Model innovation mistakes: a nine-question audit of your change plan

A charging change starts with a disputed diagnosis

The planning team is weighing a new charging structure, but its reason for change is still a collection of assumptions. To avoid model innovation mistakes, it tests its diagnosis and proposed rollout in sequence, using the current offer as a baseline before deciding what can change.

To make the arithmetic concrete, the team uses invented placeholders: list price 100, collected cash of 88 per unit, and a 40% gross margin for discount calculations. The purpose is not to treat list price as realized revenue, but to expose which assumptions must be checked before the redesign.

The team tests its diagnosis against the current return

The first test is whether the change plan states a binding constraint and measures it. Saying that the firm wants recurring revenue describes an ambition, not what currently caps it; redesigning around an unmeasured limit could leave the real problem untouched.

Before comparing alternatives, the team records the constraint and ties the claim to constraint metrics. If nobody can say what limits the business and how it is measured, the proposal has not earned a decision. The audit therefore moves from evidence to economics, rather than starting with a favourite structure.

The second check is actual cash against list price. With the scenario's list price at 100 and collected cash at 88 per unit, the team sees that the current return differs from the one list price alone suggests. Finding the gap takes an afternoon, but it changes the baseline for every later comparison.

That finding becomes the reference point for judging any replacement. The team records collected cash, rather than treating list price as revenue earned. Skipping this check would make later comparisons rely on an imagined version of the existing business, not its observed return.

Discount evidence makes a reversible test the next move

The team checks discount authority before replacing the structure. Most suspected model problems are discount-authority problems, and tightening authority can be done in a week for nothing. The available pricing architecture also supports testing a reversible approval change before committing to a structural move.

The gross-profit arithmetic gives the team a reason to make that check first. At a 40% gross margin, a 10% discount needs 33% more volume to hold gross profit, while a 20% discount needs volume to double. The approval rule is therefore worth testing before changing the model itself.

For a former-price comparison, the FTC's Guides Against Deceptive Pricing say that a former price is a legitimate comparison when it was the actual, bona fide price offered regularly for a reasonably substantial period. The team checks how it presents discounts as well as who can approve them.

The team keeps the gross-profit test beside its approval decision: at a 40% gross margin, a 10% discount needs 33% more volume, and a 20% discount needs double. That supports checking the discount rule before redesign. A revised approval process is reversible next week and requires no customer conversation.

The proposed alternative is compared with what remains hidden

The next check asks what the admired model conceals. A public price page reveals the offer, not the other firm's cost structure, channel terms or subsidies. Copying its visible half can produce the worst version of both models instead of addressing the measured constraint.

The team tests whether its charging basis suits the customers it intends to keep. The choice should reflect the customers the firm wants, not billing convenience. Invoicing simplicity is not the team's deciding objective.

The team treats charging-basis fit as a customer decision, not an invoicing convenience. It asks whether the selected basis suits the customers the firm wants, rather than choosing the option that is easiest to invoice.

Before launch, the team defines what can change

The proposal now has to specify how future changes will work. The team records the event that triggers a review, the notice customers receive, and who stays protected until when. Including these terms in the initial design makes them structural choices, not concessions sought after the offer has been presented.

Without a written trigger, notice period and protection, the company may later find it cannot move a price at all. The team treats these as design decisions, not details to negotiate after launch. Customers therefore have a stated basis for review before anyone is asked to accept the offer.

The team also limits the quarter to a single moving part. If packaging and charging basis change together, an observed result cannot be attributed to either choice. Two changes in one quarter cost the year and provide no information about either, so the design isolates one variable before reading outcomes.

A pattern is something to test, not something already proved. That is also the ordinary approach in MIT's data mining course. If the team has already changed two things, it should stop adding variables, hold the current state for a full cycle, and record what can and cannot be concluded.

The overlap receives a budget before the decision is announced

Before launch, the team sets aside a distinct transition budget. During the overlap, billing must handle both structures, sales must quote both, and finance cannot produce a clean margin figure.

These costs arrive first and appear in none of the models, so funding them from expected margin leaves the change underfunded.

The plan treats transition as a separate project cost. Companies that fail to budget for the overlap may stop halfway, the worst available position, while trying to run two models badly. The four constraints help frame the decision, while unit economics keeps the cost side in view.

The separate budget covers the actual work of operating the overlap, rather than assuming the new model will pay for its own introduction. Billing must handle both structures, sales must quote both, and finance must account for the period without a clean margin figure. The constructed worked case shows related arithmetic on one firm.

Customer protection is decided ahead of objections

The team decides who is exempt and for how long before announcing the change. Without a named list and end dates, each complaint can create a different arrangement. Recording protection in advance prevents exceptions from being improvised case by case after customers hear the offer.

Without that decision, arrangements can accumulate differently and become impossible to reconcile three years later. The team records protection terms alongside the change rule, so customers are not handled one complaint at a time. The pre-decision checklist helps sort what is ready from what still needs an answer.

The audit ends with nine practical questions

The audit's conclusion is not to redesign faster, but to record assumptions and sequence the work. Its nine questions test whether the team has measured the constraint, established the current return, set change rules and customer protections, and funded the transition.

The team makes the reversible change first, by itself, and reads the result before attempting anything less reversible. State the constraint and measure, establish what the current model earns, put review terms and exemptions in writing, and budget the transition independently. If a proposed change cannot be bounded to a group, that is an argument for waiting.

The remaining decisions are handled explicitly

The team resolves outstanding objections by separating the current test from what the evidence can support. It records what cannot be inferred and keeps urgency from turning the next move into an uninformative rollout.