Business records can reveal an interesting aspect of a specific activity. Enquiry themes, product requests and documented operational changes may help explain an issue that matters to customers or a specialist audience.
The credibility of a data story depends on the fit between the records and the claim. Establish what the information can tell you, what it cannot and whether it is appropriate to share.
Start with a question the records can answer
A fictional events supplier might examine which services appeared most often in its recorded enquiries during a stated period. Those records could describe its own enquiries. They would not automatically describe completed bookings or the national market.
Keep that distinction in both the research question and the eventual headline. Records created to help staff answer calls may contain useful context alongside missing fields and inconsistent categories.
Check quality and comparability
Identify duplicates, incomplete entries and changes in collection. Decide how to treat them before calculating results, and document that decision.
When comparing periods, check whether the service range, marketing activity or recording process changed. A new form field can reveal a previously unrecorded topic without demonstrating a change in customer behaviour.
Use counts alongside percentages where they explain scale. A percentage without an underlying number can make a small change appear more substantial than it is.
Describe who or what is represented
State the source, period, inclusion criteria and number of records. Explain whether the unit is an enquiry, customer or order. Several enquiries from one customer are not several different people.
ONS guidance explains that survey estimates are affected by sampling and other errors. Internal records also require scrutiny of coverage and collection. A clear description does not make them representative of a wider population.
If a claim requires broader research, seek an appropriate research design and specialist advice. A convenience poll does not become nationally representative simply by attracting many responses.
Match the language to the evidence
Distinguish an observed pattern from its possible explanation. Several factors may account for a change in enquiries. Describe what the records show before proposing why it happened.
The fictional supplier could explain the mix of requests it received. Claiming that all businesses have changed their priorities would require a much stronger basis. Our guide to finding a supported PR story explores that distinction.
Prepare a method note
- The question and relevance of the records.
- The source and collection period.
- The inclusion criteria and number used.
- The handling of missing fields and duplicates.
- The calculation or comparison.
- Limitations and alternative explanations.
Ask someone familiar with the records to check the analysis and someone else to check whether the wording overstates it. Use authorised material and remove identifying information; small categories can still reveal an individual.
Offer the evidence with the angle
Make the method available and identify someone who can answer questions. Put limitations where they affect interpretation, rather than below an exaggerated headline.
If inconsistent records cannot support a numerical story, they may suggest a useful explanatory guide. Talk to Empty Lemon about an angle the available evidence can support.