Skip to main content
All example use casesIllustrative example

Finding duplicates across three cash programmes

An INGO wants to find duplicate and likely wrong records across three programme lists, without building a surveillance tool.

The challenge

An INGO is merging the lists of people it supports from three cash programmes. Some households appear twice, and some records have errors. The team wants to find duplicates and likely mistakes, while collecting no more data than it needs and without building a surveillance tool.

Kind of support

  • Record-linkage approach
  • Privacy-preserving matching
  • Review of the proposed workflow

How the mentorship could run

  1. Weeks 1 and 2: map the fields each programme collects and agree which ones may be used for matching.
  2. Weeks 3 and 4: test simple matching rules on a sample, such as names with document numbers, and measure how many matches are correct.
  3. Week 5: design a review step where staff confirm each likely duplicate before any record changes.
  4. Week 6: write down the approach, the data retention rules and the steps for a person to correct their record.

What to have ready before you apply

  • A description of the fields in each programme list, without real personal data
  • How many records each list holds and how often it changes
  • Your data-sharing agreements with partners
  • Who decides what happens when a duplicate is found

Risks to plan for

  • A wrong match can remove support from a household that needs it. Every match should be confirmed by a person, and people should have a way to appeal.
  • Combining lists creates a larger store of personal data. Hash or remove fields you do not need, and delete the combined data when the work is done.
  • Names are spelled in different ways across programmes. Exact matching alone will miss duplicates, and loose matching will create false ones.

What a good result looks like

The team has a matching method it understands, a review step it trusts, and fewer duplicate payments, without collecting new data about the people it serves.

More example use cases

  • Country: Bangladesh (Cox's Bazar)

    Sector: WASH

    Water-point dashboards with breakdown alerts

    A WASH NGO wants automated dashboards and alerts when a tap stand or latrine block stops working across 30 camps.

    Time: 1 to 2 hours per week, 8 weeksIllustrative example
  • Country: Jordan

    Sector: Education

    Arabic reading practice with teacher review

    A learning centre wants an LLM to draft Arabic reading exercises, with a teacher checking everything before a child sees it.

    Time: 2 to 3 hours per week, 10 weeksIllustrative example
  • Country: Ukraine

    Sector: Shelter

    Building damage from drone and satellite imagery

    A shelter NGO wants to classify building damage from imagery so field teams visit the worst-hit areas first.

    Time: 1 to 3 hours per week, 12 weeksIllustrative example
  • Country: Nepal

    Sector: Information management

    A safe assistant over SOPs and guidance notes

    An information management officer wants to build a retrieval assistant over their organisation's SOPs, safely and on a small budget.

    Time: 1 hour per week, 8 weeksIllustrative example

Know more

Want to know more before you apply? Write to contact@theaiexpertgroup.com and a member of the Mentor Council will reply.