Frameworks & methodology: building a product launch playbook that actually drives revenue

The date is the decision. Engineering says the build is stable in five weeks, sales wants it in front of the field at January kickoff, and finance has already put a number against it in the plan. Someone has to pick. The honest answer is that you cannot pick until three things exist on paper: which tier the launch sits in, whether the organisation can carry it, and what the date itself is worth in revenue. Nielsen, which has measured consumer packaged goods launches for decades, has long put the share of new products that fail to survive their first year on shelf in the large majority. Almost none of those are creative failures. They are dating and readiness failures, locked in months before anyone wrote a headline.

Tiering rules you can defend in a room

The foundations lesson sets out the tiers themselves. What it leaves you is the harder part: assigning one when the product lead thinks their feature is the biggest thing this year and the CFO thinks it is a footnote. Score five dimensions from 1 to 5, and double-weight the first:

• Revenue at stake across the first four quarters
• Behaviour change demanded of the customer (migration, retraining, a new buying centre)
• Whether the launch makes a competitive claim you will have to defend publicly
• How many internal teams change their work: support, pricing, partners, legal, channel
• Reversibility, meaning whether you could quietly withdraw it in ninety days

Maximum score is 30. Above 22 is tier 1, 12 to 21 is tier 2, below 12 is tier 3. The tier then sets the calendar length: tier 1 gets T-90 or longer, tier 3 gets two weeks and no field enablement at all.

The rules earn their keep on the edge cases. A pricing page change with no new code can score 24: irreversible, touches every team, changes how customers buy. An eighteen-month platform rebuild can score 9, because no customer has to do anything differently, which makes it a release in the sense the foundations lesson draws, not a launch. Teams get this backwards constantly. They tier by engineering effort, because effort is the number they can see, and then spend a tier 1 attention budget on something the market cannot perceive.

One hard rule: a tier 1 designation requires a named executive sponsor who has committed calendar time, not just approval. Without that, downgrade it. A tier 1 launch with a part-time sponsor fails at the readiness gate anyway, six weeks later and at much greater cost.

Sub-concept 1: The JTBD-anchored positioningpositioningThe mental space you want your brand to occupy in your target customer's mind relative to alternatives.View full definition → sprint

Jobs-To-Be-Done, the framework associated with Clayton Christensen, reframes the product by the job a customer hires it to do rather than by what it is. Before positioning is drafted, the team answers one question: what is the customer struggling to accomplish, and why is this the best available way through that struggle?

Apple's iPod in 2001 is the cleanest illustration. The job was not listening to music; dozens of players already did that badly. The job was carrying your whole library without friction. "1,000 songs in your pocket" is a job statement, not a feature statement, and it survived twenty years of iteration because the job did not change.

In practice the sprint has three hard requirements: five customer interviews completed before a word of positioning is written, a canvas showing the customer's before and after state, and a message house where every supporting claim cascades from one core claim. If two of your five interviews describe different jobs, you have two launches, or one launch and a segment you are not ready to serve.

Sub-concept 2: The T-minus calendar

Work backwards from launch day with owners attached: T-90, T-60, T-30, T-14, T-0, T+30. At T-90 positioning and ICPICPKey Performance Indicator, a measurable value that shows how effectively you're achieving a specific objective, tracked over time against a target.View full definition → lock. At T-60 enablement material is drafted and in review. At T-30 beta customers are onboarded specifically to produce case studies. At T-14 every asset is final and in channel. At T-0 you execute. At T+30 you run the retrospective against the forecast.

The obvious objection: no company can run this for everything it ships. Xiaomi has shipped MIUI updates on a weekly Friday cadence for years. A T-90 calendar for each one would consume the company. That is exactly why tiering comes first. The calendar is a scarce instrument you spend on tier 1 and tier 2 only, and the discipline is refusing to apply it downward, not applying it everywhere.

Sub-concept 3: The readiness scorecard

Readiness is measured, not declared. Score five areas out of 20 each, at T-14 for tier 1:

• Field readiness: a random rep, not the launch champion, can name the job, the top three objections and the pricing without notes
• Support readiness: a runbook exists and support has handled ten simulated tickets
• Product readiness: telemetry ships with the feature, so week-one adoption is observable
• Proof readiness: at least two named customers on record, or one with numbers
• Commercial readiness: the quote, the SKU and the renewal treatment all exist in the system

Proceed at 80 or above with no single area below 12. The failure mode is scorecard theatre: counting artefacts instead of behaviour. Fifty assets in a shared drive scores zero on field readiness if the rep cannot use them. Test with a live call, not an inventory.

Hardware moves the binding constraint. When Apple opened Vision Pro pre-orders in January 2024 at $3,499, readiness meant available inventory and over 600 apps built for the platform, the product of developer seeding that started well before the date was public. Software teams can ship at 80. A supply-constrained launch scores on units in the warehouse, and no amount of enablement compensates.

Sub-concept 4: The revenue forecast behind the date

Build the forecast before you commit the date, then read the smallest number in it. An illustrative expansion launch: 8,000 installed-base accounts fit the new module's ICP; the historical 90-day attach rate for comparable add-ons is 3%, giving 240 deals; average add-on ACV of $12,000 produces $2.88M.

Then apply the two haircuts most teams skip. Cannibalisation: if 40% of those buyers would have expanded at renewal anyway, incremental drops to roughly $1.7M. Capacity: 40 quota reps carrying realistically two launch conversations each to close inside the quarter caps you at 80 deals, under $1M. The demand model said $2.9M and the capacity model says a third of that. The forecast's value is that it names the binding constraint before the date is public, so you can either fix capacity or restate the number to finance.

The same arithmetic prices a slip. At roughly $300k a month of incremental run rate after ramp, six weeks costs about $450k in year, more if the new date lands your buyers in budget freeze. Weigh that against launching under-ready: a tier 1 launch the field cannot explain burns the second attempt too, because reps deprioritise a story that failed them once.

Margin structure decides how much forecast error you can absorb. Xiaomi's public commitment in 2018 to cap net hardware margin at 5% means an overbuild is not absorbed by margin, it is absorbed by cash, which is why staged online sales function as demand sensing rather than scarcity marketing.

CMO action items

• Score your next three launches on the five tiering dimensions this week, before anyone argues about budget. Publish the scores. Arguments about tier get much shorter when the criteria pre-exist the launch.
• Run the readiness scorecard at T-14 as a hard stop, with the field test performed on a rep chosen at random. Track close rates for launches that passed at 80+ against those waved through.
• Refuse to announce a date until the forecast has a capacity line under it. If nobody can tell you how many launch conversations a rep can carry in a quarter, the number in the plan is decoration.

Common mistakes that kill launch results

• Tiering by engineering effort. The rebuild that consumed a year and changes nothing for customers gets the full calendar; the pricing change that rewires every deal gets a Slack message. Both misallocations are expensive, and the second one is the one that generates support escalations.
• Scoring readiness on artefacts. Asset counts, deck versions and completed training modules are inputs. Readiness is a rep explaining the job unprompted and support closing a simulated ticket without escalation.
• Forecasting demand with no capacity ceiling, then treating the shortfall as a marketing miss. If 240 deals of demand meets 80 deals of selling capacity, the launch did not underperform. The plan was arithmetically impossible.
• Skipping the T+30 retrospective. Compare actual attach rate, cannibalisation and ramp against what you forecast, and write down which input was wrong. That correction is the only reason version two of the playbook is better than version one.