Meridian Home Goods: The Investor Dashboard That Makes 0.1 Point Look Like Doubling
Meridian Home Goods sells through 14 regional e-commerce sites. The marketing team has produced a Q3 dashboard showing revenue as bars and conversion rate as a line on a secondary axis. The conversion axis starts at 2.0% and ends at 2.2%, so a region that moved from 2.0% to 2.1% looks like a doubling. One panel is a 3D pie chart with 11 slices, three of which are below 1% and labelled only in the legend. Red and green are used for 'below target' and 'above target' without text labels. The dashboard also ranks regions by conversion rate without showing order counts; the lowest-volume region had 37 orders. The CMO likes the dashboard because 'it makes the story obvious' and wants it in the investor pack. The analyst who built it has been asked to make the charts 'more visual' and to remove the footnotes because they clutter the slides.
1. Critique the visual encoding choices and explain the perceptual or statistical problems each creates.
Read the worked answer
Several choices mislead:
- Truncated secondary axis (2.0–2.2%). Starting a line axis near the data range magnifies a 0.1-point change into a visually large climb. The encoding implies a doubling, but the underlying change is 0.1 percentage point. This is a scale distortion, not a data transformation.
- Dual axes. Two axes let the designer choose scales that create an apparent relationship between revenue and conversion. Readers cannot judge the true correlation, and the crossing points are artefacts of scaling.
- 3D pie with 11 slices. Angle and area are already weak encodings; 3D perspective distorts them further. Three slices below 1% cannot be compared, and the legend forces a look-up that working memory does not handle well.
- Red/green without text. Colour alone fails for colour-vision deficiency and for anyone reading a printed or grayscale copy. 'Above' and 'below' should be labelled.
- Ranking without denominators. A region with 37 orders can top the conversion ranking by chance. Ranking without counts or uncertainty invites overinterpretation.
The dashboard optimises for a story, not for accurate comparison.
2. Redesign the key comparison for an investor audience: specify chart types, axes, ordering, labels, and uncertainty display.
Read the worked answer
For an investor audience, lead with the decision-relevant comparison: revenue and conversion by region, with volume shown. Use a sorted horizontal bar chart for revenue, ordered by value, with region labels and values at bar ends. Show conversion as a dot plot or a bar chart on a zero-based axis, with the target as a reference line. Do not put revenue and conversion on twin axes; use two aligned small multiples sharing the same region order. Annotate the absolute change in percentage points, not just the ratio. For the lowest-volume region, show n and a confidence interval or a caution marker; if intervals are too wide, aggregate small regions into an 'other' group. Use a colour-blind-safe palette and add text labels for above/below target. Put definitions and caveats in a short note directly on the page, not in a separate appendix. The investor takeaway should be a sentence with numbers: 'Conversion rose 0.1 percentage point in 9 of 14 regions; the three smallest regions are not reliably ranked.'
3. How should the analyst respond to the request to remove footnotes and make charts 'more visual'? Address the ethical and professional dimension.
Read the worked answer
The request to remove footnotes and make charts 'more visual' should be resisted in part. Footnotes carry definitions, exclusions, and data-quality caveats; removing them turns a chart into an assertion that cannot be checked. 'More visual' often means more decorative, and decoration can reduce accuracy. The analyst should propose a compromise: move essential caveats into short, visible subtitles or labels on the chart, keep a compact source/method note, and remove only non-essential clutter. Explain the risk in business terms: if the investor pack overstates a 0.1-point change as a doubling, the company may face a credibility problem when the underlying data is examined. Document the disagreement and the rationale. Professional integrity requires not knowingly presenting a distorted comparison, even under pressure from a senior stakeholder.