Case · strategic options

A patent expiry where the response did not matter

Two decision trees were built — the producer's own, and the counterparty's. The finding came out of the second one.

The situation

A specialty producer was approaching the end of patent protection on its principal product. The commercial question was the familiar one: defend on price, reposition, or exit.

The complication was the counterparty. The largest buyer of the product was also the party most capable of replacing it — able to switch suppliers, acquire a producer, build its own capability, or develop a substitute once the patent lapsed. Customer and prospective competitor were the same organisation.

The prevailing view inside the business was that price would decide it. Concede enough and the relationship holds; concede too much and the category is destroyed.

A second decision was identified alongside it — how to respond to low-cost entrants arriving from other manufacturing regions — and set aside. The tooling to model both at once did not exist at a cost the decision could bear.

Two trees, not one

The first tree was the producer's own: compete on price or hold, with existing products or with a lower-specification range added, alone or through an alliance.

The second was the counterparty's, with its options laid out as it would see them — continue buying, exit the relationship, develop a substitute at equivalent specification and face litigation, or develop at lower specification and avoid the patent altogether. Each branched again into how it would source: a new supplier, self-manufacture, acquisition, or a strategic alliance.

The producer's decision was then evaluated against the counterparty's best response to it, rather than against an assumption about that response.

The counterparty's move was the same whatever the producer did

Across every branch of the counterparty's tree, its best move was identical: develop a substitute at lower specification, and acquire a producer to make it.

That held whether the producer cut price or held it. It also held if the producer launched a lower-specification range of its own, and if it competed through an alliance rather than alone. Four quite different strategies, one unchanged response.

The reason was structural. The counterparty's decision turned on its own capability and on what it could develop or acquire — not on what it was being charged, or what range it faced. The producer's choices moved its own margin. They did not move the other side's behaviour.

One variable did change the answer

If the counterparty expected to prevail in a patent challenge, it took the path that provoked one, and the producer's expected outcomes fell sharply. If it was genuinely uncertain — no more than an even chance of winning — it avoided that path entirely, and the producer's expected outcomes rose.

The decisive quantity was not the strength of the patent. It was the counterparty's belief about the strength of the patent.

What was decided

Not a price move. The action was to signal, clearly and early, that the patent was strong and that the producer was prepared to litigate.

That would not have emerged from any amount of analysis of the producer's own options. It came out of the second tree.

The same structure, in Canada, now

IQVIA counts 180 drugs representing roughly $10 billion in annual Canadian sales losing protection between 2026 and 2030. The reflex in each of those decisions is the same: model our options, decide how much price to concede.

Canada makes the prior question unusually sharp, because for many publicly funded products the entrant's price is set by formula rather than by competition. Under the pCPA framework a single generic is priced at 75% of the brand reference price, dropping to 55% after three months of public-plan funding. A second entrant takes it to 50%; three or more take it to 25% for oral solids and 35% for other dosage forms. The three-month window is not protected — if another manufacturer arrives in month two, the lower tier applies immediately. That framework runs to 2028, so it is stable across the whole cliff.

Read that alongside the first finding. The entrant's realised price is anchored to the brand list price by formula and stepped down by how many others arrive — neither of which the originator's own price move changes. On the public-plan side, defending on price does not alter the entrant's economics at all.

Timing is equally outside the originator's control. Health Canada's generic submission backlog stood at 250 in June against 117 the previous September, and on-time review fell to 72% within 180 days in 2025-26 from 84% the year before. Entry has become a distribution rather than a date — for the entrant deciding when to commit, and for the originator deciding when the revenue goes.

None of which settles what any particular molecule should do. It establishes that the levers most likely to matter — authorised generic, contracting position, switch strategy, the timing of exit — sit outside the price response that usually absorbs the debate. Which of them is decisive has to be modelled, from the entrant's chair as well as your own.

What has changed since

This work was done manually. Building both trees, valuing the endpoints and running the sensitivity analysis took weeks, which is why so few decisions of this kind ever got the treatment.

The same analysis now runs inside a decision engine. The structure is built from a description in plain language, the endpoint values are calculated with the client's own economics, and the sensitivity work that used to take days — including the thresholds at which the answer reverses — runs in minutes.

The method is not new. What is new is that it can be afforded on a decision that has to be made this quarter.

Client and counterparty are not named. The structure and the findings are as they were.