How a Simple Process Hack Can Solve one of Development’s Biggest Problems
You’ve heard this a gazillion time: development work is very complicated.
But when you break it down to the most basic, day-to-day activities – the stuff that happens out there, in communities – things are actually relatively simple.
Let’s take an example from the world of health:
- Community-based health workers are going to people’s houses in their communities, regularly.
- These health workers provide a basic consultation and, perhaps, refer someone for a more specialized service at the local clinic;
- These local clinics need simple supplies;
If you get above right, a lot of health impact happens. The problem is that these super-simple things need to happen consistently, at scale. And that’s where things break. At scale, friction and inefficiencies compound leading to mind boggling complexity or, sometimes, halting the whole thing to a grind:
- There are tens and tens of thousands of health workers. It’s a numbers game: some of them will go to people’s houses, some of them won’t. It is impossible to know which is which;
- Of the ones that go to people’s house, some will refer patients for specialized services, some won’t. Impossible to know which is which;
- Local clinics run out of supplies, so people end up making a huge effort to go there, queue up for a whole day, only to learn that they cannot be attended to due to stock-outs. But it is hard to really know how often that really happens and what the realconsequences are;
- Millions of people go to consultations to local clinics. Some were referred in their communities, others simply decided to go by themselves. Impossible to know which is which; and then there are millions of people who never go for consultation, although they should – hard to know anything about them;
Faced with these challenges, over decades, the development industry has come up with a highly sophisticated solution:
Here is how it works in a nutshell: Health workers, community outreach agents, clinics and everyone else involved are asked to fill in paper forms for everyone they see. These forms then get sent regularly to some regional/ national hub, where people sitting in front of computers type the content into databases. They are humans – they make errors. These errors compound at scale.
One problem with paperwork of course, is you have no context: was the form filled in the field immediately after a session, or was it filled the night before the deadline, out of memory. Does it reflect real work, or is it made up? Impossible to know. Paperwork is basically input, not output (more on that here).
And even in the unlikely case that you actually solve above problems, paperwork still takes a lot of time, which creates a whole set of new problems. Most significantly, it slows down a process that was critically slow to begin with:
- The community health worker can only see 3 households/ day instead of 5 – that is a 40% productivity loss. At scale, those are millions of people who do not get a service they need.
- The nurse at the clinic can only see an average 16 of the patients sitting in front of the clinic since 5am, instead of 35, because she needs to do paperwork. More critical damage as her productivity takes a 55% hit.
- Meanwhile, the implementing agency has a 3-6 months lag between activity and data about that activity, which means they are not able to adapt based on this data.
Using an app instead of a paper form won’t change a thing. It will add more friction, though. And costs.
At Triggerise we take a completely different approach to these things. We are working around most of these problems by insisting on a simple validation process every time one of these simple, basic activities happen in the field. A validation is basically an automated confirmation of a personal interaction between two actors – health worker/ patient. Clinic/ patient. Shop/ buyer. Distributor/ clinic.
- every visit a health worker does to a household gets validated;
- Every client that walks into a clinic (whether they were referred or not) participates in a validation;
- Every woman that has been referred to a service, participates in a simple validation and the clinic will know if she was referred or not and by whom.
- The supplier validates delivery of commodities at the clinic;
We are running highly complex validation engines built around two simple steps:
- Actor A triggers a One Time Password (OTP) – basically a 4 character code that carries a number of predefined variables (expiration, how many times it can be used, whether or not the user needs to be qualified in any way etc). This code can be triggered by SMS, by a missed call or on an app. Actor A gives this OTP to actor B (shows it to them, writes it down, dictates it)
- Actor B actor sends this code to be validated – by SMS, by missed call or in an app.
That is it. Takes about 30 seconds. None of this costs anyone any money. No-one needs internet. All that everyone needs is the most basic phone on the market and a valid SIM card. This is something that most people have already, and the numbers of those who don’t have a phone in our target communities go down exponentially. That means that within 18-24 months phone penetration will reach near absolute numbers.
This validation process unlocks entirely new worlds of opportunities: we have real time data. We can automate personalized follow-up models, based on that data. We can build predictive models that ensures highly personalized interactions built on top of that validation.
The validation of a referral (i.e. a woman that was referred to a clinic by Community Health Worker X actually goes to the clinic) unlocks an instant payment to the health worker’s account, along with a congratulation message. The Health Worker can check at any point which of the people she referred have been to service and which haven’t – then she can easily follow up on those who haven’t.
Nurses go back to doing health work. M&E data comes in in real time, along with insightful, actionable contextual data.
No more paperwork. No more need for people in front of computers who type data into databases. Huge savings for everyone. Less friction. More visibility into high-scale operations. More actionable insights into a behavioral universe few understand.
The big question, of course, is why would anyone bother to do any validation at all? I mean, sure it only takes 30 seconds – a lot shorter than filling in a form – but it still is work, and we know that people don’t like extra work.
Well, they bother to do it because we reward them for it. Read about it here.
EDIT: there is a follow up post to this, explaining how you can collect richer data & even qualitative data on top of this validation.