Varsha Venugopal

There may be some huge exponential gains we could make by reducing costs by, say, 40%.

For instance, if we could find the right influencer without having to go through the whole nomination process, for instance.

So this third piece is higher impact, but lower likelihood.

So I think there is something unusual about the impact of our funding as a startup nonprofit compared to some of the more established GiveWell programs, because our model is still quite malleable and developing.

And some of the lessons we can learn from this funding can help us get more future funding, which could be used more effectively.

Yep.

So the gossip intervention is in the original J-PAL study, the surveyors were sent to a random set of 17 households and they asked several questions.

But the main question is, if there was a fair in town, who is most likely to tell you about it?

We then run an algorithm to identify the top gossips or community influencers and the surveyors then go back.

to these top influencers and recruit them as immunization ambassadors.

So that is the ambassador program in the J-PAL study.

What we did as a result of COVID was pivot to a remote model, which already brings down the cost significantly in that we call up a random set of households to identify these influencers and then call these influencers to recruit them as ambassadors.

So compared to the SMS model, this model at the moment is still not as mature and we're still iterating to find the most cost effective way of doing this.

That's the standard ratio we're using, yes.

We have some standard language we use.

And in fact, we are iterating with some of this, which also seems to increase the success rates.

But broadly, it is saying that you have been identified as a community influencer and we would like you to be an immunization ambassador.

There is no monetary cost involved in this.

It is a voluntary exercise.

We would be sending you regular reminders on immunization camps.