When a form builder stops being enough

Most monitoring teams do not choose their first tool badly. They choose KoboToolbox or ODK because it works, it is free, and it puts a form on an Android phone in an afternoon. That decision is usually right. This article is about the point where it stops being right, and how to recognise it.

The honest signals

None of these signals arrive with a label. They show up as ordinary work that quietly takes more of the week.

• It is the third Tuesday of the month and someone is exporting CSVs again, because the dashboard the donor wants cannot be built inside the collection tool.
• A spreadsheet exists whose only job is to reconcile two systems. It has a version number in its file name, and a person maintains it by hand.
• A monthly report takes four days to assemble, and three of those days are copying numbers, not writing.
• Two people have asked for the same indicator this quarter and received different numbers, and both numbers were correct when they were pulled.
• A monitor corrected a submission last week, and the report that went out on Friday still carries the old figure.

Any one of these is survivable. Three of them together, every month, is a system telling you something. Collection was the bottleneck once. Now reporting is.

Being fair to the tools

KoboToolbox, for what it is built for, is excellent. It is free, open source and well documented, and the form designer is genuinely good. If your pipeline ends in a spreadsheet and the people who read that spreadsheet are happy, keep it. ODK is the most capable collector in the field. Moving off either one to feel modern is a bad reason to move.

The argument is not about collection. It is about what happens after. A collection tool answers the question of what we captured this week. A monitoring system answers what we achieved this quarter, against which target, split how, with what evidence behind each number. Those are different questions, and stretching a collection tool to answer the second one is where the spreadsheets come from.

Where the time goes

Watch a reporting cycle closely and the pattern is consistent. The data sits in the collection tool. The indicators sit in a spreadsheet, because the collection tool does not know what an indicator is. The narrative sits in a document template. Every month a person carries numbers from the first to the second, from the second into the third, and then rebuilds the charts the template expects.

Each hop is a chance to paste the wrong figure, and each hop has to be redone when a late submission arrives. This is why a quick monthly report takes days. It is not the writing. It is the assembly.

What changes

The fix is structural, not motivational. When the indicator, the dashboard and the report read one dataset, the assembly job disappears. A monitor corrects a submission, and the indicator, the chart and the report draft all move, because they are views of the same record. Nobody reconciles anything, because there is nothing to reconcile against.

• Indicators defined once, with their formula and disaggregation, not rebuilt per report.
• A dashboard that reads live submissions, so the export step disappears.
• Reports that start as drafts with the numbers already placed, so the month starts with writing, not copying.

Collection does not get worse. It gets easier to defend, because every figure on the dashboard links back to the submission behind it.

The reconciliation spreadsheet deserves its own warning. It works, which is the problem, because anything that works never gets replaced. Eighteen months in, the only person who understands it is the person who built it, and when that person takes leave or takes another job, the reporting cycle stops. A pipeline that lives in one person's file is an operational risk, and it usually announces itself as a missed submission deadline.

If this reads like your last reporting week, the honest next step is small. Bring one programme across, run it alongside the collection tool for a month, and count the days you get back. The tools you have are good tools. The question is whether the work around them has outgrown them.