Representation
Do examples cover the people, periods, cases or conditions where the analysis will be used?
Data · Readiness checklist
How can I tell if my data is ready for analysis or AI?
More data does not always mean better results. This guide helps assess whether information has enough context, quality and coverage for the question.
Before you start
Readiness depends on the question, not a universal score.
A dataset is ready when it can be connected to a specific task and its limitations are understood. Requirements differ when describing results, comparing groups or training a predictive model.
At T2G: we first connect the question to available sources. Then we review what evidence exists, what is missing and what can responsibly be concluded before proposing analysis or a model.
A practical review
Check these four things before modelling.
A review may find that data is ready, needs preparation or requires better examples to be collected.
- 01
Question and outcome
Define what you want to understand or predict, for whom and how a useful answer will be recognised.
Conceptual schematic01 / 04 - 02
Source and meaning
Confirm where each field comes from, what it means, when it was recorded and whether it can be used for this purpose.
Conceptual schematic02 / 04 - 03
Quality and coverage
Review missing values, duplicates, unusual ranges, time periods and groups that may be underrepresented.
Conceptual schematic03 / 04 - 04
A way to evaluate
Check whether enough examples exist to examine results and keep an evaluation separate from model fitting.
Conceptual schematic04 / 04
Signals to check
Quality is more than a table without empty cells.
Reliable labels
If a model learns from known categories, are they consistently defined and reviewed by someone familiar with the task?
Traceability
Can you explain how data was generated and avoid including information that would only be known after the outcome?
Judgement before automation
Having data does not mean you should use AI.
- 01
A clear metric, query or descriptive analysis may answer the question better.
- 02
Permissions, privacy and retention need review based on the data source and purpose.
- 03
Readiness does not prove a model will work; that requires evaluation against agreed criteria.
Next step
Explore the service and keep learning.
Talk about your challenge
Have a question and some data, but unsure whether it is enough?
We can help review sources, limitations and a realistic way to evaluate the next step.
Book a diagnostic session directly with the partners