A data analyst notices that a line chart showing monthly sales over the past two years has a steep drop in one month. Upon investigation, the analyst discovers that a new sales region was added mid-month and the data was not normalized. Which of the following best practices should the analyst apply to communicate this insight accurately?
Trap 1: Remove the month with the drop from the report.
Deleting the month conceals a real data-quality event and breaks the two-year trend, misleading readers about performance. It is tempting because removing outliers is legitimate when values are erroneous, and would be correct if the drop resulted from corrupt or duplicated records rather than an unnormalised region.
Trap 2: Use a bar chart instead to show the data.
Switching to a bar chart changes the visual encoding but leaves the unnormalised figures intact, so the misleading drop persists and the trend is lost. It is tempting because bar charts handle discrete comparisons well, and would be correct if the data were categorical totals rather than a continuous time series.
Trap 3: Present the data as-is and let stakeholders interpret the drop.
Presenting the unnormalised drop lets stakeholders infer a sales collapse that never occurred, since the added region distorted the monthly aggregation. It is tempting because raw, unadjusted data feels objective, and as-is presentation is correct when the metric is already comparable across periods and no composition change has occurred.
- A
Remove the month with the drop from the report.
Why it fails: Deleting the month conceals a real data-quality event and breaks the two-year trend, misleading readers about performance. It is tempting because removing outliers is legitimate when values are erroneous, and would be correct if the drop resulted from corrupt or duplicated records rather than an unnormalised region.
- B
Use a bar chart instead to show the data.
Why it fails: Switching to a bar chart changes the visual encoding but leaves the unnormalised figures intact, so the misleading drop persists and the trend is lost. It is tempting because bar charts handle discrete comparisons well, and would be correct if the data were categorical totals rather than a continuous time series.
- C
Normalize the sales data by region and explain the data anomaly in the report.
Normalising by region removes the distortion caused by the mid-month addition of a new region, so the steep drop reflects a reporting artefact rather than genuine sales decline. Documenting the anomaly in the report satisfies the accuracy constraint, preventing executives from misreading an unnormalised aggregate as a real performance drop.
- D
Present the data as-is and let stakeholders interpret the drop.
Why it fails: Presenting the unnormalised drop lets stakeholders infer a sales collapse that never occurred, since the added region distorted the monthly aggregation. It is tempting because raw, unadjusted data feels objective, and as-is presentation is correct when the metric is already comparable across periods and no composition change has occurred.