Executive Research Reporting

How do I write a research summary for a board or executive team?

Build it backwards from the decision, not down from the report. Write the decision as a sentence containing a choice, then work out what the decision-maker has to believe to choose either way, and include only the conclusions that bear on those beliefs. A shortened report contains the most important findings. An executive artefact contains the findings that bear on a choice, which is a different and usually smaller set.

How do I communicate uncertainty to executives who do not like it?

Convert it into something they can use. Name what would change the conclusion, which turns a caveat into a monitoring instruction. State what it costs to be wrong and whether the decision is reversible, which is what they are actually calculating. And separate the part of the conclusion held at high confidence from the part held at moderate, rather than qualifying the whole thing. A caveat with nothing attached to it is an obstacle and will be ignored; "further research is required" tells nobody anything.

Should I say how confident I am in each finding?

Yes, in words, on every headline conclusion. It feels like it weakens a conclusion and it does the opposite: it tells the reader where to spend their scepticism, which is the scarcest thing in the room. An executive acting on a moderate-confidence finding as though it were certain is the most expensive failure available in research, and it is a communication failure rather than an analytical one, because the analysis knew. Say strong things strongly too: hedging a high-confidence finding is miscommunication, not modesty.

What goes on a one-page research summary?

About six things: the question, the answer with its confidence, three to five supporting conclusions each with confidence, the implication or recommendation, the principal limitation in body text, and the basis (method, sample, base, fieldwork dates). Three things are not compressible whatever the space pressure: confidence, base, and any limitation that would change what the reader does. The test for whether you have compressed too far is not length, it is whether someone reading only the page would reach a different decision from someone reading the report.

How do I tell a client the research contradicts a decision they have already made?

Six things, in order. Verify it harder than anything else in the study, and do not raise its confidence to help it survive. Separate the finding from the verdict: the research says something about the world, and whether that makes the decision wrong depends on things the research does not contain. Lead with the evidence rather than the conclusion, which inverts the usual answer-first rule for this one case. Name what would have to be true for their position to hold, which converts a confrontation into a testable question. Give them somewhere to go: what could be adjusted rather than reversed. And escalate how it is handled to a named human, because that is a relationship judgement rather than an analytical one.

What do I do when a stakeholder asks for a recommendation the data does not support?

Give four things instead of either a refusal or a fabrication. What the evidence does support, at its real confidence, which is often more useful than the request. What a recommendation would require: the specific missing evidence, its cost and its timeline. The options laid out as a decision under uncertainty, with what each assumes and which assumptions the evidence speaks to. And your professional view, clearly labelled as judgement rather than as a finding, where you have one and it is wanted. What is never available is a recommendation presented as evidence-based when it is not, because senior audiences act on recommendations identically whether or not they were earned.

How short can a research summary be before it becomes wrong?

The test is the different-decision test: would a reader of the summary alone reach a different decision from a reader of the full report? Run it explicitly by giving both to two colleagues rather than assuming. Also run the onward-travel test: assume every sentence will be quoted alone in an email by someone who read nothing else, because that is what happens to executive material. A sentence that is only true in context will be untrue within a week.

How do I handle the hard question in a research debrief?

Write down the three hardest questions before the meeting, honestly, as an intelligent sceptic would ask them, and write the honest answers including where the answer is that something cannot be ruled out. Pre-empt the one that goes to your central conclusion in the document itself, briefly, because meeting a challenge on the page is far stronger than meeting it under pressure. Prepare the others and do not raise them, since pre-empting everything reads as defensiveness. And pre-empt with evidence, never with a statement about how confident you are, which invites exactly the challenge it was meant to prevent.

What is a decision brief and when is it better than a report summary?

A decision brief leads with the recommendation, then the reasoning in a sentence or two, then the evidence with its bases, then the confidence and what would change it, then the options not recommended and why. It is the better format whenever a recommendation is owed and supportable, because it forces the research team to say what follows rather than leaving that inference to a reader with less context than they have. A findings summary is right when the decision is not yet framed or the research is one input among several.

What should I do when the research says the same thing as the last two studies?

Say so. The problem is organisational rather than evidential, and presenting a repeat finding as a discovery wastes the opportunity. Show the previous studies and shift the question from "is this true", which is settled, to "what has prevented action", which is open and more useful. Senior audiences generally respect this move even though it is uncomfortable to make.