What unstructured document search costs West African institutions

The problem

A program officer at a development agency based in Dakar needs to prepare a note on water access in northern Senegal. She knows a similar report was produced two years ago by a colleague who has since left. She searches her files, emails former colleagues, explores the shared server whose folder structure has not been updated in three years. After 45 minutes, she gives up and starts the analysis from scratch.

This scenario is not an isolated case. It is the daily reality of most West African institutions that produce and accumulate reports, studies, databases, and internal notes — without a system to retrieve them.

An internal study across five partner institutions (development agencies, ministries, consulting firms) provides an order of magnitude: an agent spends an average of 2.3 hours per week searching for documents that already exist in the system.

What the research shows

A recent thesis on digital mobility in West Africa (Kodjo, 2024) analyzed digital usage in West African administrations. Beyond its main subject, it highlighted a cross-cutting finding: most organizations in the region use digital storage tools (shared servers, Google Drive, OneDrive) without an effective internal search engine. Documents are stored but not indexed in a way that allows them to be found through semantic search.

The result is a double loss:

• Immediate time loss: the 2.3 weekly hours spent on document search represent 6% of total working time. For a team of 50, this is the equivalent of a full-time position entirely dedicated to searching for documents.
• Intellectual capital loss: produced analyses cannot be capitalized. Each departing employee takes with them the knowledge of where key documents exist and reside. Reports exist, but no one knows where they are.

What this means in practice

The impact of an internal semantic search engine — based on vector document indexing (RAG) — can be measured at three levels:

1. Time regained. The 2.3 weekly hours per agent become 15 to 20 minutes. The saved time can be reallocated to analysis, strategic thinking, and producing new knowledge.
2. Information reliability. No more duplicated analyses or notes based on partial data because the full report was not found. Every document is indexed, every search returns relevant sources — even ones the agent did not know they were looking for.
3. Institutional memory. Departures and internal mobility no longer erase knowledge of what was produced. The system knows where every document is, regardless of who created it.

Why this problem persists

The reason is not technical. Semantic search engines exist, RAG models are mature, and infrastructure costs are accessible. The real barrier is elsewhere:

• Managers underestimate the cost of non-structuring. 2.3 hours per agent per week does not seem serious until translated into full-time equivalents and euros.
• Technical teams and managers do not speak the same language. A program officer does not ask for a "vector engine with semantic chunking" — they ask to find the report they wrote last year.
• Generalist solutions do not solve the problem. Google Drive, SharePoint, or Nextcloud have search functions, but they are based on file names and dates, not semantic content. Finding "the report on sanitation in Cotonou" among 50,000 files named report_v12_final_v3.docx is impossible without semantic indexing.

This is precisely the gap Lektaris fills: an internal knowledge engine that indexes document content, not just their names, and lets you find in seconds what already exists on your servers.

Further reading

• Digital mobility in West Africa
• Network reliability and quality of service in Togo
• Digital cognition in children
• Explore our RAG knowledge engine

Academic source cited: Sena Kodjo. Digital mobility and digital usage in West African administrations. Information Sciences. University of Lomé, 2024. hal.science/tel-05391367