You’ve Got the Data. So Why Can’t Anyone Answer a Simple Question?

“How many requests are waiting for a response?”

It should be a simple question. There is a case-management system, a spreadsheet used by the team and a monthly report sent to managers. Yet the three sources give different answers. Before anybody can explain the position, someone has to work out which requests count, when the figures were last updated and whether the spreadsheet includes work that is paused.

That is an illustrative situation, not a description of a particular Processive client. But the underlying difficulty is familiar to anyone who has tried to run a service using information assembled from several places. The data exists. A trustworthy answer is harder to find.

The same question can produce different answers

Imagine two people reporting the number of open requests. One counts every record that has not been formally closed. The other excludes requests waiting for a customer, a supplier or a decision outside the team’s control. Both are using the information available to them. Neither is necessarily making a mistake.

The disagreement sits in the definition of “open”. If that definition has never been agreed, a more attractive dashboard will simply display the disagreement more neatly.

Other apparently ordinary words can cause similar trouble: completed, overdue, active, referred, resolved, available. Every team develops sensible local interpretations. The difficulty appears when those interpretations are brought together and treated as if they are the same.

This is why data quality is about more than removing duplicate rows or fixing date formats. The Government Data Quality Framework describes quality in terms of whether information is fit for its intended purpose. That is a useful test for any organisation, not just government.

Five things a reliable answer needs

Before trusting a figure in a meeting, a leader should be able to establish five things:

1. Meaning: What exactly are we counting? What is included, and what is excluded?
2. Source: Which system or record is authoritative? Are there other sources that change the answer?
3. Timing: When was the information last updated, and what period does it describe?
4. Ownership: Who can explain the figure and resolve errors or exceptions?
5. Use: Is this information sufficiently reliable for the decision we intend to make?

None of those questions requires expensive technology. They require an understanding of the work, the people doing it and the information produced along the way.

An answer with a clear definition and a known limitation can be more useful than an apparently precise figure with no explanation. The important thing is to know what the number can — and cannot — tell you.

Why the reporting team ends up holding everything together

The problem often begins long before anyone opens Excel or Power BI. A frontline team records something because it helps them manage today’s work. Another team needs the same information for a different purpose. Someone exports a report, adds a few columns and keeps a local tracker so the process will still work.

At each point, a practical workaround fills a gap. Over time, those workarounds become part of the operating model, even when they are not written down. The report depends on steps that sit outside the system.

This creates a particular kind of fragility. A colleague who understands all the exceptions can produce a dependable management report. Their successor may inherit the files without knowing why certain records are excluded or which numbers need checking.

If a board report is only reliable because one person spends Friday afternoon reconciling it, that reliability belongs to the person rather than the process.

The answer is not to criticise the people keeping things together. Their workarounds are often a good place to discover what the system and process are failing to provide.

Start with one question, not a business-wide clean-up

It is tempting to start by asking for a complete inventory of systems, spreadsheets and documents. That may be necessary for some projects, but it can make a manageable problem feel enormous.

A better first move is to choose one question the leadership team regularly struggles to answer. Pick something that affects a decision: how long work is waiting, what actions are overdue, how much capacity is available or which cases need attention.

Then follow the answer back to its origins.

1. Write down the question. Ask the people who use the answer what decision they will make from it. This gives the work a purpose.
2. Reconstruct the current answer. Identify the systems, spreadsheets, manual checks and conversations needed to produce it.
3. Agree the definition. Decide what is included, which date matters and how exceptions should be treated.
4. Fix the problem where it starts. If missing fields or inconsistent rules cause the reporting work, change the collection or workflow where practical instead of adding another checking step.
5. Make someone accountable. Agree who maintains the definition, checks the exceptions and reviews whether the answer is still suitable.

The government’s supporting data quality guidance similarly recommends tracing problems through the data journey and fixing them as close to the source as possible. It also advises weighing the cost of fixing an issue against the cost of leaving it alone.

What a decision-ready answer looks like

Return to the original question about requests waiting for a response. A useful answer might include a number, but it should also say what the number represents, when it was taken and whether anything material is missing.

The detail is not bureaucracy for its own sake. It lets somebody make an informed decision, spot when the information has changed and challenge the answer constructively.

It also makes it easier to produce the answer next time without reconstructing the same logic from scratch.

Do we need to replace the software?

Sometimes. But software should not be the first assumption.

If two departments use different definitions, moving both into one platform will not resolve the disagreement. If a critical field is never completed, a dashboard cannot make it accurate. If nobody owns exceptions, automation can move questionable information faster without making it more useful.

Many improvements begin with agreed definitions, better capture rules, fewer duplicated steps and a clear owner. Existing Microsoft 365, Google Workspace or line-of-business systems may already have enough capability to support those changes.

There are circumstances where integration work, a new application or specialist data architecture is justified. The decision should follow a clear understanding of the need, the available options and the ongoing effort to maintain the result.

Good enough for what?

No organisation can make every record perfect. Nor should it try.

A weekly staffing overview may reasonably use an estimate, clearly labelled as such. An important compliance decision may need evidence checked against the source record. The right standard depends on the consequence of getting the answer wrong.

The Office for National Statistics describes quality partly through relevance, accuracy, timeliness, accessibility and comparability. That reinforces a sensible approach: be explicit about the intended use and focus effort where reliability matters most.

That also means recognising when a manual check is justified. Not everything needs automation. The aim is to avoid repeatedly rebuilding trust in the same information.

When information becomes genuinely useful

Once the organisation can explain its figures, it can begin to use them with more confidence. Routine reporting becomes less dependent on individual interpretation. Teams can spend more time investigating what needs attention and less time debating which spreadsheet is right.

The same foundations help with future automation and AI. An assistant drawing on business information needs to know which sources are appropriate, what terms mean, what is current and what the user is permitted to see. Access to more files does not, by itself, provide that understanding.

There is a practical distinction here: the ambition is not to collect as much information as possible. It is to preserve the information and context that make useful action possible.

A useful question for your next management meeting

Choose a figure that everyone relies on. Ask: “If I asked someone outside the team to reproduce this answer tomorrow, would they get the same result — and could they explain it?”

If the answer is no, you have identified something worth examining. It may be a definition, an undocumented handoff, a reporting workaround or a genuine data-quality issue. That is a much more useful place to begin than commissioning a new dashboard.

How Processive approaches this

We start with the question you need answered and follow it through the real workflow: where the information begins, how it changes, where interpretation creeps in and who is accountable for it. We then help improve the process and the information supporting it, using existing technology where sensible.

That might mean clarifying definitions, simplifying data collection, restructuring records, bringing sources together or automating a stable part of the workflow. Where the organisation needs ongoing support, the design should also cover how quality and ownership will be maintained.

The first discussion does not need to involve an organisation-wide transformation plan. One important question that is unnecessarily difficult to answer is enough to start.