Reading an industry census like an operator
Most people read industry surveys for validation. An operator reads them for allocation. Five numbers worth acting on.
Guide 05 / Topic
Competitive intelligence
The premise
Useful intelligence changes a decision; everything else is a collection of links.
A living guide \u00b7 last tended 26 August 2026Competitive intelligence is not a long document about everything a company says. It is a disciplined way to reduce uncertainty around a decision.
The work begins by naming the choice, the evidence that could change it and the limits of what public or vendor-provided information can prove.
01
Begin with the decision
Research expands endlessly when the decision is vague. A vendor review, market scan and competitor analysis need different evidence even when they mention the same companies.
Write the decision in one sentence. Then list the few criteria that could move it. This becomes the filter for sources, questions and comparison.
02
Separate claims, evidence and inference
A company website can prove what a vendor claims. It cannot by itself prove delivery quality. A customer quote can describe one experience. It cannot prove the typical experience.
Label each item as a direct claim, independently supported fact or analyst inference. The discipline prevents confident language from outrunning the source.
Claim
What the company or interested party says about itself.
Fact
What a reliable source directly supports.
Signal
A pattern worth investigating but not yet decisive.
Inference
A reasoned conclusion that remains dependent on assumptions.
03
Compare operating fit, not feature counts
Long feature tables reward the vendor with the longest list. They often ignore implementation effort, ownership, integration, exception handling and the work required after launch.
Compare the full operating model: what changes for users, which team maintains it, how errors surface, what data moves, what the contract protects and how the organisation can exit.
04
Make uncertainty visible
Good intelligence does not remove every unknown. It shows which unknowns matter enough to test before a commitment.
Close the work with a recommendation, confidence level, critical assumptions and a short verification plan. The next step might be a reference call, a controlled pilot, a commercial clarification or a decision to stop.
The operating model
01
Decision-led
Research only what can inform a real choice.
02
Source-aware
Match the confidence of a claim to the quality and incentive of its source.
03
Operating-fit
Compare ownership, workflow and risk, not only features.
04
Uncertainty-open
State what remains unknown and how to test it.