What a messy evidence workflow costs a reporting team

What does a messy evidence workflow cost?

It costs time and confidence. Time goes on searching for the right source or version, rechecking claims that should already be traceable, reconciling copies of the same table and answering review questions late. Confidence goes when nobody can say with certainty where a reported claim came from.

Messy workflows rarely look dramatic. They look busy. The same pattern repeats every cycle: repeated searching, late clean-up, duplicated figures, weak version control and slow review. The underlying problem is a broken chain between source and claim, which is why the fix starts with keeping every claim in a report traceable to its source.

The cost is easy to miss because it looks like ordinary project work. It tends to sit in four places:

None of this appears as a line in a budget. It appears as late nights, slipped deadlines and a report the team is less sure of than it should be.

What are the signs that an evidence workflow is broken?

The clearest sign is that information can be collected but cannot be reused cleanly. Files move, names change, logic sits in people's heads, and reports are assembled by stitching fragments together near the deadline.

Look for these signs:

• people spend time looking for the right source or version
• the same claims or figures are checked again in every review round
• the same figure appears in several places with different values
• review comments live in emails, document comments and chat threads
• drafts contain claims nobody can trace quickly
• reporting depends on one or two people remembering how things fit together
• every reporting cycle feels like starting from scratch

If several of these are familiar, the team does not need one more patch. It needs the structure underneath the reporting looked at as a whole.

Why does reporting start to hurt?

Reporting is usually where a system failure becomes visible, not where it starts. The trouble begins earlier, when inputs are captured loosely, stored in separate places or reviewed without shared rules.

Spreadsheets are doing a database's job

Spreadsheets are good for quick analysis and light tracking. Trouble starts when a workbook becomes the system of record for evidence, review and reporting, with copied tabs, hidden assumptions and no stable master copy.

The research on spreadsheet errors is a useful reality check. In an audit of 25 operational spreadsheets from five organisations, researchers confirmed 117 errors. Forty per cent of those had no effect on the results and 9 of the 25 spreadsheets had no errors at all, but the largest single error was worth US$100 million (Powell, Baker and Lawson, Decision Support Systems, 2009). The study is old and small. Its lesson still holds: most errors are harmless, a few are not, and without checks you cannot tell which is which.

If the work needs IDs, field rules, validation, status tracking and source links, it needs more structure than a loose workbook can carry. That does not always mean replacing spreadsheets. Often it means adding what is missing: defined fields, checks and links back to source.

Tool sprawl splits the evidence

A form tool, a notes app, a shared drive, a project board, a tracker spreadsheet, a reporting document and a few chat threads holding the rest together. Each tool makes sense on its own. Together they mean duplicate entry, broken review chains and evidence with no clean path back to its source. Teams export, paste, reconcile and repeat the same checks across systems that were never designed to work together.

The answer is not one giant file. It is one agreed home for the core records, their statuses and their source links, with other tools feeding it.

There are no shared standards

Shared standards sound minor until a team has to defend a finding, reproduce a table or explain which version of a file fed the final report. Without them, file names drift, folders become personal, fields get renamed and review comments apply to different drafts at once.

The fix is unglamorous: naming rules, version rules, field definitions, review checkpoints and a register of sources. Research data guidance recommends file names that show content, status and version, applied consistently across the team, with the naming convention written down (CESSDA Data Management Expert Guide). A register of what material exists, with a stable ID for each source, is set out in how to build a source register.

The process lives in people, not systems

If the workflow only works because one person knows the naming logic, coding rules or reporting sequence, the team has a dependency, not a system. New staff learn by guesswork and old email threads, and outputs vary because each person rebuilds the process their own way. A data dictionary, written file rules, review notes and handover material keep the work moving when roles change.

The load falls on people's attention

When much of the week goes on searching for files, switching between tools, re-entering information and chasing status updates, the team is tired before the analysis or writing starts. That strain is often treated as a staffing problem. It is usually a design problem.

Broader workplace research points the same way. In Microsoft's 2025 Work Trend Index, a survey of 31,000 knowledge workers in 31 markets, 80% of employees and leaders said they lacked enough time or energy to do their work, and Microsoft's own usage data showed employees interrupted every two minutes during core hours by meetings, email or chat (Microsoft, Work Trend Index, April 2025). These figures come from a software vendor and cover knowledge workers in general, not reporting teams. Microsoft's 2026 edition focused on AI agents and did not repeat the measure. They are context, not a measure of your team.

Firefighting is the result, not the cause

Late nights, last-minute checks and the rush to fix version problems before something goes out feel like the problem because everyone sees them. They are the result of messy intake, scattered storage, too many tools, undocumented process and output logic that arrives too late. Adding people may help for one cycle. The friction stays and returns at the next one.

How can you find out what it costs your own team?

Measure one reporting cycle before changing anything. Published figures describe other people's work; your own record of where time goes is the only cost estimate that will survive a budget conversation.

For one cycle, ask the team to log:

1. Time spent looking for a source, file or version.
2. Each time a source is reopened to confirm a claim already in the draft.
3. Each review comment that asks where a claim or figure came from.
4. Each figure that had to be reconciled between two versions.
5. Work redone because it was based on an outdated file.

Keep the log light: a shared sheet with date, person, task type and time is enough. At the end of the cycle, the totals show which part of the chain carries the most hidden work, and that is where to start.

What does a better evidence workflow look like?

A better workflow keeps the chain from source to claim intact from intake to sign-off, so review checks claims instead of rebuilding them. It usually has five traits:

1. A clear intake layer. Inputs arrive with IDs, dates, statuses and consistent fields.
2. One working evidence layer. Core records, source links and coded evidence stay aligned in one agreed place.
3. Visible standards. Naming rules, version rules, source notes and checks that people can follow.
4. Reporting shaped early. The evidence base is built with the output in mind, whether that is a funder report, a board pack or a public report.
5. Handover built in. The team can run the system after the person who set it up has moved on.

How the drafting stage should run once the evidence is in order is covered in the report writing workflow. For programmes that report to funders or government every quarter, the monitoring and evaluation reporting workflow sets out the same chain for indicators and partner data. Consolidating partner reporting so that each reported indicator links back to its source is the work behind the Social Employment Fund evidence portal.

When is a simple setup enough, and when do you need more structure?

A simple setup is enough when the workflow is small, one person controls the source material and the output is low risk. More structure is needed once evidence is reused across reports, reviewers or teams.

A simple tracker is usually enough when:

• the workflow is small and one person controls the source material
• the output is low risk and not reviewed repeatedly
• a single shared tracker fixes the current search and version problems

A more structured system is worth considering when:

• the same evidence is reused across reports, decisions or teams
• review rounds stall because claims and sources are hard to check
• your own log shows that searching, rechecking and clean-up take a material share of each cycle

If you are not sure which side you are on, start with a data system audit that shows what is wrong and what fixing it takes. The questions are the same either way: which outputs matter most, what scrutiny they face, where the data lives now, where the chain breaks and how much traceability the team needs.

How do you score your own workflow?

Score each item from 0 to 2: 0 if the issue is rare, 1 if it happens sometimes, 2 if it happens often. Add up the scores.

Messy evidence workflow scorecard (0 = rare, 1 = sometimes, 2 = often)

[ ] People spend time looking for the right source or version
[ ] Figures or claims are checked again in every review round
[ ] Review comments live in different places
[ ] Drafts contain claims that are hard to trace
[ ] Final reports need evidence clean-up just before submission
[ ] Different people work from different versions of the same material

Total:   0–3 low risk   |   4–7 moderate risk   |   8–12 high risk

A score in the moderate or high band does not mean the team is doing poor work. It means the structure underneath the work is making good work harder than it should be.

Common mistakes

• Treating the cost as admin time. Searching, checking and rework are part of delivery. If they happen every cycle, the workflow is costing more than it appears.
• Fixing the final report instead of the evidence flow. A rewrite rescues one output and leaves the source and review problems in place.
• Waiting for the deadline to clean the evidence. Late clean-up adds pressure and weakens review.
• Buying software before fixing the structure. A new tool on top of unclear fields and missing IDs moves the mess somewhere else.
• Reading the cost from someone else's statistics. Published figures are context. Your own cycle log is the estimate that counts.