Portfolio-Wide Impact Listening Using AI Voice Interviews in Latin America
In partnership with IDB Lab, Fortell Impact deployed AI-powered voice interviews across three pilot projects in Venezuela, Chile, and Ecuador to support portfolio-level impact monitoring and learning.
Watch the full interview with Yuri Soares
Overview
In partnership with IDB Lab, Fortell Impact deployed AI-powered voice interviews across three pilot projects in Venezuela, Chile, and Ecuador to support portfolio-level impact monitoring and learning.
The pilots tested whether conversational AI could provide a scalable, cost-effective way to capture lived-experience data across diverse sectors and geographies, while generating structured insight that could be reused across programmes and investment contexts.
Rather than relying solely on discrete quantitative indicators, the approach focused on capturing the richness of respondents' experiences with IDB Lab–supported projects — with a view to strengthening portfolio-level intelligence, beneficiary segmentation, and decision-relevant impact measurement.
Why this mattered for a development finance context
IDB Lab manages a diverse portfolio of early-stage and growth-stage ventures across multiple sectors and countries, each operating at different levels of data maturity. While some portfolio companies have the internal capacity to collect and analyse impact data, many do not. This creates a structural challenge for development finance institutions: how to generate consistent, comparable, and decision-useful impact insight across a heterogeneous portfolio without imposing heavy reporting burdens or incurring prohibitive evaluation costs. Historically, IDB Lab relied on randomised controlled trials (RCTs) to assess impact. While rigorous, these studies often cost between USD 200,000 and 300,000 per project and were not practical to deploy systematically across a large portfolio. IDB Lab therefore sought a more scalable monitoring approach that could:
- lower the cost of impact measurement
- capture qualitative dimensions of impact
- work across multiple sectors and countries
- generate insight that could be compared across portfolio companies
- support ongoing learning rather than one-off evaluations
Methodology
Respondents engaged in structured, conversational interviews that explored:
- how they experienced the IDB Lab–supported intervention
- what challenges they faced in their daily lives
- what benefits they perceived as most important
- what barriers limited their ability to benefit from the project
"With Fortell Impact, we were able to significantly lower costs by deploying AI-powered voice interview instruments. Unlike traditional instruments that collect discrete data on income or demographics, these surveys focused on capturing the richness of respondents' experiences with the project. By using artificial intelligence, we were able to deconstruct commonalities among respondents, identifying their challenges and the benefits they perceived as most important."
What IDB Lab learned
The pilot surfaced several vulnerability dynamics that are often underrepresented in income-only assessments.
The pilots demonstrated the value of capturing lived-experience data alongside traditional indicators, particularly in contexts where quantitative metrics alone provide an incomplete picture of impact.
Richer insight than traditional instruments
Unlike traditional surveys that focus on narrow indicators such as income or service access, the AI-powered voice interviews captured nuanced information about respondents' experiences, perceived benefits, and unmet needs across different project contexts.
Portfolio-level pattern detection
By using a consistent conversational framework across three countries and sectors, IDB Lab was able to identify recurring themes and patterns across otherwise unrelated projects. This created a foundation for cross-portfolio learning, enabling more coherent synthesis of insights than project-specific, one-off evaluation tools.
Increased openness in respondent disclosure
IDB Lab observed that respondents were more open in sharing their experiences when interacting with digital avatars, with longer interviews and improved data quality, particularly when using female avatars.
Why Fortell Impact's approach was different
Unlike traditional evaluation tools, Fortell Impact's platform combines digital avatars for conversational data collection, voice-based access for low-connectivity environments, automated transcription and translation, structured qualitative coding, and synthesis of insights across projects.
This makes it possible to transform large volumes of voice data into structured, decision-useful insight without relying on fully manual qualitative analysis workflows.
The approach is designed to be reusable across projects and geographies, enabling development finance institutions to move from one-off impact studies toward continuous, portfolio-level impact listening.
Implications for portfolio monitoring and learning
IDB Lab's experience highlights how development finance institutions can move beyond episodic impact evaluations toward more continuous, scalable monitoring models. By embedding voice-based interviews into regular monitoring workflows, institutions can:
The ERI framework demonstrates how vulnerability screening can move beyond static poverty thresholds toward a more dynamic and multidimensional model that is better suited for:
- track changes in beneficiary experience over time
- identify emerging issues earlier
- compare impact patterns across projects
- generate more consistent evidence for reporting and learning
- reduce reliance on expensive, one-off evaluation studies
Looking ahead
Building on these pilots, IDB Lab and Fortell Impact plan to explore the use of AI-powered voice interviews as a repeatable portfolio-level monitoring tool across additional countries, sectors, and investment contexts.
The longer-term vision is to integrate conversational AI into regular impact monitoring workflows, creating an always-on channel for beneficiary feedback and lived-experience insight that can support ongoing learning and decision-making.
The pilots
The pilots were implemented across three distinct contexts: **Venezuela** — Focus: Education system in a crisis context (243 completions) **Chile** — Focus: Financial products for migrants, in partnership with Galgo (602 completions) **Ecuador** — Focus: Economic opportunities for migrants and host communities (111 completions) Across all three contexts, Fortell Impact deployed AI-powered digital avatars to conduct voice-based interviews directly with respondents in their own language, without requiring literacy or stable internet connectivity.
Results by country
"We found that the respondents were more open to sharing experiences, the length of the interview was longer, and the quality of the data was a little bit better with the female avatars. We extended our understanding from real-life experiences with human beings to the digital experience with avatars, and I think that was tremendously interesting."
Decision Impact
Results are shaping IDB Lab's approach to portfolio-level impact monitoring and learning.
Have a similar challenge?
We'd be happy to discuss how we might approach your specific context and requirements.