How to set data quality rules in DataLens
A quality problem you found once will come back. A rule catches it every time, without anyone remembering to look.
What you will do
You will turn expectations about your data into rules that are checked rather than assumed, and see which datasets meet them.
Quality Rules is a Pro capability, along with the Dictionary and Column Naming tabs it sits beside.
When this is useful
- A quality problem has recurred and you want it caught automatically next time.
- Data feeds a report or a decision and silent degradation is unacceptable.
- You need to show that quality is monitored, not just occasionally checked.
- Several people load data into the same place and standards need to be explicit.
Before you start
- A Pro entitlement — Quality Rules, Dictionary and Column Naming carry a Pro badge on the Govern tab row. Without it the tab shows an upgrade prompt.
- A profiled dataset — Profile first. Rules should encode what you have actually observed, not what you assume.
- Explicit expectations — What should never be null, what format a field must take, what values are valid.
Steps
Profile the dataset first
Open Govern → Quality & Lineage → Profiling and look at the columns you intend to write rules about.
Writing a rule that current data already fails is a common and frustrating way to start. Look before you write.
Open the Quality Rules tabPro
Select Quality Rules from the Govern tab row. It is marked Pro.
The Quality Rules screen, or an upgrade prompt if the workspace is not on Pro.
Define rules on the columns that matter
Write rules for the columns whose quality actually affects something — the join keys, the amounts, the identifiers.
A rule on every column produces noise nobody reads. A rule on the ten that matter produces signal.
Record expected formats in the dictionaryPro
The Dictionary tab holds each column's Business Label, Business Meaning, Domain and expected Format / Pattern.
The dictionary is where the expectation is written down; the rule is where it is enforced. Both are Pro.
Standardise naming while you are herePro
Column Naming applies a naming convention across your datasets, so the same concept is not called three things in three places.
Assign stewardship
The Stewardship tab records who is responsible for a dataset, and Approvals covers sign-off.
A rule with nobody accountable for its failures is a rule that fails quietly.
What happens next
Rules and lineage complement each other: rules tell you something is wrong, lineage tells you where it came from.
Once quality is enforced, publishing and data products both rest on a firmer footing.
Example
A finance dataset feeds a monthly board pack. Rules require that account_id is never null and amount is always numeric. When an upstream export changes format, the rule fails before the board pack is built rather than during the meeting.
Tips
- Profile before writing rules, so they encode reality rather than hope.
- Rule the columns that matter, not all of them. Alert fatigue is a real failure mode.
- Write the expectation in the dictionary as well as the rule — the rule enforces it, the dictionary explains it.
- Name a steward. Accountability is what makes a failing rule get fixed.
Limitations
- Quality Rules, Dictionary and Column Naming are Pro capabilities and are marked as such in the product.
- Rules check data inside DataLens. They do not enforce anything in your source systems.
- A rule reports a failure. Fixing it is a separate action.
- Rules apply to the datasets you define them on.
Related how-to guides
Profile a dataset and find PII
See what is really in each column, and find the personal data before it ends up somewhere it should not.
Trace where a column came from
Follow a field back from the dataset you are looking at to the source it came from, step by step.
Clean a messy dataset
Drop nulls, fill gaps, remove duplicates and cap outliers — with a live preview before anything is applied.
Build a data pipeline
Turn a sequence of steps into a repeatable pipeline with a defined execution order.
Related questions
More of these on the DataLens FAQ page.
Does DataLens support data quality rules?
Yes, on the Pro tier. The Quality Rules tab under Govern lets you define expectations that are checked rather than assumed, alongside a Dictionary for recording what each column means and its expected format, and Column Naming for consistent naming. All three are marked Pro in the product.
What is the difference between profiling and quality rules?
Profiling tells you what your data looks like right now — a one-off observation. A quality rule encodes an expectation and checks it every time, so a problem you found once is caught automatically when it recurs.
Try this in DataLens
DataLens is in private beta. Request access and work through this guide on your own data.
Request beta access