Data → Customer Behavior → Data Visualization
How Charts Can Mislead
The same underlying numbers can tell very different stories depending on how they're charted. Each example below pairs a technically-accurate-but-misleading version against a better, honest version of the same data.
1. Truncated Y-Axis
Why this misleads
Both charts plot the identical numbers — Desktop 30.4%, Tablet 28.8%, Mobile 26.4% — but truncating the y-axis to a narrow 20%–32% band inflates the visual size of the gap between bars, making a real difference of 4.0% percentage points look like Mobile converts at roughly a third of Desktop’s rate, when the honest, zero-baseline chart shows all three devices convert in a fairly similar range. Axis truncation is one of the most common ways a real but modest effect gets visually exaggerated into a dramatic one.
2. Inappropriate Aggregation (Volume vs. Rate)
Why this misleads
Raw total revenue conflates channel volume with channel quality. Organic Search tops the revenue chart with ₹1,73,56,089, but its conversion rate is only 27.0% — ranked #3 of 6 by rate. It leads in revenue mainly because it drives the most sessions (13,449, the highest of any channel — confirmed as Organic Search). Meanwhile Email converts best per session (31.3%) but ranks lower on total revenue because it has fewer sessions. In this dataset the ranks genuinely diverge: a channel can look like the “best performer” on a totals chart purely by having more traffic, while converting that traffic at a below-average rate. The normalized, per-session metric is the appropriate one for judging channel quality; raw totals are the appropriate one for judging channel scale — they answer different questions.
3. Distorted Categorical Comparison
Why this misleads
Upper-Middle customers (32.7%) are only about 2.04× as numerous as Low customers (16.0%). But human perception of a circle’s size is dominated by its area, not its diameter. Scaling diameter directly to the proportion (the misleading version) makes the area ratio blow up to roughly 4.17× — visually implying Upper-Middle customers outnumber Low customers by far more than the real 2.04× gap. The better version scales the circle’s area (so diameter grows with the square root of the proportion), so the visual size ratio matches the true data ratio. This same distortion appears whenever bubble charts, icon arrays, or 3D pie/donut charts scale a linear dimension (radius, height, side length) directly to a data value instead of scaling area to value.
4. Cherry-Picked Comparison
Why this misleads
Picking only the single lowest month (Jun 2024, ₹26,53,545) and the single highest month (Oct 2024, ₹44,88,861) and presenting them as adjacent bars manufactures a growth story of +69.2% that the two points alone cannot support — it ignores every month in between and implies a sustained upward trend. The full 24-month series tells a different story: revenue fluctuates in a fairly narrow, seasonal-looking band around ₹32,25,399 per month, Oct 2024 stands out as a single spike (likely a seasonal or promotional event) rather than the start of a trend, and revenue from the first month to the last month actually changed by only +1.9%. Cherry-picking two non-adjacent, favorably-chosen points is one of the clearest ways to imply a trend that doesn’t exist in the full data.
These three examples use the same real, precomputed figures as the rest of this site — only the presentation choices differ. The lesson generalizes: always check the axis baseline, prefer normalized (rate) metrics over raw totals when comparing groups of different size, and look at a full time series before trusting a two-point comparison.