Data for education and non-profits

Separate learner or participant outcomes from administrative measures such as enrolments, attendance, processing times or expenditure. These can provide context, but do not by themselves prove learning gains or social change. Define the population, period and method behind each outcome.

Make definitions concrete by recording the owner, source, measurement point, exclusions and useful breakdowns. A term such as “active” or “available” often needs a sector-specific agreement. Also note missing data and any provisional interpretation. Teams can then check a figure before linking it to a decision.

Data about pupils, participants, clients, donors or staff is not automatically needed for every analysis. Define the purpose, limit access and granularity, and use aggregated or pseudonymised data where possible. Clarify which fields are needed and avoid reuse for another purpose without appropriate assessment.

Connect only data needed for the agreed question. Record source, refresh frequency, transformations and limitations. Check for missing values, duplicate records, unexpected shifts and changes to source fields. When data is not comparable, show that distinction rather than silently averaging it away.

Start with a question a team genuinely wants to answer, such as changes in participation across cohorts, without making individual records more widely available. Inventory definitions, periods and sources, discuss interpretation with subject-matter professionals and check that indicators are responsible and understandable.

A focused first step is often to choose one question, inventory relevant sources and definitions, check a small dataset and validate findings with users. Decide in advance who owns the work and which decision the result should support. Only then select a reporting format or further automation.