A data quality assessment is only as valuable as what happens next

Something I've noticed from the monitoring side of M&E over the years: lots of organisations are open to hearing recommendations on the quality of their systems and data. They're less good at doing anything with them.

A few years ago, I conducted a data quality assessment for a major international health programme, reviewing six monitoring tools covering dozens of countries and cities. The findings were familiar: missing data, inconsistent formatting, gaps in reporting timelines, and reporting templates that varied across sites in ways that made consolidation a nightmare.

Based on a systematic review, we produced a set of recommendations. These included standardising indicator definitions, locking Excel cells, agreeing how to record zeros versus missing values, and setting up Stata do-files for consistent cleaning.

What made this case different was a fully engaged client, who saw the assessment as a tool for improvement. The working relationship was collaborative from the start, which made it easier to have honest conversations about what wasn't working and why.

When I returned to conduct a follow-up assessment two years later, the picture had changed significantly. Data availability and completeness had been meaningfully improved, and the consistency of data across sites was much stronger. The organisation had formed working groups, worked directly with regional hubs to support roll-out, and built our recommendations into their standard operating procedures.

By the time of the follow-up assessment, those active changes had moved the conversation forward. Rather than focusing on fixing basic errors and inconsistencies, we could now focus on genuinely improving the monitoring system and the skills of the people running it. The next set of recommendations included: automating submission pipelines, shifting error correction to hub level to build local ownership, and formalising guidance for new staff.

A data quality assessment is only as valuable as what happens next. But actively taking that next step is rarer than it should be.