Use case · Public data only
Half the revenue was heading off patent. The evidence showed what would replace it.
In 2018, three medicines carried about half of Roche Pharmaceuticals' revenue, and all three were losing patent protection. We read Roche's public pipeline and patent record to see whether the replacements were already there. Seven years later, we checked.
The franchise handoff
A gap of known size, years in advance
Avastin, Herceptin and Rituxan made up roughly USD 21 billion, about half of Roche Pharmaceuticals' 2018 revenue. On Roche's own assumption of 60–70% biosimilar erosion, sales were expected to fall by about CHF 13 billion. This is the situation EVOS is built for: a growth commitment, a hole of known size, and a question about what fills it.
The replacements were in the pipeline before they were in the revenue
We coded every molecule entering Roche's pipeline from 2008 to 2018 by who originated it, how Roche accessed it (build, partner or buy) and when it came in. Three things stood out: Genentech originated 32% of new molecules and 49% in oncology; most entered at Phase 1, when cost is lowest; and Roche's success rate was about 8%, against roughly 4% for the industry.
Access mode, 162 new molecules
Entry stage
The lost franchise wasn't defended. It was replaced.
By FY2025 the three blockbusters had dropped out of Roche's top ten. Five successor medicines brought in CHF 21.4 billion, and Roche Pharmaceuticals grew to CHF 47.7 billion. Ocrevus and Vabysmo, in the neurology and ophthalmology positions the 2018 data flagged, are now among its largest products.
The same method, built into the platform
Roche is not an EVOS customer.
| What we did with Roche public information | Where it lives in EVOS | What it shows |
|---|---|---|
| Sized the revenue at risk from expiring patents | Focus: growth goals and gap | Your gap, as a number and a date |
| Coded every pipeline asset by origin, access mode and stage | Evidence and Registries | Your opportunities, each with a source |
| Compared build, partner and buy routes | Focus: scenario plans | Options side by side against one goal |
| Tracked patent filings and partner activity | Signals | Alerts when the evidence moves |
| Checked the prediction against reported revenue | Results | Actuals against plan, re-forecast early |
What this does and doesn't prove
This is one company and one seven-year window, and the predictions matched. It is not a back-tested signal with a measured hit rate across many companies. We say so on purpose: the discipline that makes the method worth trusting is the same one that declines to overclaim it.
Your company has a public record too
Bring one product line. We'll show you what the evidence says about where your next revenue will come from, and the gap if it doesn't.
FAQ
Is Roche an EVOS customer?
No. This use case is built entirely from public sources and implies no relationship with Roche or Genentech.
Does this only work in pharma?
The patent-cliff mechanics are specific to pharma. The method (reading where growth will come from in the evidence before it shows in revenue) applies to any technical company.
Who did the analysis?
Matthew Wahlrab (Rapid Alpha) and Christian Hartmann (RWS), published as Accessing Innovation, Parts 1 (2022) and 2 (2026).