One year as a Charity Entrepreneur

When we founded NOVAH (No Violence At Home) in April 2024 through the Charity Entrepreneurship incubation program, we had a research-backed hypothesis that edutainment (mixing social messages in entertaining content) could effectively reduce domestic violence (technical term: intimate partner violence, IPV). Now, twelve months later, our first radio drama, Twubakane (meaning "Let's build together"), has reached ~30,000 listeners in Rwanda, where 24% of married women experience partner abuse annually.

This post reflects on our first year building an intervention from scratch, sharing lessons and surprises along the way that might be valuable for others considering similar paths. 

In previous posts, we discussed more about the background of NOVAH and its evidence base, and we presented why gender-based violence is becoming a more established and recognized cause area among the EA community.

The right partners accelerate progress 

As a young organization with limited resources, finding the right balance between going fast and being thorough was sometimes a challenge. We decided to commit six weeks to finding and interviewing potential partners. At the time, we weren't sure if this time commitment would pay off, but one year in, we can say it has. By building partnerships with both RWAMREC (an established gender-based violence prevention organization) and Urunana DC (an experienced radio production company), we got certain benefits we wouldn't have had otherwise:

• The opportunity to immediately start operations, anywhere in the country (gaining 2+ months);
• The ability to leverage their 18 years of expertise and strong international credibility to connect with global researchers and gain introductions at key conferences;
• The ability to learn from and access high-quality and efficient content production processes.

Iterative research to solve unexplored problems

Once we decided to launch in Rwanda, we were left questioning which specific messages would resonate in rural Rwanda. What should characters say and do? What advice should they receive?

Academic studies showed that 60% of edutainment programs aimed at reducing violence against women were effective. However, the research remains unclear about what kind of messages could be effective. We do believe that good edutainment could be designed by following certain steps, and we wrote a blog post about it.

We combined different approaches (academic literature, expert interviews, and formative research) and ultimately chose to run our own small-scale A/B test by producing three standalone recorded episodes and using focus groups to compare them in terms of engagement and educational impact.

For a reasonable cost, it helped us get a unique and context-specific answer to this question. It revealed key insights about what to avoid and emphasize in our final storyline and script. We ultimately focused on two key dimensions:

1. A gender equality message: Household financial decisions should be made jointly by spouses, and this makes the household better off
2. A conflict resolution message: it’s important and possible to seek help to prevent violence escalation, mediation being a strong option.

Without this small A/B test, we would have lost a lot of time in producing a final content that wouldn’t have resonated with the audience, and that could have even had some negative impact.

AI accelerates content production and improves monitoring 

When starting NOVAH, we hypothesized how AI could be a great asset to accelerate some of our processes. Now, twelve months later, we can confirm it has. To name a few specific instances in which AI has proven particularly useful for our intervention:

1. Automating broadcasting checks. For the intervention to work, we need our content to be aired at the agreed-upon times (e.g., 8 pm every day). AI and coding helped us automate verification that our content aired as scheduled, significantly reducing monitoring time (which especially proves valuable when airing promos multiple times a day across time zones).
2. Rapid translations from Kinyarwanda to English. It accelerated script iterations as we were able to translate content from Kinyarwanda to English, with high quality and almost for free, in less than an hour (where it previously might have taken 4 days and $200). Recent progress by LLMs was very valuable for us as Kinyarwanda is spoken by only 15m native speakers, and thus had not been prioritized by software companies (e.g., there’s no good speech-to-text yet).

Looking ahead, we envision AI to further optimize our process, enabling faster operations:

• Transcribing (if software improves) and translating interviews and focus groups that are key to understanding local context and how listeners perceive our content
• Adapting existing scripts to different cultural contexts
• Creating graphical content (e.g., videos) if we expand to visual formats like YouTube.

Small-scale M&E is very valuable to steer your program 

With limited resources, we designed a modest but informative qualitative evaluation study with 51 participants across rural communities. 

Despite its small scale, it provided some encouraging signals:

• 67% of male participants reported increased shared decision-making
  • “The drama helped me a lot. Before, my husband used to make decisions alone, but now we discuss everything.” - a female participant
• 60% of female participants reported increased willingness to seek help
• 90% reported relationship improvements (notably better communication)

  • “That has changed now—since my wife and I started communicating better, I also feel happier, and it is visible when I interact with others.” - a male participant

        

It confirmed we were on the right track (in terms of message selection, production quality), notably as gender equality (to which shared decision-making is a component) and communication are two known IPV factors.

To measure reach, we surveyed 500 people and estimated 27,000 active listeners (7% reach), with 90% of married listeners reporting positive relationship changes. This proved very useful as in Rwanda, radio stations do not have any listenership data (would it be 1, 5, 20%?).

(For more detailed results, see this infographic or our impact report.)

It gave us the confidence and key insights to continue our work, until the next phase, where we’ll improve the evaluation’s robustness.

A lean model at scale thanks to cyclical work and cheap media

One of the most encouraging findings that keeps us very optimistic about the future is the confirmation of the broadcast costs in Rwanda. For approximately $5000, we can air 14 episodes of our drama on Radio Rwanda, reaching up to 1 million people. Even when you include promotion and journalist fees, the total remains remarkably low. 

This cost-efficiency is what enables us to stay small and focused. NOVAH is a team of just three people (the two co-directors and one project manager in Rwanda), and we don’t plan to grow much beyond that - instead, mostly partnering with local organisations. Our work is cyclical by nature (research, writing, production), which allows us to operate lean while tapping into world-class external support when needed: from academics, advisors, and volunteers to expert implementation partners.

We’re inspired by the AMF model, which we learned about during the incubation program: a core team of just 13 staff coordinates with other partners and actors and protects hundreds of millions of people. This shows that small, evidence-driven teams can scale effective solutions to neglected problems.

What keeps us up at night

While early feedback is promising, our biggest uncertainty is the actual effect size of the radio program on IPV. While qualitative feedback is promising, we need more rigorous quantitative data to accurately estimate our impact.

In our preliminary cost-effectiveness analysis, we've applied a 50% discount factor to effect sizes from comparable studies and estimate that our radio drama could reduce IPV by 12.5%. Confirming this number through a more rigorous quantitative study will be our focus in year two. If this 12.5% decrease holds and we secure airtime on Radio Rwanda (the national broadcaster), the cost-effectiveness could be remarkable: $30 to protect a woman from IPV for a year.

What we would have done differently

When starting, we decided that it would be too expensive and tricky to compare IPV before and after airing our first show. You can imagine that asking women if they’ve been beaten recently is difficult and requires a referral system, which isn’t always feasible. This makes surveys longer, more sensitive, and more expensive than typical health surveys.

In hindsight, we could have taken a more creative approach to still have an inclination of the effect size in a natural setting. For example, by measuring change in subjective wellbeing , mental health, or asking general questions about relationship communication and decision-making patterns.

How you can support us to reach more people and generate evidence

Hearing Twubakane positively impacted relationships, and seeing the data on our reach, gives us the confidence to move forward. By expanding our programming and rigorously measuring impact on IPV, we aim not only to advance our own work but also provide insights on how to replicate this intervention (and how cost-effective it can be). 

This first year has been both challenging and deeply rewarding, and we’re grateful to all supporters. In particular, we are incredibly grateful for all the support we’ve received so far from the EA community since receiving our seed funding. It’s encouraging to receive support from both larger institutions and individuals.

If you believe in our approach, you can support us in different ways:

Donate: We are currently fundraising for our second-year budget, and every amount donated would help us to protect women at scale.

Connect: We'll be at EAG London and would love to share more details about our work and hear your thoughts.

Share: Know someone who might be interested in innovative approaches to gender equality? Please share our story. 

Follow our journey: Sign up for our newsletter to stay up to date on our progress.

If you have any questions after reading this post, please reach out to info@novah.ngo.