Getting Teachers Where They’re Needed Most: Insights from Sierra Leone’s GIS-Supported Preference Matching Algorithm

This summary discusses the conversation from a virtual meeting that covers lessons and what worked with the Sierra Leone GIS preference matching algorithm and lessons from other regions including South America.

---

Sierra Leone faces long-standing teacher deployment challenges: too few qualified teachers in the rural and remote schools, high turnover in certain regions, subject specialist shortages, and striking underrepresentation of female teachers on payroll in remote communities. These inequalities undermine learning outcomes and widen the gap between urban and rural learners.

To address this, the Teaching Service Commission (TSC), in partnership with EdTech Hub, the Learning Generation Initiative (LGI), and Fab Inc., introduced a GIS-supported preference matching algorithm during the 2024/25 deployment cycle. Inspired by the Nobel-prize winning matching algorithm used to solve health worker allocation issues in the US & Ethiopia, the algorithm aims to place the right teachers in the right schools using transparent, data-driven criteria. Sierra Leone is among the few countries in the world to use such an innovative approach to deploy teachers to schools.

This webinar brought together policymakers, researchers, and technical experts to share what worked, what didn’t and what Sierra Leone’s experience means for other countries facing similar challenges. The webinar also brought in comparative insights from Latin America, where similar centralised teacher assignment reforms have been implemented in countries such as Ecuador, Peru and Chile.

Watch the webinar here

Featured speakers 🔉

• Taskeen Adam, Senior Research Lead, EdTech Hub
• Lans Keifala, Chairman, Teaching Service Commission
• Katie Godwin, Head of Research, Learning Generation Initiative
• Tomáš Koutecký, Education Economist, Fab Inc
• Madleen Madina Frazer, Country Lead, EdTech Hub Sierra Leone 
• Gregory Elacqua, Principal Economist, Education Division, Inter-American Development Bank

Key takeaways 📝

1. Sierra Leone’s deployment challenge is multifaceted

The Teaching Service Commission highlighted several structural challenges that the new matching algorithm is designed to tackle. Remote schools continue to face shortages of qualified teachers, while some regions experience high turnover that disrupts continuity in learning. At the secondary level, many schools lack subject specialists, and female teachers remain underrepresented in hard-to-reach areas. The Commission also noted persistent inequities in how teachers are distributed across the system. 

To ensure that only qualified teachers are deployed, the commission, also introduced a licensure examination. However, too few teachers passed the exams, with distance education trainees recording the highest failure rates, and some qualified teachers ultimately excluded from deployment after being linked to falsified certificates.

These issues underscore broader systemic constraints—problems rooted not in individual schools, but in the design and functioning of the wider education system.

2. The matching algorithm brings more equity and transparency

The GIS-supported algorithm introduced several critical improvements:

How it works:

• Schools are prioritised based on the pupil-payroll teacher ratio (PPTR) and need.
• Teachers submit school preferences for deployment via the Teacher Management Information System (TMIS)
• The matching rule considers level, qualification, and distance to schools in need
• Tiebreakers include religious preference matching and a gender-based tiebreaker in favour of female teachers

Immediate Impact:

• More qualified teachers deployed — 100% of selected teachers passed the licensing exam
• More teachers placed in high-need schools — 64% of teachers were deployed to schools with PPTR>100
• Increased allocation to remote and disadvantaged districts — 57% of teachers were deployed to disadvantaged districts
• For the first time, every teacher added to the payroll was able to indicate their preferred schools. Overall, 54% of teachers were matched to their preferred schools. Only 7% of selected teachers were allocated to a school more than 5 km from their preferred option.

The algorithm is an open-source online tool that is flexible and can be used by other countries. The link is available under Resources.

3. The insights from the webinar revealed both progress and gaps

What worked well:

• Clear reduction in political interference as politicians and senior officials can no longer influence placements
• More transparent and impartial decision-making, as the TSC can justify its decision through the online system
• Improved equity in placements as districts with the highest PPTR were given priority, irrespective of the size of the district.
• Evidence of cost savings and efficiency gains from the digitisation of the system. The Chairman of the TSC  no longer had to physically sign every teacher registration form, and TSC staff no longer had to travel across the country to verify information submitted by Deputy District Directors.

Challenges:

• Inconsistent communication with teachers about placements initially limited teacher’s buy-in
• Insufficient public sensitisation of the new process led to some confusion
• The condition to pass the licensing exam limited to number of teachers that could be considered for deployment, reducing the algorithm’s ability to match teachers to their preferred school 
• Data integration issues between TMIS and other education data systems (e.g. Annual School Census) 
• Limited capacity within the TSC to manage the new digital requirements

These insights highlight that technology alone cannot solve deployment challenges; system strengthening is important.

4. Lessons from Latin America

Our discussant, Gregory Elacqua, shared reflections on the Sierra Leonean case and emphasised lessons from Latin America:

• In both Sierra Leone and Latin America, centralised systems improved the matching of teachers with schools and saved costs, showing the promise of this approach.

• In Sierra Leone, only one preferred school is selected by a teacher. In countries like Ecuador and Peru, teachers can rank up to 10 or more schools during the application process. This helps the system understand their true preferences and leads to better, more accurate matches. When teachers are allowed to select only a few schools, they often try to “play the system” by choosing schools they think they are more likely to get rather than the ones they genuinely want.

• In Latin America, Deferred Acceptance (DA) algorithms generate fairer outcomes when teachers rank more schools: teachers are informed when it’s not possible to deploy them to their preferred schools and are asked to select other schools. In Sierra Leone, teachers are immediately to the closest school to their preferred schools.

• While there is high teacher application density in Latin America, in Sierra Leone, because not enough teachers are passing the licensing examination, the pool of teachers eligible for deployment is relatively small.

• Machine learning tools (such as risk alerts, recommendation engines) have the potential to improve allocation efficiency. In Ecuador, machine learning tools are used to send alerts to teachers when their preferred schools are oversubscribed, prompting them to select alternative options. This has increased the number of teachers deployed to schools. In Brazil, recommendation engines are used to promote hard-to-staff schools on commercial platforms.

5. Evidenced approaches to increase female teachers

Countries across Latin America have adopted several strategies to improve gender balance in the teaching workforce. Key approaches include:

• Targeted recruitment: In Chile more female candidates are recruited into the preservice training program. 

• Scholarships for women: Financial support is provided to female teachers 

• Monetary and career-path incentives can increase female teachers’ uptake in remote areas. However, monetary incentives are expensive, sometimes 16-30% of a teacher’s base salary. Peru provides monetary and non-monetary incentives to female teachers in remote areas. Guyana offers housing support

• Making information available on incentives and the impact teachers could have on schools, the public could also increase female teachers’ uptake. Though this has minimal cost, its effect is more moderate. 

• Adjusting licensing requirements: Colombia reviewed the weight of its licensing test and made adjustments to ensure fairer outcomes for female candidates.

Questions 🙋🏾‍♀️

Are TTIs delivering some of the modules teachers are taking exams on in the TSC?

Response: Yes. The intention is that any newly trained teacher graduating from a Teacher Training Institution (TTI) should be fully prepared to pass the national licensure examination. The exam syllabus is publicly available. Importantly, the exam questions were developed by TTI lecturers themselves and cover core areas including English, Mathematics, ICT, Education, and Professional Standards. 

Is the algorithm equally effective in primary and secondary schools?

Response: Overall, we did not observe major differences in how the algorithm performed for primary versus secondary school placements. However, secondary schools are fewer in number, which means there are fewer nearby alternatives if a teacher’s preferred school is already well-staffed. This limited set of options could influence the matching process and may have a slight effect on how efficiently the algorithm allocates secondary school teachers.

What is the actual number of teachers we are giving the % record for? 100% of how many?

Response: 2,341 teachers passed the licensing exam and were recruited in the 2024 deployment round by TSC.

What are the specific “rules/conditions” (besides grade level and specialism) that the algorithm uses to match teachers to schools? (e.g., certification status, geographic zone, minority language skills)?

Response: The algorithm applies a range of criteria to ensure that teachers are matched fairly and appropriately. While the full list is detailed in our quantitative report, the key factors include: district, school level, PPTR, licensed, school approval status, distance, remoteness, qualification, and gender.

Which specific programming language or platform is the open-source algorithm built on? ​Is the code publicly available for review and replication?

Response:  See the public GitHub here.

Are the TTIs offering the Professional standards modules, which are part of the exams?

Response: The Professional Standards referenced in the exam are the Teaching Service Commission’s Professional Standards, for teachers and school leaders, which have been in place for several years. The full standards document is available here.

Resources 📚 

Frazer, M. M., Adam, T., Godwin, K., Mackintosh, A., & Haßler, B. (2025). Transforming Teacher Deployment: Lessons from a matching algorithm tool [Policy Brief]. EdTech Hub. https://doi.org/10.53832/edtechhub.1114. 

Godwin, K., Cameron, E., Frazer, M., Dube, G., Mackintosh, A., Koutecký, T., Adam, T., & Haßler, B. (2025). Shifting power dynamics in education decision-making: Investigating the role of a matching algorithm to improve teacher deployment in Sierra Leone [Technical Report]. EdTech Hub. https://doi.org/10.53832/edtechhub.1103.  https://docs.edtechhub.org/lib/IRFH43RX.

Mackintosh, A., Koutecký, T., Godwin, K., Adam, T., & Frazer, M. (2025). Data-driven teacher deployment in Sierra Leone: Practicalities and quantitative analysis of using a matching algorithm in the 2024/25 deployment cycle [Technical Report]. EdTech Hub. https://doi.org/10.53832/edtechhub.1115. 

Frazer, M. (2024). From algorithm outputs to classroom impact: A conversation with Marian Abu, director of teacher management at the Teaching Service Commission. Available at: https://edtechhub.org/2024/12/04/strengthening-the-use-of-data-for-decisions-on-teacher-deployment-in-sierra-leone/.

The Impact of GIS-supported teacher allocation in Sierra Leone. https://edtechhub.org/evidence/edtech-hub-research-portfolio/impact-of-gis-supported-teacher-allocation-sierra-leone/.

Teacher Deployment Algorithm https://github.com/Fab-Inc/FabMatch_SLE.

For a full list of reports related to this project