expiry, send what the buyer built and what they lose. And give trial conversion one owner; when nobody owns the funnel, trials get longer, access gets more generous, and conversion drifts.
Marlow Analytics, $12M ARR, ran 30-day trials with full access. Conversion was 4.8 percent, so the team blamed product fit and funded features. The product analytics said otherwise: most trial users never ran a second report after the first session. Marlow cut the trial to 14 days, gated the advanced report templates, and added a day-3 prompt for anyone without a second report. Sixty days later conversion was 11.3 percent. The product did not change.
Fennwick People Ops, $61M ARR, launched a shift-scheduling module with a 60-day free trial and a 25 percent trial-to-paid target. Buyers had it working by day 10, then waited seven weeks. Conversion came in at 11 percent. Fennwick cut the trial to 14 days and printed the list price on the trial page. The next cohort converted at 23 percent.
Freemium is a different decision, and the variable that decides it is decision authority. A consumer upgrading your app spends their own money; the conversion happens once the habit is strong enough that the free limit is genuinely annoying, and no prompt works before that. A B2B user usually upgrades with a budget they do not control, so a B2B free tier is designed for organizational visibility: the limit should hurt the team rather than one person, message history or shared workspaces rather than a per-user cap, so that someone with budget authority feels it. Before you touch the upgrade prompt, run the day-7 diagnostic: split free users by whether they were still active at day 7 and compare paid conversion for each group. If the retained group converts at a multiple of the base, the problem is habit formation, not the prompt or the price. The three free-tier models and the consumer cases are in the B2B free tier that cut consumer conversion from 6.2 to 0.8 percent.
Success at 30, 60, and 90 days means three numbers. At 30 days: activation, the share of launch accounts that reached the value moment you defined. At 60 days: conversion or attach rate, and pocket price against list on closed deals. At 90 days: retention of the first cohort and the first expansion signal on the value metric. Cannot write them before launch? You have a cost with a landing page.
Protect It at Launch
The first ten deals set the discount baseline. If the first ten close at 30 percent off, 30 percent off is the price, whatever the list says. Discounting is usually a symptom, and at launch the cause is one of three: reps do not believe the price, cannot explain the metric, or get paid the same either way. Deal governance protects margin; sales comp alignment keeps it.
Write the exception policy before launch. Which discounts are standard and printed: multi-year, volume on the value metric, a launch cohort discount with an end date. Which are exceptions, and who approves each level. What the buyer gives in return. And what is never available: the module bundled free into a core renewal, a 100 percent discount that never appears in the discount field. An illustrative ladder for the first 90 days: reps hold no discretion, the sales leader approves up to 10 percent with a give-get, and anything above that goes to a deal desk.
Deal desk rules for a new product are narrower than for the core product, because you have no history. No exception discount in the first 90 days without a documented competitive alternative and a give-get. A discount reason code on every deal, so the 90-day review can tell competitive pressure from habit. And pocket price tracked from day one: Marn and Rosiello's pocket price waterfall (Harvard Business Review, 1992) is still the cleanest way to see what a deal collected after every concession. The core product's governance layer is in the SaaS pricing strategy guide.
Brief sales on five things. Why the price is what it is, in the buyer's units. How the value metric works and how the buyer forecasts the invoice. The three objections you expect. What the rep can offer and what they cannot. And how they are paid: if comp pays on bookings at any price, discounting is rational. Pay on pocket price and the policy enforces itself.
Larkspur Analytics, $73M ARR, launched a data-quality module at $18,000 list with no exception policy. The first ten deals closed at an average of 31 percent off. The eleventh buyer opened with a reference call in which a launch customer had mentioned their price. Realized price settled at $12,400 while the list said $18,000. Watch the first ten deals yourself.
The 90-Day Review
Book the 90-day review before launch and name who owns each number. Five measures decide whether the price moves.
Win rate by tier
Against the core product's win rate in the same segment. High and rising with little negotiation says the price is low.
Tier mix
Where new customers landed. One tier holding most revenue means the fence is not sorting buyers: a packaging problem first.
Discount rate and reason codes
List to pocket, by rep and by reason. Competitive pressure and habit look identical in the average and nothing alike in the codes.
Expansion on the value metric
Attach into the existing base and metered growth inside the launch cohort. The metric works when the invoice grows without a sales conversation.
Churn and downgrade by cohort
Launch cohort against later cohorts. Early churn on a new module is usually a misunderstanding of what was bought.
Move the price only when two signals point the same way. High win rate, low discount rate, and buyers who rarely negotiate together say the price is below value: raise it. Losses that cite price alongside rising discount requests say check packaging first, because packaging beats pricing. If you ran the cohort test, this is where the winning price becomes list.
Move without resetting the reference price. Raise list for new logos and give the launch cohort a grandfathering window with a date. Add a tier above the current top instead of cutting the base, so the reference moves up. Change the included volume on the value metric rather than the unit price. And communicate it as a packaging change, because price increases fail because of communication, not the price. For a product already in market, see value based pricing at Series B.
Dunmore Controls, $95M revenue, equipment monitoring for manufacturers. It launched a predictive-maintenance module at $150 per monitored asset per year. The 90-day review showed a 78 percent win rate, and a 3 percent average discount. Rather than raise the unit price, Dunmore added a Best tier with failure prediction at $240 per asset and moved the launch edition to the middle. Realized price per asset rose 41 percent over two quarters.
Three Places This Framework Breaks
Three places this framework breaks, what it costs, and what to do instead.
Value-in-use pricing when the buyer will not count the value
The framework is economic value estimation. It breaks when the value lands on a team the buyer does not manage, or as soft savings a CFO refuses to book. A workflow module priced at $48,000 against a $190,000 value model built on 3,000 saved analyst hours saw 40 percent of deals stall in procurement, because the CFO would not count hours no one would lay off. The correction: price against a cost the buyer already budgets, such as the external review fee or the license they will cancel.
Good-better-best for a module sold into one segment
The framework is three tiers at launch. It breaks when the module goes to existing customers who are alike and the base edition does the whole job. 85 percent of buyers chose the base tier, the Better tier's fence was a capability the segment did not need, and the two extra columns added 20 minutes to every discovery call for a 6 percent revenue lift. The correction: one edition and one add-on at launch, and tiers when usage shows where buyers split.
Sequential price tests at low deal volume
The framework is cohort price testing. It breaks in enterprise sales at eight deals a quarter. Two cohorts of four deals cannot separate price from rep and quarter, and after one loss at the higher price the reps declared the test over and discounted an average of 20 percent on every deal after it. The correction: treat the first 20 deals as qualitative, test the anchor through proposal structure, and hold the price for a fixed number of deals rather than weeks.
The Eight Weeks Before Launch
The sequence above, laid against a calendar.
- Week 8: write the launch segment and the job it hires the product for in one sentence, signed by product, sales, and finance
- Week 8: list three to five candidate value metrics and score each against the four tests
- Week 7: run twelve reference-price interviews and record the alternative, its cost, and the budget line
- Week 7: build the value-in-use model for three real accounts
- Week 6: classify every capability as leader, filler, or killer, group by use case, and draft the base edition, the fence, and the add-ons
- Week 6: pull usage depth from any beta or existing feature, rank accounts by it, and run the 30-second test on the draft tier page
- Week 5: set the list price and the cohort test: the non-price split, one variable per cohort, the decision criteria, and how many deals it holds
- Week 5: measure time to value from the beta, choose the offer, and write its 30, 60, and 90-day success numbers
- Week 4: write the exception policy, the discount authority ladder, and the three deal desk rules
- Week 3: brief sales on the value math, the metric, the objections, the offer boundaries, and the comp plan
- Week 2: configure the CRM with the SKU, the tier field, the metric quantity field, the cohort field, and the discount reason codes
- Week 1: publish the list price, book the 90-day review, and name an owner for each of the five measures
Company names and figures in worked examples are illustrative.
New Product Pricing Strategy and Packaging: Common Questions
How do you price a product that has no competitors?
Should a new product be priced by seat or by usage?
How many pricing tiers should a new B2B product have?
Should we launch with a free trial or freemium?
How do you test willingness to pay before launch?
How do you test two launch prices without showing buyers two prices?
When should you change the launch price?
How do you keep sales from discounting a new product?
Set the price before the first deal sets it for you
Bring the product brief and any usage data you have; you leave the call with a value-metric hypothesis, a packaging sketch, and a scoped next step.