Competitive intelligence with open sources: a reading of Sun Tzu

Knowing a competitor’s offer is not the same as knowing all their plans. Competitive intelligence begins with a specific question and requires distinguishing between what is published, what is verified, and what is only a hypothesis.

Competitive intelligence inspired by Sun Tzu can be approached as a process of questions, checks and reviewing conclusions. Within Sun Tzu applied to business, this reading focuses on public information and responsible analysis. It does not present industrial espionage as a lawful activity by definition.

What is taken from Chapter 13 and what changes when it is brought into the company

Chapter 13 links knowledge of the adversary to information obtained from people, rather than replacing it with guesswork or calculations. The five kinds of spies in Chapter 13 develop that military argument. Their application to open sources is a contemporary adaptation, not an equivalence between each ancient agent and a digital tool.

The business question needs a limited scope. “Knowing everything about the competition” does not indicate which information would change a decision. Asking what an offer includes for a certain type of customer makes it possible to identify relevant documents and recognise what those documents still do not explain.

It is also necessary to define what is meant by competitor. Another company may be of interest, but so might the possibility that the customer keeps their current process. Sun Tzu’s vocabulary in business helps to separate that comparison from a view in which any interlocutor becomes an enemy.

In this guide, working with open sources means starting from published and accessible information, without accessing other people’s accounts or requesting confidential material. Accessibility does not prove that any subsequent processing is allowed. Specific practices require legal and privacy review; this article does not determine their legality.

Collecting, comparing and interpreting are different tasks

Collecting from a question

Imagine a fictional company that wants to understand what support a competitor offers during the implementation of a service. The question is not whether that competitor is “better”, but what work their published offer includes and what part the customer would have to resolve by other means.

For that example, the commercial description, user documentation and the public terms of the service would be relevant. A general article on the provider’s history might add context, but would not necessarily resolve the question about a specific implementation.

Each finding needs to retain its source, consultation date and scope. A document may describe an earlier version or an offer available only to certain customers. Recording those conditions helps prevent a limited statement from ending up presented as a universal feature.

It is also wise to note absences without turning them into conclusions. “It is not explained in the documents reviewed” is different from “it does not exist”. The first sentence describes the result of the search; the second makes a statement about the company that may require further checks.

Comparing without counting repetitions as independent evidence

Suppose a commercial note announces support during go-live. The public documentation describes the tasks included and makes clear that certain adaptations are excluded. The second item allows the announcement to be made more precise. It would not be rigorous to summarise both by saying that any implementation work is included.

If several media outlets reproduce the same note, their similarity does not necessarily provide independent verification. There may be a single original statement repeated several times. When assessing the support for a conclusion, what matters is how the information was obtained, not merely how many pages contain similar wording.

A discrepancy should also be preserved. If two documents appear to describe different conditions, it is necessary to check whether they relate to different dates, versions or recipients. When that cannot be clarified, the report should point out the uncertainty rather than silently choosing the most convenient version.

Interpreting and explaining what is still not known

In the fictional case, what is confirmed could be that there is published support with a specific scope. A different hypothesis would be that the provider is trying to reduce adoption difficulties. That intention may seem plausible, but it is not demonstrated solely by the service description.

The consequence for our company would be to review what its own customer needs during the launch. Copying the competitor’s offer does not automatically follow from the finding. First it would be necessary to check whether the same difficulties appear and whether the team has the capacity to deal with them.

The risk of ignoring the enemy helps to explain why internal analysis can be insufficient. The reverse caution also matters: knowing part of another party’s offer does not authorise reconstructing its whole strategy or future decisions as facts.

A useful report distinguishes between what is observed, the proposed explanation and the data that could change that explanation. If it cannot specify which decision the finding affects, it may be collecting interesting information without answering the original question.

Big data, mystery shopping and former employees: the limits of the analogy

The analysis of large volumes of data does not eliminate those distinctions. Accumulating records can repeat the same mistake, mix versions or amplify an incorrect classification. Before interpreting a pattern, it is useful to know what each datum represents and what was left out of the collection.

When personal data are involved, the question is not limited to whether they can be found. The purpose of the processing, its scope and the access conditions all require specific review. A reference to Sun Tzu does not resolve privacy obligations or authorise building unnecessary personal profiles for the business decision.

Mystery shoppingmystery shopping should not be treated as an automatically permissible practice either. Its assessment requires attention to the context, the form of interaction and the data intended to be recorded. This reading does not propose false identities or procedures for obtaining restricted information.

Hiring former employees deserves an equally clear separation. Assessing experience and skills is not the same as requesting internal documents, customer lists or confidential plans from another company. Bringing a person into the company does not turn their previous access to information into permission to use it.

In the strategy of a start-up against a giant, external information helps to check a specific difference. That is the practical limit of the comparison: understanding a need and reviewing your own proposal, without attributing to the competitor intentions that have not been documented.

Before taking a conclusion as final, identify its evidence, its limits and what data would force it to be reviewed. For any specific method of obtaining or processing information, request legal and privacy review.

View all articles in Sun Tzu in business: resources, competition and decisions