Failing Fast: 3 Things We Have Learned about Engaging and Empowering Communities (BLOG)


March 16, 2016

In the second post of T4D's new blog series, Courtney Tolmie shares three lessons we have learned about engaging communities from 'failing fast' in not one, but six pre-intervention pilots.

By Courtney Tolmie

If you’re looking for evidence that answers the question of whether it’s possible to empower people and save babies at the same time, you are not alone.  We launched the Transparency for Development program in 2013 to answer this exact question.

To gather this evidence, we’re working with local partners in Tanzania and Indonesia to design a citizen engagement model that we hope will lead to improved health outcomes for moms and babies. Now we’re testing the model in 200 communities and conducting a rigorous, long-term evaluation to see whether it actually works.

But before launching this work in 200 communities and long before we have results, we planned to have a pilot – one pilot that might test a few different approaches to engaging and empowering communities to address maternal and newborn health service breakdowns.

And one pilot may have been enough, if a few things had been true:

  1. We had a clear best choice approach to transparency and accountability that we just needed to tweak around the edges;
  2. We wanted to test that one approach to see how it worked in a single context; and,
  3. The general approach had a short causal chain with just one or two short-term outcomes to trigger.

For better or worse, none of these things are true.

Just thinking through the chain of outputs and outcomes that needs to be triggered to get to healthier babies and moms makes this very clear:

T4D 2


This chain is long and messy, and a lot can go wrong along the way.  But this also has an upside. Steps 1-3 happen very early in the approach, and if they fail, they fail fast.

Which gets back to the issue of piloting.  While one pilot model implemented in a very condensed timeline was not appropriate for this project, we instead had an opportunity to test a lot of pilot models (six, to be exact) in an iterative manner to see what was failing fast – and what actually seemed promising.

Community representatives developing social action plans in one pilot village in Tanzania.

Community representatives developing social action plans in one pilot village in Tanzania. © Transparency for Development/Jessica Creighton.


Over the course of conducting these 6 pilots and as a result of failing fast, we discovered three key ingredients for success:

1) Working with a small, balanced group of citizens works best.

As a result of testing many different approaches, we found that we needed to bring together a small group of citizens (around 15 people) to discuss maternal and newborn health problems in their villages and work together on developing solutions.  We tried groups that included health volunteers and informal health workers, but these individuals tended to dominate the discussion and disrupt the balance of the group. We also tried having different balances of informal leaders and normal women in the village, and what we found was that a balanced mix of average citizens and informal community leaders was the sweet spot for getting people engaged and involved.  Finally, we found that the most active and engaged community activists or representatives were those who expressed a personal interest in and/or experience with maternal and newborn health.


2) You’ll need to equip that group with the right mix of information.

Information is not sufficient to trigger action and accountability; but it is necessary. Because information comes in many different shapes and sizes (it can be focused on inputs or outcomes, it can be localized or national and it can even be perception-based), a key question is what kind of information will help trigger action. We tried providing different kinds of information and ultimately found that a hybrid approach was best. We focused on the underlying problems and potential causes of maternal and newborn health problems. This included inputs (health worker attitudes and privacy in health facilities), but it also included outcomes (proper antenatal care) and perceptions (women prefer to give birth with a traditional attendant). We also tried focusing the community discussions on national and local statistics, but we found that community representatives wanted to discuss their own experiences — regardless of whether these experiences matched the statistics or contradicted them. As a result, the final model includes statistics collected from local facilities and women, but it also provides space and structure for communities to discuss what their experiences reveal to be the most critical problems in maternal and newborn health.


3) Citizens design interesting actions and adapt based on how targets respond.

When we got the first two pieces right, we saw that community members designed detailed action plans.  Initially, many of the actions focused on education campaigns targeting women of childbearing age, but during follow up observations we saw that some communities expanded their actions to include collaborations with midwifes and even “long route accountability” actions of asking government officials to fix problems happening at health facilities.  Communities were not required to take action, nor were they paid to carry out these actions. They did so based on the results of their actions.

Is this a sure sign that this approach will work in 200 communities across Indonesia and Tanzania?  No.  But we do know that we found a promising approach that was worth scaling up and testing at a larger scale.  And for us, it highlighted that maybe that persistent idea that piloting should be short and sweet is something we need to rethink.

Display Title: