How do you present an analysis so that management can act without confusing uncertainty with ignorance? · 5.5 Communicating implications and decisions: the inte

A recommendation is not a long summary

The quality of an intelligence product depends as much on its foundation as on its usefulness for a specific decision. A good briefing separates relevant information from implications, avoids hiding uncertainties, and sets out what is recommended now, what could change that recommendation and who should review the matter.

A useful structure for a two-page briefing is:

Decision at stake: the question, recipient and deadline.
Executive conclusion: the most defensible reading today, with a cut-off date.
Key findings: three to five, each with a source or an evidence identifier.
Interpretations and rival hypotheses: why the findings matter and what other readings are possible.
Options: possible courses of action, reversibility, costs or dependencies to be quantified.
Conditional recommendation: what we would do and under what assumptions.
Review indicators: future signals that reinforce or change the recommendation.
Limitations and owners: coverage, doubts, missing data, next update.

The synthesis must not erase disagreement. If R&D doubts the technical maturity and sales sees emerging demand, the briefing can include both perspectives and explain which experiment would narrow the discrepancy.

Interactive · Put in order

Build the briefing structure

Put the eight parts of a two-page briefing in order.

Activity: put the eight parts of a two-page briefing in order.

Communicating uncertainty without paralysing

There are differences between an unknown source, a missing data point, contradictory evidence and an inherently uncertain future. Each calls for a different response: finding another source, measuring, opening an expert review or designing flexible decisions. Avoid the catch-all “according to the AI”. A model can summarise arguments, but it does not replace provenance and does not sign off on conclusions.

Confidence is better described by the quality and convergence of the evidence than by numbers of false precision. A qualitative rating needs an explanation and consistent criteria.

Diagram · uncertainty

Four uncertainties, four responses

Not every uncertainty is resolved the same way. Tap each type to see what it requires.

Four uncertainties, four responses: this activity needs JavaScript.

Company A case · simulation

Fictional case

Company A case: briefing for management

Decision: determine whether to start conversations with potential partners for technology exploration in the coming half-year. Current reading of the fictional case: there is exploratory activity, but insufficient evidence of broad adoption and industrial profitability. Options: devote no resources; O3 targeted monitoring with periodic review; O2 contact partners and design a limited test; O1 invest in own capacity (the codes O1–O3 are those of 5.4). Example conditional recommendation: combine O3 with low-commitment exploratory contacts, subject to reviewing real costs and demands. Review: changes in verifiable industrial projects, independent results and specific customer requests.

This recommendation is a teaching example, not economic or technical advice. The sheet must make visible which future conditions would force a change.

Educational simulation: fictional information and organisations.

Do it yourself

Do it yourself · The briefing that allows an uncomfortable conversation (30 minutes)

Produce a one-page note for management. Include one recommendation, two provisionally discarded options, three pieces of evidence, one rival hypothesis, two gaps and an update rule. Then answer: “what data point would make you retract?”. If you cannot formulate it, go back to the analysis.

Deliverable: traceable executive briefing and decision/follow-up sheet.

Common mistakes: confusing more pages with higher quality; presenting scenarios as forecasts; omitting opportunity costs; recommending something because the model said so; not dating the analysis; forgetting the next monitoring cycle.

R14 · Intelligence product

Intelligence product specification and briefing

Define the need and deliver actionable conclusions.

See the template →

Check it · self-assessment

Check your briefing

This part needs professional judgement: there is no automatic marking. Complete your work in the practice, rate it with the lesson's rubric and compare it with the indicative solution.

What your answer should include

One recommendation, two provisionally discarded options and three pieces of evidence.
One rival hypothesis and two explicit gaps.
An update rule and the data point that would make you retract.
Cut-off date and each fact linked to its evidence; Company A flagged as fiction.

Self-assessment rubric (0–2 per criterion)

An answer that asserts unsupported facts cannot reach the maximum score. It is not a professional certification.

Traceable evidence

Alternative interpretations

Proportionate decision

Review and limits

Indicative solution

"The simulated information supports asserting exploratory activity, not proven commercial demand; a limited pilot is proposed, subject to validation of operation and cost, or follow-up if there is no partner. Owner and review date defined".

Mistake to avoid: Moving from observation to "invest now" without a reasoned path.

Illustrative solution for a fictional case: reasoned alternatives are valid.

Optional extensionExtension · Integrating challenge: From the first sign to the management committee

Indicative duration: 75–100 minutes with the Company A corpus or a real case of your own, subject to legitimate access to the sources.

Formulate a question about the adoption of a technology and the decision it can change (5 min).
Build a brief profile of an actor with two alternative hypotheses (15 min).
Assess a technology on five dimensions and point out gaps (15 min).
Classify six signals; separate duplicates, evidence and interpretations (10 min).
Build two contrasting scenarios and compare three options (20 min).
Write a one-page briefing with review conditions (15 min).

Final package: an intelligence file made up of actor sheet + radar + signal log + scenarios + briefing. It can be completed with the participant’s own office or AI tools. It does not need to be connected to Innguma.

Assessment criteria: traceability; distinction between fact and hypothesis; temporal coherence; consideration of alternatives; explicit connection to a decision; honesty about gaps; possibility of review. Narrative creativity alone does not make up for a lack of evidence.

Optional extensionDeep dive · A briefing that can be defended before management

An executive product is not a compressed summary of the dossier. It must answer which decision is on the table, with what alternatives, which evidence discriminates, which conditions would invalidate the recommendation and who must act. Uncertainty is not hidden in footnotes: it is placed where it affects the decision.

Company A briefing · entirely fictional example

Decision (fictional October): should resources be allocated to exploring industrial thermal storage over the next three months? Recipient: innovation committee. Cut-off date: 18/10/2026 (teaching date only). Coverage: simulated sources from the Company A corpus; it is not real research on the sector.

Executive summary (under 100 words): “The available evidence points to trials and exploratory activity, but does not demonstrate repeatable commercial demand or economic viability in our application. We recommend a limited pilot with a partner only if costs, operating conditions and right of use are confirmed; absent those guarantees, keep targeted observation. We do not recommend acquiring irreversible industrial capacity. Review in six weeks in light of independent operating data or verifiable contracts”.

Evidence: there are simulated signals of technical activity; press releases do not demonstrate results or contracts. Alternative hypothesis: the activity is mainly exploratory and communicative. Limits: simulated search, no economic data or external validation. Options: O1 invest, O2 pilot, O3 observe. Next decision: authorise or not an exploration of limited scope.

Decision / objection	Verifiable answer in the briefing	Risk of poor wording
Why act?	Learn about operation before allocating capital	“The whole sector is moving” without evidence
Why not invest now?	Lack of evidence on integration, cost and adoption	Presenting caution as a definitive rejection
What would change the stance?	Installation, cost and contracting data	Saying “we will keep monitoring” with no condition
Who does what?	Innovation defines the pilot; legal reviews IP; the analyst verifies facts	Delegating a decision to the AI

Five-sentence audit

Before dissemination, the reviewer marks with a different colour verified facts, inferences, recommendations, absence of evidence and assumptions. If a sentence mixes two categories, it is split. Compare the headline with the data; check dates, units and provenance; ask which rival explanation remains unresolved and document who approved the version. A real quotation does not guarantee that the quoted content supports the claim exactly.

Optional extension (35 min): turn the fictional file into R14 and write a one-page briefing with three options, proposed decision, uncertainties, triggers, owner, review date and source of each fact. Then swap it with another person: they must be able to find the decision in 30 seconds and be able to question three assumptions. Acceptance: the note withstands questions without resorting to invented information and distinguishes evidence from advice.

Interactive · Tag the text

Audit of five sentences

Before dissemination, the reviewer marks each sentence with a colour. Do it yourself with this excerpt from the Company A briefing.

Activity: label each sentence of the briefing as fact, inference, recommendation, absence of evidence or assumption.

References and limits

Innobasque · training material. Curso de escenarios (presentación, 2024) (2024).Stages of building scenarios and critical uncertainties.Innguma training material, not public.
Innguma · training material. Inteligencia competitiva (presentación Innguma · CEEI) (undated).Competitor analysis.Innguma training material, not public.

Company A is a fictional case: its data, companies and figures do not describe real markets or organisations.
The analyses do not assign probabilities or scores without an explicit rule. Scenarios are plausible worlds, not forecasts.

Intelligence Lab · L09

Intelligence prompt factory

Ready to apply it? Open the Prompt factory in the Lab and generate a reusable prompt for your next intelligence product. Output: reusable prompt. Works in the browser, no server, no account and no connected AI. The result may need reviewing and adapting to your case.

Open the tool Back to the learning path

Communicating implications and decisions: the intelligence product that can be questioned