In our last post on PBS Idea Lab, NextDrop, which informs residents in India via cell phone about the availability of piped water, was trying to scale up in a very short period of time. How did we fare?

i-a8b03020188307e40774bc3b1bc1ffe6-nextdroplog.png

Well, I think we discovered the first step to winning: Just get good data about yourself. Period. Even if it’s ugly. Because after admitting there’s something wrong, the second hardest part is wading through the mess and figuring out what exactly that is!

Let me try to lay out everything we discovered about our service.

Customer Side

Goal: Bill everyone possible and make money.

Immediate problem: Billers wasted a lot of time because even when they found houses (which many times proved difficult), a lot of people were getting late messages, weren’t getting messages at all, getting them intermittently so they didn’t want to pay for the service (no argument there), or just didn’t want the service.

Immediate solution: Make a list of areas that have been getting regular messages for the past two weeks, and then call all those people before we actually go out and bill.

Immediate Systems We Put in place

Creation of the “Green List”: We look through all of our valvemen data, and using the all-mighty Excel, we figure out which areas received at least four calls within the last two weeks. Our logic here is that since the supply cycle is once every 3-4 days now, if they are getting regular messages, valvemen should call in at least four times in a 2-week span. This system is by no means perfect, but it’s a start, and at least gets us to the next level.

Conduct phone surveys: After we see all the areas that are on the Green List, we then call all the customers in that area. We spent two weeks piloting the survey to even figure out what categories/questions we should ask, and we’ve finally got some classifications the sales team feels good about.

Here are the different categories of NextDrop potential customers:

  • Could Not Contact (people who had phones turned off, didn’t answer the call, possibly fake numbers)
  • Satisfied Customers
  • Pay (want to pay for service)
  • Continue
  • 1-month Free Trial (again)
  • Deactivate
  • Unsatisfied Customers
  • Not Getting Messages
  • Wrong Messages

Bill: We just bill the people who are satisfied and want to pay, or who are satisfied but want another free month trial (and have already had one).

our customer cycle

Here’s a great flow chart that our sales manager made of our customer cycle (and if any engineers out there think this looks familiar, you’re right! It is, in fact, a State Diagram. This is why I love hiring engineers!) And let me say, this may look easy, but it took two weeks to analyze customer behavior to even figure out what states to include and how to go from one state to another state.

i-a1e57812d1d29bb8cf443862c3c69985-customercycle.png

When we finally had data, we discovered some really interesting things about our service:

  • Total number of people called: 1,493
  • Total number of people we could contact: 884 (59%)
  • Total number of deactivated customers: 229
    15% of total customers
    26% of contacted customers
  • Total number of continuing customers: 655
    44% of total customers
    74% of contacted customers
  • Total billable customers: 405
    27% of total customers
    46% of contacted customers
  • Total billed customers: 223
    15% of total customers
    25% of contacted customers
    55% of billable customers
  • Total number of people who paid: 95
    6% of total customers
    23% of billable customers
    43% of billed customers

As you can see, the two major problems we identified were 1) we were unable to contact 41% of the customers we tried to contact, and 2) a majority of the people who we were able to contact were getting incorrect messages (54% of the contacted customers).

troubleshooting problems

And that’s where we’re at: trying to troubleshoot those two problems. Here are the immediate solutions we’re putting in place to increase the people that we contact, and to put customers in the correct valve area.

Instead of taking “Could Not Contact” customers off the billing list, we are going to try to contact them. We’re in the process of seeing what percentage of the “Could Not Contact” customers we can actually find and contact when we bill.

We have an intern, Kristine, from UC Berkeley, who will be working with us for the next six months to figure out how to place people in the correct valve area (because that is the critical question now, isn’t it?) Kristine’s findings are pretty interesting (and definitely deserves its own blog post), but our first prototype is to test a guess and check methodology:

  • First we call customers and find out when was the last time they got water.
  • Then sort through our data and see what areas got water on that date (plus or minus a few hours). This should at least eliminate 50% of the areas.
  • Then, to narrow it down even further, we only consider those areas that are geographically close to the customer. This should narrow it down to within 4-5 areas to check.
  • We subscribe the customer to these areas, and see when he/she gets the correct message. (We will find out through the phone survey.)

That’s what we are going to try — we’ll let you know how that goes.

steps toward progress

In any case, I think the tunnel has a light at the end of it, so that’s all we can really ask for — progress!

And, as always, we will keep you updated on our progress, what seems to work, what doesn’t, and more importantly, why.

Additionally, and most importantly, we’re hiring! We are looking for enthusiastic and passionate individuals who want to be a part of our team. If you love problem solving, and finding creative solutions to problems, we want you!

As always, please feel free to write comments, offer insight, ask questions, or just say hi. Our proverbial door is always open!

A version of this post first appeared on the NextDrop blog.

Anu Sridharan graduated from the University of California, Berkeley in 2010 with a master’s degree in civil systems engineering; she received her bachelor’s degree from UC Berkeley as well. During her time there, Sridharan researched the optimization of pipe networked systems in emerging economies as well as new business models for the dissemination of water purification technologies for arsenic removal. Sridharan also served as the education and health director for a water and sanitation project in the slums of Mumbai, India, where she piloted a successful volunteer recruitment and community training model.