Data Prompts
Chart & Visualization Interpreter
Describe a chart, graph, or dashboard and get a structured interpretation of what it shows, what's notable, and what questions it raises.
Prompt
Interpret the following chart or data visualization. **Chart type:** [CHART_TYPE] (e.g., line chart, bar chart, scatter plot, heatmap, funnel chart) **What the chart shows:** [CHART_DESCRIPTION] (Describe axes, data series, time range, units, and any visible trends, outliers, or anomalies — or paste the underlying data if you have it) **Business context:** [CONTEXT] (What metric is being tracked and why it matters) Provide: 1. **What the chart shows** — a factual one-paragraph description of the pattern, trend, or distribution visible in the data. 2. **Key observations** — 3–5 specific, noteworthy findings (peaks, drops, inflection points, surprising comparisons, missing data). 3. **What's likely driving it** — plausible hypotheses for the pattern (2–3 candidates with brief reasoning). Be explicit that these are hypotheses, not conclusions. 4. **What to investigate next** — 2–3 follow-up questions or analyses that would either confirm or disprove the leading hypothesis. 5. **What the chart does NOT show** — limitations or confounds not visible in this view that could change the interpretation.
How to Use
Describe your chart in [CHART_DESCRIPTION] — axes, labels, what the lines or bars represent, any visible spikes or drops. If you're using Claude or another model with vision, you can attach an image directly and simplify the description. Set [CONTEXT] to explain what the metric means to the business.
Variables
| Variable | Description |
|---|---|
| [CHART_TYPE] | The type of chart or visualization |
| [CHART_DESCRIPTION] | Description of what the chart displays — axes, series, visible patterns |
| [CONTEXT] | Business context: what metric this is and why it matters |
Tips
- Models with vision (Claude, GPT-4o) can interpret chart images directly — paste or attach the image and simplify the description field.
- The "What the chart does NOT show" section is the most valuable part for preventing overconfident decisions — don't skip it.
- For dashboards with multiple charts, run this prompt on each chart separately, then ask: "Synthesize these interpretations into a coherent narrative about [BUSINESS_AREA]."