Dan Gardiner | 31 Jul 2025

Frontline workers know the reality on the ground: they see the problems, they understand the nuances, and they often have ideas to solve them. But there's a persistent, frustrating gap between these critical observations and the decision-makers that can change things. That vital nugget of information, born from direct experience, frequently gets waylaid, lost in translation, or simply never makes it anywhere.

With the rise of digital technology and AI, we're finally equipped to bridge this systemic challenge. At Colectiv, we've developed a unique approach that transforms raw, qualitative data into structured, useful intelligence. Let us illustrate this with a real-world example from a recent project with UNICEF Tanzania on Mpox. A single frontline idea powerfully demonstrates the impact of our analytical journey.

1. Capturing the original insight

"Kuwepo na sehemu ya Isolation kwaajili ya mgonjwa kupelekwa."
(There should be an isolation area for the patient to be taken to)

This insight is critical: if Mpox patients must isolate to prevent transmission, designated areas must be provided, since many people cannot isolate at home.

2. Mapping to our topics of interest

Our researcher (i.e. human) led analysis starts by defining topics, and then getting AI to search through all our qualitative data for quotes relevant to each topic. In this case, our researcher defined topic that picks up our quote is:

• Topic: "Recommendations for strengthening the local Mpox public health response"

3. Ensuring verbatim accuracy

We're using AI, so let's make sure our quote isn't hallucinated. At this point we use a non-AI algorithmic process to check every quote used is 100% verbatim in our transcript. Yes, this quote is real - as are all the others in our analysis.

4. Clustering into themes

At this stage, we apply a hybrid machine-learning / AI approach to cluster semantically similar quotes into themes. Our quote is grouped into the following theme with feedback from 15 participants:

• Theme: "Operational preparedness and resource management"

5. Deriving actionable insights

For each theme (sitting within a topic) we generate insights that summarise the quotes we've picked out. In this way, our quote is one of 7 within the following insight:

• Insight: "Prepare and equip designated isolation facilities or treatment centers with necessary medicines and equipment for Mpox cases."

And this is what we end up with. Nice organised qualitative data that programme managers and innovators can use to plan their work, iterate and test changes, and achieve more impact.

We make it easy to drill down from topics, to themes, to the individual quotes they came from using our interactive dashboard.

If you want to see what this looks like across a whole study, or even better - with your own data - in our dashboard, just send us a message at hello@colectiv.tech.