Insight
Good Enough Data Can Be Surprisingly Valuable
Waiting for perfect inputs can mean postponing useful opportunities. Start by asking what the next decision actually needs.
“We need better data first” can be a sensible observation. It can also become a reason no customer-facing idea ever reaches a test.
The useful distinction is between data that is imperfect in general and data that is not reliable enough for the decision in front of you. Those are not the same thing.
Define good enough for the use case
Start with the action the data will support. A broad analysis of customer behavior, a message to a specific person and a change to an account all carry different consequences when the information is wrong.
Ask what needs to be known, how current it needs to be and what happens when it is missing. Quality becomes more concrete when it is tied to a use, rather than expressed as an ambition to clean everything.
Use the information that answers the question
Consider a team exploring whether customers who recently completed a first purchase would find a short product guide useful. As a hypothetical first step, the team might need a reliable purchase event, an appropriate way to contact those customers and a clear way to observe the response.
It may not need a complete lifetime-value model or every historical interaction joined into one record. Adding those requirements before the team has learned whether the guide is useful could make the experiment larger without making the first question clearer.
A useful data-quality boundary
- Purpose
- What decision will this support?Use a specific action or learning goal to determine which information matters.
- Reliability
- What can we reasonably trust?Identify gaps, freshness and uncertainty rather than hiding them.
- Consequence
- What happens when it is wrong?Keep the first use proportionate to the consequences of a mistake.
Imperfect is not the same as careless
Good enough is not permission to ignore uncertainty, use data without an appropriate basis or send messages to the wrong people. Some gaps rule out a particular use. Others can be handled through a narrower audience, a manual check or an explicitly limited experiment.
Make those choices visible. A modest data set with understood limitations can support a more defensible decision than a large one whose meaning is assumed.
Let the work reveal the next improvement
A focused use case can make data work easier to prioritize. Instead of an undifferentiated cleanup backlog, the team can see which missing field, unreliable event or disconnected system is preventing a useful action.
That does not remove the need for a sound foundation. It gives the next foundation improvement a purpose.
The goal is not to celebrate imperfect data. It is to avoid making perfection a condition for every useful step.
