Validate the data before asking for a convincing summary
A polished report can count repeated rows, turn a blank into zero or mix currencies. Define the rules first, then distinguish what you can measure from what still needs source evidence.
The destination is in Spanish. This is a waitlist, not guaranteed immediate access or a free plan.
Situations to review
A sales report needs a total, but one row has no amount.
Count amounts present against all rows and preserve blanks as missing. Identify the source and owner needed to resolve them instead of inventing values or substituting zero.
Readers see the denominator and limitation. A partial subtotal is not presented as a complete total.
The same ID appears twice before sales are aggregated.
Identify the repeated group and distinguish rows, distinct IDs and rows whose ID occurs exactly once. Review whether repetition is an error, legitimate line items or an unsuitable identifier rule.
An alert does not delete records or choose which one to retain. Proposed corrections need evidence and approval before changing a system.
A date is impossible and the dataset contains USD and COP amounts.
Check formatting and calendar validity separately and keep currencies apart. If conversion is required, agree the method, source and date before calculating.
A parseable date does not establish the true transaction date. Without an approved conversion basis and sufficient data, there is no reliable combined total.
How to evaluate the work
- Define the purpose, period and unit of observation. Document required fields, types, uniqueness by ID or composite key, date rules and currencies. Completeness, uniqueness, consistency, timeliness, validity and accuracy are separate dimensions, not one universal quality score.
- Preserve the original and inspect an authorized copy with only necessary data. A collaborator can organize sources and proposals using their computer, groups and context, a compatible model and authorized tools. Do not assume a specific integration or autonomous corrections; external actions require approval.
- Publish each formula and denominator. Distinct IDs divided by rows and rows with non-repeated IDs divided by rows are different measures. Date validity needs the agreed rule and calendar check; accuracy needs source evidence. Do not combine dimensions into a global percentage without a justified method.
- Template: field/rule/unit/denominator/flagged row/source/owner/decision. Fictional CSV, with Spanish field names preserved for reproducibility: id,fecha,moneda,importe V001,2026-11-01,USD,100 V002,2026-11-02,USD,200 V002,2026-11-02,USD,200 V003,2026-02-30,COP,300 V004,2026-11-04,USD, Local standard-library checks found amounts present in 4/5 rows and calendar-parseable dates in 4/5 rows. Row 4 has an invalid date; row 5 has a missing amount. There are 4 distinct IDs among 5 rows (80%), but only 3/5 rows have an ID occurring exactly once (60%); rows 2 and 3 form the repeated group. One excess repetition does not identify which row to delete. Do not add USD and COP together. Accuracy and timeliness were not measured. These are checks on fictional data, not customer results or an integration.
Download the fictional dataset and expected results
Extracted from the existing 13 November fixture. Spanish headers and values are preserved for reproducibility. This is not a new test or a Think² integration.
Frequently asked questions
Does a valid format prove a value is accurate?
No. Validity checks a rule; accuracy requires evidence that the value matches reality. A real calendar date can still be the wrong transaction date.
Should I delete a row with a repeated ID?
Not automatically. Repetition can reflect an error or a different unit of observation. Investigate the identifier and source, preserve originals and document approved corrections.
Can I add USD and COP into one total?
Not as though they were one currency. You need a conversion policy with a source and date, as well as resolved missing values and repeated records. Otherwise show separate groups and limitations.
Sources and scope
- UK Government — The Government Data Quality Framework — Data-quality dimensions and fitness for purpose; a methodology, not dataset certification.
- UK Government — Data Quality Framework guidance — Prioritizing and measuring quality rules; not proof of a Think² integration.
Keep exploring
The destination is in Spanish. This is a waitlist, not guaranteed immediate access or a free plan.