A favourable survey result often travels faster than the account of who answered it. In one of our fictional programmes, 400 participants received an invitation, 100 replied and 85 gave a positive answer to the defined question. The campaign draft reduced that sequence to 85% approval. The arithmetic was correct for respondents. In a presentation about the entire programme, however, the wording made it easy to hear a conclusion about all 400 participants. The 300 people who did not reply had disappeared from the story.
That disappearance was not a small reporting preference. It changed the population the evidence described. We knew something about the answers supplied by 100 people and nothing from that survey about the views of the other 300. The missing answers were not a reserve of approval waiting to be counted. They were also not proof of dissatisfaction. Both interpretations would have filled the same gap with an assumption that suited a particular argument.
The denominator is part of the meaning
I stopped the campaign statement and retained the fuller account: 85 positive replies among 100 respondents from 400 invitations. Some colleagues felt the revised wording placed too much emphasis on absence. They wanted to recognise the encouraging response we had received. I agreed that the response deserved recognition. Naming the denominator did not erase those 85 answers. It prevented us from asking them to represent a larger group without the evidence needed to do so.
The word approval also needed attention. The questionnaire concerned a defined aspect of the programme experience. It did not establish learning, application at work or a commercial return. A positive response to one question should not become a favourable answer to several unasked questions as the result moves through a campaign. Even if every participant had replied, the meaning of the question would still limit the conclusion. A complete response count would not turn satisfaction into a different outcome.
This is why I want the commercial team involved before the survey result reaches a slide. We should know which decision the research is intended to inform and how the proposed question relates to it. Otherwise, a convenient measure may become the answer to a business question it was never designed to examine. The problem is not solved by producing more attractive charts or by insisting that the percentage is mathematically accurate.
Unknown does not mean inconveniently negative
When we challenged the original statement, one reaction was to ask whether we were assuming that the silent participants disliked the programme. We were not. Nonresponse can arise for many reasons, and this case had not established which reasons applied. Treating the missing group as negative would be as unsupported as treating it as positive. The honest account needed a category for what we did not know, even if that made the presentation less conclusive.
I approved ₹23 lakh for further research intended to examine missing perspectives and questions the first survey had not answered. The programme remains unfinished in this account. That status is important. An approved research budget does not repair the earlier evidence by itself. It creates a responsibility to investigate, and its eventual findings may support, complicate or change the account. Marketing should not borrow certainty from work that has only been commissioned.
We also needed to preserve the difference between improving response coverage and improving the substance of the enquiry. More answers to the same question might make the account of participant reaction broader. They would not automatically establish workplace application or explain why a programme had produced a particular result. Different questions may need different methods and owners. My role was to keep the commercial decision connected to the enquiry, not to decide every methodological detail without the relevant expertise.
The cost was a campaign we could not publish as planned
There was a practical trade-off. The account team wanted to use the result while the programme was recent. A simple favourable percentage would have made an effective promotional line. Stopping it meant losing that moment and returning to colleagues with a more limited statement. The decision did not come with a measured increase in trust, a later sale or a retention improvement that could cancel out the inconvenience. Those outcomes were not established in the case.
I think it is useful to retain that discomfort. If every example of careful communication ends with an immediate commercial reward, leaders learn to justify accuracy only when it can be presented as a growth technique. Sometimes the reason to correct a statement is that it describes the evidence badly. The company still has to make commercial choices afterwards, and the corrected account may leave it with less certainty than the original slide suggested.
That does not mean research should become a reason never to act. Decisions often have to be made with incomplete evidence. The important distinction is between choosing with uncertainty visible and presenting uncertainty as though it has already been resolved. We could decide whether to continue a campaign, commission more work or narrow a proposition while acknowledging the limits. We could not honestly make the limits disappear by letting a respondent percentage stand for a different population.
A result may move from a research note into an executive summary, a sales presentation, a website and a partner's conversation. Each move creates an opportunity to shorten the wording. I want the population and the question to survive that journey. They are not ornamental footnotes. They explain what the number means and prevent a useful observation from becoming a broader endorsement through repetition. The shorter the statement, the more carefully its core meaning needs to be chosen.
For a board discussion, I would ask management to identify the population named in the conclusion and the population that actually supplied evidence. If they differ, the paper should explain why the inference is reasonable or where it remains uncertain. Directors can ask that question without pretending to be survey specialists. They can also ask whether approved further research is being reported as unfinished work or quietly used to imply that the initial claim has been validated.
The case has changed what I look for in a favourable result. I still want to know what respondents said, and I do not treat enthusiasm as suspicious. I also want to know whose voice is absent and which question has been answered. Those details make the result more useful because they show what the organisation can learn from it now. Eighty-five positive replies can inform a decision. They cannot supply the answers of 300 people who did not give them, however appealing the resulting percentage might look.