A few months ago, I negotiated what looked on paper like a straightforward mobile data collection setup.
The conversation began like many others in the deal pipeline. An IT lead at a non-governmental organization wanted a mobile data collection system built on KoboToolbox, hosted on a private server on their premises, with live data streaming into Power BI dashboards.
He explained that more than eighty people would use the system: enumerators in the field and project managers overseeing operations.
Technically, it was a very good proposition. The data architecture they suggested made sense, and the tools were familiar. Nothing about the request itself was unusual.
The most striking thing, however, was how lightly system training was treated.
When the discussion got to training and capacity building, they assumed that two hours a day for three days would be sufficient for the team to design and manage their own digital forms with skip logic, validation rules, and all attendant functionalities.
The client expected that once the system was installed, people would simply pick it up like fish in water. Any suggestion that more time was needed was met with shocking resistance, framed as an unnecessary cost rather than a critical investment.
The negotiation now became tense. Costs were scrutinized, timelines compressed, and training steadily watered down.
He was argumentative, continuously insisting that budgets were tight, donors were watching, and more “strategic review” meetings were looming on the horizon.
There was a growing sense that we were talking past each other. We were discussing organizational capability while they were negotiating a software purchase.
NGOs Overengineer Technology, Overlook Human Capacity
This is one of the most common mistakes NGOs make with mobile data collection: treating it as a technical installation rather than a change in how an organization thinks, works, and makes decisions.
Tools like KoboToolbox are often described as “user-friendly,” and they are, to a certain extent. It can be easy to create a form. It is much harder to design one that consistently produces clean, reliable, and analyzable data across dozens of enumerators, multiple locations, and real-world conditions. Form logic, validation rules, skip patterns, question phrasing, and data structure all have consequences that only become visible once data starts flowing into live dashboards and reports.
When training is rushed, those consequences may not appear immediately. Forms will still submit, the live dashboard will reflect, and numbers will still move. Everything looks like it’s working.
The Hidden Cost of Poor System Training
Poorly trained teams do not usually produce obviously broken systems. They produce data that is just not good enough to trust once things get real.
Cheaply trained enumerators interpret questions differently and yield rubbish data with constraints missing or misapplied, fields inconsistently filled, or overlapping categories. Managers look at live dashboards with confidence, unaware that the underlying data has structural weaknesses baked in from the start.
Problems pop up downstream: during analysis, during reporting, or worse, during program evaluation. At that point, the cost of fixing data is far higher than what would have been spent on proper training.
What came out of this negotiation was the imbalance in priorities. Significant attention was given to private servers, real-time data streaming, and sophisticated visualizations. Very little was given to the people who would actually run the thing, generate and interpret the data. Eighty users, many of whom were said to be new, were expected to adopt new tools, new workflows, and new responsibilities, with little training.
Many engineers will tell you that complexity is not negotiable. It can only be postponed.
You can reduce training hours on paper, but the complexity shows up later in the field or during data cleaning.
This then translates into delayed decisions and frustrated teams down the road.
For NGOs, bad data has serious implications. Data collected through mobile systems increasingly informs interventions, funding decisions, and claims of impact. When that data is weak, interventions are misdirected, communities are misrepresented, and accountability suffers. In sectors like health, governance, justice, and humanitarian work, the consequences of bad data are not abstract. They affect real people.
NGOs should stop negotiating or spending endlessly on technology. It should be about people, discipline, and institutional learning.
A modest system in the hands of a well-trained team will always outperform an advanced system operated by people who were never given the time to understand it properly.
Failure starts at the moment where training is dismissed as an inconvenience rather than recognized as the foundation on which everything else rests.
