The same data supports several tasks; draw a chart for each

What it is

Task-specific views over shared data provide a chart or view state suited to each reading operation when one dataset supports different questions. Sales data can support category comparison, temporal trend, distribution, and variable relation, but these tasks need different aggregation, order, scale, and detail. “A chart for each” may mean juxtaposed small multiples, coordinated views, or switchable modes—not a new static chart for every question.

Why it happens

One dataset is not one analytical object: field roles, filters, aggregation units, and reference baselines change with task. Packing every task into one view makes encodings and interactions compete; mechanically splitting into many charts adds search, switching, and correspondence cost. Effective multiple views give each view a primary task while reusing stable entity identity, filter state, and necessary scales, diverging only where task semantics require it.

Studying it

Derive task sequences from real workflows rather than listing isolated operations, then compare one general-purpose chart, several task-specific charts, and a switchable view. Measure subtask accuracy, time, view transitions, backtracking, omission, learning, and final decisions, noting when juxtaposition is necessary. Vary task order and data state to test whether coordination preserves selection, filtering, and semantics. Evaluate the complete path for compound tasks instead of merely summing isolated scores.

Where it stops holding

One chart may suffice when tasks share the same reading operation; splitting can increase mental integration when trend and composition must be combined simultaneously. Exploration can create and discard views dynamically rather than anticipate every question, while explanatory output should remove views without audience demand. Cross-view consistency does not require every domain to match: direct magnitude comparison needs shared scales, whereas local shape or standardized comparison can use clearly disclosed independent domains.

Applying it

• Build a task–data transformation–view matrix. Merge tasks sharing a reading operation and create separate views only for conflicting aggregation, order, or encoding needs.
• Use juxtaposition with synchronized entities, filters, and time range for direct comparison. For sequential exploration, provide named modes, a clear return path, and state preservation.
• State measure definition, unit, domain, and answer scope in every view. Do not let common color, position, or scale imply comparability across different semantics.
• Make view switching, current mode, filter state, summaries, and data tables available by keyboard and screen reader; validate the end-to-end compound task, not charts in isolation.

Related

• Same group: U2.01.1 Comparison, trend, distribution, composition, and relation each map to different charts · U2.01.2 Define the question before choosing the chart
• Adjacent: U2.15.1 Small multiples replace overlay with repeated like-for-like panels · U1.11.1 The same entity keeps the same colour and position across a chart series
• Search terms: task-specific views · coordinated multiple views · task sequence