Before the Slide Speaks: A Chart Integrity Review for UK Board Packs

UK board packs are full of charts that are factually correct and argumentatively incomplete. A line can be plotted from audited data and still encourage a conclusion the data does not support. The reason is rarely dishonesty. It is convention: default software settings, inherited templates and the pressure to make a slide legible from the back of the room. As organisations approach the end of 2026, with planning cycles compressed and budgets contested, the geometry of a chart deserves the same scrutiny as the number in its title.

At Jordan Research, we treat the chart as a decision instrument. The question is not whether it looks professional, but whether it gives the board a fair basis for the decision in front of it. A short chart integrity review, run before the pack is circulated, can catch most of the recurring problems.

The chart is an argument with a scale

Every chart makes choices that the data alone does not make. Where the axis begins, whether the series is indexed, how many scales are overlaid and which cells are coloured are all editorial decisions. They are legitimate decisions, but they are not neutral. A truncated axis can turn a two-point share movement into a dramatic climb. An index chart can make a stable market look turbulent, or a turbulent one look stable. A dual-axis exhibit can imply a relationship between two series that the evidence does not establish.

The practical test is simple: can a reasonable board member read the chart and state the decision it supports without being misled about magnitude, direction or uncertainty? If not, the chart is not ready for the room.

Five recurring geometries to inspect

1. The truncated vertical axis. Bar charts should normally start at zero because the length of the bar is the visual message. A market share chart running from 40% to 44% exaggerates a modest movement; the same data on a zero baseline may look almost flat. Line charts can legitimately use a narrower range to show trend, but the break should be labelled and the absolute change stated in the title or annotation. The audit question is not whether truncation is allowed, but whether the audience can see that it has happened.

2. The index-to-100 chart. Indexing is useful for comparing series with different bases, but it removes the denominator from view. A tracker moving from 100 to 104 could represent a four-point rise on a small base or a much larger absolute shift on a large one. Before the board sees the index, ask for the underlying values and the base sizes. If the absolute movement is material to the decision, put it on the slide rather than in an appendix.

3. The dual axis. Two vertical scales allow almost any pair of series to be made to look related, because the crossing points are chosen by the chart builder. For board use, dual axes should be exceptional. Where two measures genuinely need to be compared, small multiples with a shared time axis are usually clearer. If a dual axis is retained, the chart should say why the two scales are comparable and should not imply causation through proximity alone.

4. The stacked bar with a moving base. Stacked bars compare parts of a whole, but they become difficult to read when totals vary across periods or segments. A category that looks larger may simply sit on a bigger base. For composition questions, a 100% stacked bar can work well, provided the base is stable and the shift in share is the point. For magnitude questions, use separate bars or a table. The audit should confirm that the visual comparison matches the question being asked.

5. The heatmap that hides its sample. Colour scales are persuasive. A deep red cell can signal crisis before the reader has checked whether it contains twelve respondents or twelve hundred. Heatmaps in UK tracker decks should show the base size per cell, or at least flag cells below a stated threshold. Colour should be used with restraint, should work for readers with colour-vision deficiency, and should not imply precision where the sample is thin. Missing data should be visually distinct from a zero or a neutral score.

Redraw tests before the meeting

A useful discipline is to redraw the two or three exhibits that will carry the most weight in the discussion. Take the market share chart and plot it on a zero baseline. Take the tracker index and add the absolute values beneath it. Take the heatmap and grey out cells below the minimum base. The exercise is not about producing prettier slides. It is about discovering whether the storyline survives a change in geometry.

If the conclusion changes when the axis changes, the board is not looking at a finding. It is looking at a framing choice. That does not make the finding invalid, but it does mean the framing should be disclosed and debated.

A ten-minute checklist for the insight lead

Before a deck is circulated, run a short review against the following questions:

1. What decision is this chart meant to support, and does the title state the finding rather than the topic?
2. Does the vertical axis start at zero where bar length carries the message, and is any truncation clearly labelled?
3. Are base sizes, denominators and weighting information available for every headline chart?
4. If the chart is indexed, what are the underlying absolute values and the base period?
5. Is a dual axis genuinely necessary, and has the reason been stated on the slide?
6. Do stacked bars compare like with like, or are changing totals distorting the visual comparison?
7. Does the heatmap show sample sizes, handle missing data clearly and use an accessible colour scale?
8. Are fieldwork dates, source and any modelled or synthetic components disclosed?
9. Would a sceptical reader be able to reconstruct the main conclusion from the chart and its footnotes?
10. Has the chart been redrawn under one alternative convention to test whether the story is robust?

The audit is a governance habit, not a design review

Chart integrity is not a matter of taste. It is part of the evidence trail that boards rely on when they approve a budget, change a supplier or enter a market. The insight lead does not need to win every argument about chart design, but the team should be able to explain why each convention was chosen and what it does to the reader's interpretation.

A short record of those choices, kept alongside the data pack, is often enough. It signals that the chart was treated as a decision, not a decoration. In a year when UK boards are being asked to make consequential calls with imperfect evidence, that discipline is a small but useful form of assurance.