Back to Insights
Research and Data Analytics

Bad Data Can Make a Good Strategy Expensive

Imagine a management report showing that sales have increased. The business approves more spending, yet cash remains tight. On closer examination, the report measures invoices issued, while management assumed it reflected payments received.

The figures may have been calculated correctly. The decision still rested on a misunderstanding of what they represented.

A similar problem arises when a customer survey is presented as evidence of market demand, even though it reached only a narrow group. Strategy inherits the assumptions and limitations of the information behind it.

Define the number before interpreting it

Terms such as active customer, completed project and service turnaround time need agreed definitions. If departments use different rules, their reports can appear comparable while measuring different things.

Record the period covered, the source and the calculation behind each important indicator. Show the denominator behind a percentage and distinguish recorded activity from the outcome the organisation wants to achieve.

The UK Government Data Quality Framework describes dimensions including completeness, uniqueness, consistency, timeliness, validity and accuracy. They offer useful questions for organisations assessing whether information is fit for a particular decision.

Prioritise the checks that matter to the use. An outdated address and a duplicated revenue entry have different consequences, even when both count as data quality problems.

Ask whose experience the research represents

A large number of responses can still leave important groups out. Consider how participants were recruited, who could participate and whose views may be missing.

AAPOR's survey guidance stresses the relationship between the sample and the population being studied. A survey distributed only through an organisation's online channels may provide useful feedback from participants, while offering limited evidence about people those channels do not reach.

Explain these limits in the findings. Identify the population, collection method and response base, and assess how far the conclusions can reasonably extend. Where a decision carries substantial consequences, consider whether additional evidence is needed.

Make the evidence traceable

Keep a record of how data moved from collection to the final report, including corrections and exclusions. Assign an owner to important datasets and maintain a short dictionary of key measures.

Separate what the evidence shows from the explanation proposed for it. If performance improves after an intervention, consider other changes that could have contributed before claiming that the intervention caused the improvement.

For each major recommendation, identify the supporting evidence, the uncertainty and what further information could change the decision. This gives managers a way to judge how much confidence to place in the analysis.

Well-governed data makes it possible to challenge a conclusion constructively, reproduce the calculation and improve the next decision.

Talk to DCS about strengthening research design, data quality and the reporting behind your decisions.

Start the Conversation

Sources

Ready to Strengthen Your Digital Resilience?

Speak with Data & Cybersecurity Limited about your cybersecurity, data protection, AI governance, analytics, or digital transformation priorities.