6 quality dimensions — score your dataset 0–100% on each
Do data match reality? Check: sample verification against source (10–20 records), known control sums.
Are required fields filled? Metric: % empty per critical column. <95% on keys — a problem.
Do systems contradict each other? CRM sum = accounting sum; same status in both systems.
How fresh is the data? Metric: max record age vs process need. 'Current' contacts 3 years old are dead.
No duplicate entities? One customer = one record. Metric: % duplicates by key.
Do values match format/range? Emails have '@', dates are real, sums >0, phones formatted.
80%
Acceptable (70–85): fine for operational decisions; clean up before financial reporting
Scores are subjective without metrics — where possible, compute real % (see hints). Your weakest dimension is your main risk: e.g., 60% completeness at 95% accuracy means a third of the picture is missing.