Pharmaceutical competitive intelligence is the systematic collection and analysis of information about named competitors, produced to support a specific decision. The decision is the part that matters. Pricing, positioning, trial design, launch sequencing, indication prioritisation, whether to in-license an asset — CI exists because somebody has to commit money against an uncertain competitive future, and would prefer to be less wrong.
That framing rules a lot of things out. A weekly digest of competitor press releases is not intelligence. Neither is a 90-slide landscape deck that nobody can act on. Both are collection. Intelligence is what happens when someone with domain knowledge looks at the collection and says: this changes the plan, and here is how.
The short version
- Unit of analysis: named competitors and their assets, not the market in aggregate
- Test of quality: did it change a decision, or just get circulated
- Hardest input: not data — judgment, and access to people who were in the room
- Most common failure: one undifferentiated output for five audiences who each needed a different cut
CI versus market research
These get conflated constantly, often because the same agencies sell both and the primary-research methods are nearly identical. The difference is the unit of analysis.
Market research studies the market: patient flow, unmet need, prescriber behaviour, payer attitudes, willingness to pay. The output describes an environment.
Competitive intelligence studies competitors: what asset a specific company holds, what trial they are running and in whom, what label claim they are likely to win, when they will read out, and how they are likely to price and position. The output describes an opponent.
In practice a good commercial team runs both and the outputs interlock — you cannot forecast a competitor's uptake without understanding prescriber behaviour. But conflating them produces the classic failure where a team commissions a landscape study, receives a beautiful description of the disease area, and still cannot answer whether the competitor's Phase III is going to hit.
What CI teams actually produce
| Deliverable | Consumer | Cadence |
|---|---|---|
| Landscape / indication map Every asset in an indication by phase, mechanism, line of therapy and sponsor |
Brand, clinical development, BD | Refreshed quarterly, rebuilt annually |
| Readout calendar Expected data disclosures, regulatory dates, congress slots |
Commercial leadership, forecasting | Rolling, reviewed monthly |
| Congress coverage What was presented, what it means, what the room thought |
Nearly everyone | Per congress, within 24–72 hours |
| Competitor profile One company or asset in depth, including likely strategy |
Brand teams, BD diligence | On demand |
| Early warning / alerting Material change, flagged fast |
Whoever owns the affected decision | Continuous |
| War game / scenario work Structured rehearsal of competitor response |
Launch teams, senior commercial | Pre-launch, pre-major-readout |
Notice how differently these move. A readout calendar changes when a company amends a trial registry entry. Congress coverage has a 48-hour useful life and then becomes archive. Early warning has to be continuous or it is not early. A function that tries to serve all six on one weekly cadence will serve none of them well.
The source base
Almost everything a pharma CI team uses is public. That surprises people who expect the work to be clandestine; it is not, and the practitioners who treat it that way tend to create legal exposure without generating better insight.
Secondary sources
- Trial registries — ClinicalTrials.gov, EudraCT/CTIS, WHO ICTRP, jRCT. Enrolment criteria, endpoints, site counts, estimated completion, and the amendment history, which is often more revealing than the current record
- Congress abstracts and presentations — the primary disclosure route for oncology data
- Regulatory outputs — approval documents, review packages, CHMP opinions and minutes, advisory committee materials
- Company disclosure — earnings calls and transcripts, investor days, 10-K and 20-F filings, pipeline pages
- Publications and preprints — plus what has conspicuously not been published
- Patent and IP filings — composition of matter, formulation, method-of-use, and the timing of continuations
- Job postings and conference sponsorship — genuinely useful signals of where a competitor is building capability
Primary sources
- Treating physicians and investigators — the read on tolerability, on whether a statistically significant result is clinically meaningful, on what would actually change prescribing
- Payers and formulary decision-makers — how a claim will be received, what evidence gap will be cited
- Former employees and advisers — with real care, and inside a compliance framework
The line that matters. Public information gathered in the open is fair game. Inducing someone to breach a confidentiality obligation is not, regardless of how it is framed or who is asked to do the asking. A CI function without a written elicitation policy and trained interviewers is carrying risk it has not priced.
Why the same landscape deck gets rebuilt every year
Because it was built as a document rather than as a data structure. An indication landscape assembled in slides is stale the week it ships and cannot be diffed against last quarter, so the next refresh is a rebuild rather than an update.
The teams that escape this treat the landscape as a maintained record — asset, sponsor, mechanism, phase, indication, line of therapy, biomarker population, trial identifiers, expected readout — and generate documents from it. That is unglamorous and it is the single highest-leverage change most CI functions can make.
It also makes the difference between "we track 400 trials" and "these three moved this quarter, and here is what they do to the forecast." The first is a boast about collection. The second is intelligence.
Where this gets genuinely hard
Oncology. Not because the sources differ, but because the granularity does — a single tumour type fragments into lines of therapy, biomarker-defined subsets and combination regimens, each of which is effectively its own competitive market with its own set of assets. Tracking "NSCLC" as one thing is close to useless.
That is a separate problem with separate methods, covered in why oncology CI is a different job. In practice it means maintaining a landscape per indication rather than per tumour type, which is a larger job than it sounds — an indication-level coverage map gives a sense of how many separate landscapes that turns into.
How to tell whether a CI function is working
Not by output volume, and not by whether stakeholders say they find it useful — they will, because it is free and it is interesting. Two better tests:
- Can someone name a decision that changed? A price point, a trial arm, a launch sequence, a walk-away on a licensing deal. If nobody can name one in the last year, the function is producing content rather than intelligence.
- Did anything arrive as a surprise that shouldn't have? A competitor readout, an approval, a label expansion that the commercial team learned about from the trade press. One of those is a bad quarter. A pattern of them is a broken function.
Related
- Primary vs secondary intelligence in pharma — what each source type can and cannot tell you
- Congress intelligence: ASCO, ESMO, AACR, ASH — the abstract-to-podium cycle
- Why oncology CI is a different job — indication-level granularity