The problem

In humanitarian sector, a typical Information Management workflow we met looks like:

• 📥 Data collection → KoboToolbox
• 🧹 Cleaning & calculations → Excel
• 📈 Visualisation → PowerBi
• ⚙️ Sometimes a bit of automation → PowerAutomate

Collection works but everything after export is chaos, for instance:

• When an activity or a project is spread in different surveys, how do you compute consolidated indicators?
• When someone edits data in Kobo, how do you keep Excel in sync, and vice versa?
• When multiple teams use the same form, how to isolate data?
• With high staff turnover, how to keep permissions aligned over time?
• When suspicious changes appear in a spreadsheet, how you know who changed what and why?
• When a form evolves, how to prevent dashboards and reports from breaking?

The consequences

This kind of workflow has 3 major downsides:

1. Time-consuming and repetitive
   Each step required a significant amount of manual work
2. Fragile and hard to maintain
   Excel formulas, PowerAutomate processes, PowerQuery transformation, all of these easily breaks especially when a new survey version is deployed. When surveys are too complex, it becomes almost impossible to harmonize the data.
3. Data duplication and divergence
   Changes will be made in Excel copies, or in KoboToolbox to re-trigger calculate fields, but it never propagates in the workflow.
4. Weak data protection
   Access cannot be strictly controlled: staffs can see data they should not, and permissions quickly become outdated with staff turnover.

Trying to automate with PowerAutomate or similar tools come with trade off: improve #1 but worsen #2.

The most critical issue are #2 and #3: data duplication. It is what makes reporting to donors or clusters inaccurate and a constant headache. Every time, we ask ourselves:

"Why is this figure different from the previous period?"

"How do you get that numbers? I can't reproduce. Let's start over."

"Let's round it to have something that looks coherent."

It leads to frustration and a loss of trust, making us fight against the data instead of relying on it.

The diagnosis and our solution

Let's focus on the core issues #2 and #3:

• Data inconsistency
• Difficulty harmonizing complex surveys

These problems are not new, software engineering solved them years ago.

The key idea: instead of having data scattered and transformed in many places (Excel files, dashboards, manual edits), we:

1. Store data once, in a single, secure source
2. Apply transformations using "pure functions"

What does that mean in simple terms?

A pure function is just a rule that:

• Always gives the same result for the same input
• Doesn't depend on hidden context or manual edits
• Can be run again and again with identical results

In practice: Every time a survey submission is created or updated, it goes through these rules and produces a clean, standardized output.

Why this changes everything

Your data never diverges

There is only one source of truth and one way to transform it. No more mismatches between files, dashboards, or reports.

You are not limited by tools like Excel

• No fragile formulas
• No broken cell references
• No "this logic is too complex to implement"

The logic is written in code (here TypeScript), so:

• It can handle any level of complexity
• It auto-validated by the compiler telling you if what you are writing is relevant.

Changes are safe and controlled. If a survey changes:

• The system automatically detects what is no longer valid
• You must update the logic before anything breaks

👉 This is something Excel or PowerQuery cannot guarantee.

It's easier to understand and maintain

Instead of:

• Hundreds of interconnected cells
• Hidden dependencies

You get:

• Clear, explicit transformation rules
• Logic that can be read and audited

Real-world impact

This is the philosophy behind NexusPortal. It allows us to:

• Handle 200+ inconsistent Kobo forms
• Harmonize them automatically
• Generate reports in one click
• Ensure data is accurate, consistent, and auditable

Summary

When surveys become complex, manual tools break down.
A centralized system with deterministic transformation rules is the only way to scale reliably.