Installation is as easy as downloading the data collection app (ODK Collect) from the Android market. Out of the box it transmits form data via an HTTP POST to ODK Aggregate, a server that runs locally or on Google’s App Engine and provides an interface for viewing and exporting your data. A selection of demonstration forms are included. One of these, Geo Tagger, allows you to capture a photo and GPS location, enter a brief description, and upload the result to Aggregate.

I thought it would be useful to be able to skip Aggregate and send a geo tagged photo directly to a website in Django, which turns out to pretty straightforward. It comes down to parsing the XML form data included in a multipart attachment to the HTTP post from Collect, which can be done with a few lines of code. If you’re interested in the details I’ve posted some of them here. The end result looks like this (click screenshots to enlarge):

which with a quick template becomes:

And there you have it: a basic mashup of two open source tools that is directly applicable to field-based data collection.

Things get interesting when you start to think about use cases. Over the past month I have had the opportunity to talk with several folks working on international development initiatives about their needs for data from the field. Examples include a private company selling solar energy products in Sub-Saharan Africa, groups distributing improved cookstoves in Haiti and Tanzania, and an NGO that offers a range of household energy and health services in Central America. What stands out to me is that these disparate organizations have strikingly similar data collection and reporting needs. They want to know who and where their users are, understand which sales staff and distribution channels are most effective, and optionally collect survey data and multimedia to use for internal business administration, reporting to funding organizations, and communicating with the public.

I think that the recent availability of robust, field-tested, open source data collection platforms (ODK, RapidSMS, JavaRosa, and FrontlineSMS are a few examples) presents an opportunity to start addressing this need for data in a very general way. Since the tools are open there are no barriers to linking them with each other and with the web (i.e. through Drupal, Django, etc.), and given the range of potential uses it makes sense to think carefully about how to define those interfaces. Ideally data could be passed between applications according to clear, open standards that are consistent with modern monitoring and evaluation practices. These may not need to be much more complicated than the proof of concept integration above, but they will have to be better planned.

My ideas on the subject and about what standards would be helpful are still evolving, so feel free to contact me if you’re interested in this topic.

Suraj talking stoves, unfinished stove bodies behind him to the left.

Suraj Wahab is passionate about cookstoves. Indeed, efficient charcoal burning stoves like those made by his company, Toyola Energy Limited, offer a lot to be passionate about. For hundreds of thousands of families in Ghana who cook using traditional methods, these simple metal and clay devices provide a cleaner, safer, more efficient way to prepare their daily meals, while also reducing greenhouse gas emissions and deforestation. The stoves are sold in markets and door-to-door by Toyola “evangelists”, individuals who record each sale in a notebook and then are paid on commission. With 50,000 stoves projected to be sold this year and double that possible in 2010, the paper records are becoming increasingly difficult to maintain. Over the past six months I have been working with Toyola and E+Co, a US-based non-profit investment company investing in Toyola, to develop an SMS-based recordkeeping system for their stove sales.

Assembled stove bodies waiting to be finished.

The stoves are assembled from scrap metal so they're painted black for uniformity.

Toyola evangelists are quick to point out that well designed stoves use 50% less charcoal than traditional models, meaning they pay for themselves in fuel savings after a few months. Increased efficiency also means that charcoal production and deforestation are avoided, reducing the greenhouse gas emissions associated with cooking by about 1 ton per stove per year. If carefully documented and then verified by a third party these emissions offsets can be sold to organizations in developed countries wishing to reduce their environmental impact. The process is called carbon finance, and is potentially a tremendous source of revenue for stove businesses.  By a conservative calculation the value of a stove’s emissions reductions over three years is more than double its purchase price. This is capital that can be used to grow the business, reduce retail prices, and extend credit to customers, increasing the reach of clean cooking technology and making it more accessible to low income households.

New advertisement: "Don't burn your money - use Toyola improved coalpot".

It’s not easy though. The most widely used standard for developing cookstove emissions offsets, a methodology published by the Voluntary Gold Standard, requires a rigorous registration and verification process. Vendors must maintain sales records which are later used by trained evaluators to follow up with stove owners about their fuel use and cooking habits (see http://www.cdmgoldstandard.org/ for details of what is required). Toyola staff and evangelists’ notebooks are currently the basis for these records, and compiling them is an ongoing challenge due to lack of infrastructure, literacy, and incentive

Paper records we're replacing with SMS.

Last week I traveled to Ghana with Erik Wurster, director of carbon finance for E+Co, to set up a pilot of SMS recordkeeping software intended to partially replace the notebooks. The application is called ASSIST and is based on RapidSMS, an open source messaging framework developed by UNICEF to support their operations in the developing world. With ASSIST evangelists can use fixed format SMS messages to update a sales database directly from the field. The application was designed primarily to track sales for carbon recordkeeping, but it also offers limited supply chain management and back office functionality. Managers access the database through a web interface that provides aggregate statistics and the option to export an auditor-ready excel spreadsheet of stoves sold.

The whole system: Asus EeeBox server, Multitech GPRS modem, and two 10 GHC top-up cards on a dusty bookshelf in E+Co's Accra office. A backup battery is on the floor below.

The message used to record a cash sale looks like this:

sale <user first name> <user last name> <stove serial #> <price paid> <mobile number> <region code (0-9)> <village> <free text description>

Where region codes are specified on laminated reference cards given to evangelists. For example, to sell a stove for 10GHC to a customer named Marindame Kombate who lives in Kumasi with mobile number 0241273802, the message sent would be:

sale marindame kombate hmn123456 10 0241273802 5 kumasi near ecobank

“near ecobank” is a brief free text description that can be used to help the seller remember the stove’s location.

Explaining the software to Toyola management.

Erik and I spent almost a full day at one of Toyola’s production facilities explaining the software to management and training employees. Their initial reaction to the system was positive. Suraj was most excited about the ability to view stock and sales figures in real time, but also appreciated that the software would make maintaining carbon monitoring data more efficient and accurate, and virtually eliminate the risk of double counting stoves. Some employees were concerned about the length of sale messages. They are admittedly cumbersome, and I’m working with Suraj and his staff to make them shorter. In the coming months I will be improving and refining the system based on their input.

Training Toyola staff on how to record sales using SMS.

I’m encouraged by Toyola’s reception of ASSIST, and curious to see whether it proves itself as a sustainable solution. I am also actively looking for future projects and interested in learning how to develop more sophisticated and flexible tools. I think that mobile tech presents exciting possibilities for development work, and look forward to finding other innovative applications.

Thanks:
Erik Wurster, E+Co – funding and design input
Bikash Pandey, Winrock International – providing my time
Berkeley Air Monitoring Group – early discussions and design input
UNICEF Innovation – RapidSMS

Questions or comments? thebenedict at gmail