The Cut That Changes the Story: Cross-Tab Discipline for UK Board Packs
The summary is not the finding
Board packs reward brevity. A single percentage, a clean chart and a confident recommendation will usually travel further than a page of caveats. That is understandable: directors are time-poor, and research needs to be usable. But the pressure to compress evidence into a headline creates a familiar risk. The number that reaches the board is often the least decision-relevant number in the dataset.
Top-line results are necessary. They tell the board what the average respondent said. They do not tell the board whether that average is stable, whether it is driven by a particular group, or whether it conceals a contradiction that matters. The most useful finding is frequently buried in a cross-tab that nobody requested — the second question that follows the headline.
Why sub-groups matter more than they appear
Aggregates smooth out heterogeneity. In UK consumer and B2B samples alike, attitudes and behaviour vary by region, firm size, sector, tenure, usage frequency and role in the buying committee. A headline satisfaction score of 60% could reflect a stable 60% across the base, or 80% among one group and 40% among another. The two datasets support very different decisions.
There is also the familiar problem of contradictory responses. A survey can show high stated satisfaction alongside low advocacy, or strong intent to renew alongside declining usage. Individually, each measure looks reassuring. Together, they suggest that the headline is measuring something narrower than the board assumes. Cross-tabs help expose those tensions before they become commercial surprises.
A worked example: the expansion that looked obvious
Consider a UK professional services firm that surveyed 800 clients after introducing a new account management model. The top-line result was encouraging: 61% of clients said they were satisfied with the new model. The board paper recommended a national rollout, with a phased timetable and a modest increase in account management headcount.
The second question changed the recommendation. When responses were cut by client revenue band, satisfaction among small clients was 74%, among large clients 58%, and among mid-market clients — those spending between £25,000 and £250,000 a year — just 43%. The mid-market group represented 55% of revenue and most of the firm's recent growth. A further cut by contact frequency showed that the most engaged clients were also the least satisfied, which cut against the assumption that familiarity was breeding contentment.
The headline was not false. It was simply an average that masked the segment the firm could least afford to disappoint. The board paused the rollout and asked for a diagnostic review of the mid-market service model. The finding that changed the decision was not hidden in a complex model. It was in a standard cross-tab that had not been asked for.
The discipline of the second question
The lesson is not that every board pack needs dozens of cuts. It is that the cuts should be chosen deliberately, before the headline is written. A practical habit is to ask, at the commissioning stage: which sub-group result would cause us to change our decision? That question forces the board and the research team to identify the cuts that matter, rather than discovering them after the presentation.
Discipline also means respecting sample sizes. A UK survey of 800 respondents may support robust national estimates but not fine-grained cuts across every region and sector. Sub-groups of 30 or 40 respondents are noisy. Reporting the base size and an indication of uncertainty is not pedantry; it prevents a small-cell anomaly from being treated as a strategic signal. Significance testing helps, but it should not become a veto on judgement. Direction, consistency across related cuts and plausibility all matter.
A short checklist for board reporting
- State the decision the evidence is meant to inform.
- List the pre-agreed cuts, with a one-line rationale for each.
- Show the base size for every sub-group reported.
- Flag any cut where the direction reverses, and explain the commercial significance.
- Pair attitudinal measures with behavioural or operational measures where possible.
- Label exploratory cuts clearly, and keep them separate from confirmatory findings.
- Include a one-line decision implication for each material cut.
- Put full tables in an appendix, but keep the material cuts in the main pack.
- Record what would change the recommendation.
- Assign an owner to follow up contradictions before the next board cycle.
Reporting without clutter
A one-page cut register can sit alongside the headline summary. It need not be elaborate: a short table showing the cut, the base size, the result and the implication is usually enough. A traffic-light convention can help directors see at a glance where the aggregate holds and where it does not.
For example: overall satisfaction is 61%. However, among mid-market clients (n=142), satisfaction is 43%, and this group accounts for 55% of revenue. The recommendation is therefore conditional. That is decision-grade reporting. It respects the headline without being captured by it.
Conclusion
Top-line results start the conversation. Cross-tabs finish it. For UK boards, the second question should be routine rather than exceptional: what does this look like by segment, and does that change the decision? The cost of asking is small. The cost of acting on an aggregate that conceals the decisive cut can be considerable.
