Project Everest

Work Update

[Work Update]: SoCon Malawi - LESSONS LEARNED: Gathering Data - July 2018

Henry Baldwin
Henry Baldwin | Jul 26, 2018 | in Knowledge Base

As the first SoCon (Social Consulting) team on the ground in Malawi, it’s our task to set the groundwork and build foundations for future teams to operate on in Malawi. We decided to make this post so that for future teams can reflect on the measures we’ve used and hopefully to learn on our experiences on how best to obtain their own data. As SoCon is so new to operating in Malawi, it is critical that the assumptions we make have been validated, as cutting corners with poor validation may lead to a serious negative impact down the track.

Our hope is by the end of our month, our handover document will not only include verified assumptions, but additionally a solid knowledge base for future teams to run an experimental framework off of. We’ve included little quirks that we’ve found that may make the process run more smoothly.


Interviews have been enormously helpful in our month on project through developing relationships with key stakeholders and acquiring information that otherwise would not be possible. This includes meetings with members of commercial banks, village banks, members of NAYO and others. Having a solid meeting agenda has really helped us define what we need from an interview and helped us keep on track. Utilising a semi structured interview (a preset list of information that you want and questions, with room for flexibility)  have proven to be the most effective method, as full structured interviews don’t allow enough fluidity for a conversational nature, and unstructured interviews can lack consistency and direction. We strongly recommend continuing to use interviews and meetings to gain information due to how useful they’ve been in the past.

Quirk: We found the most success when coming in after lunch for the banks and businesses we inquired to. I suspect is this is due to usually more work being done in the morning compared to the afternoon, and as such they may be more susceptible to answering our questions.

Surveys: Face to face

For our surveys we chose to adopt a more rigid format with room for flexibility in answers, and give the surveys a more conversational tone. A more structured set of questions has lead the data we’ve obtained to be more quantitative in nature, and as such can be better analysed for patterns and trends. In the July period we’ve surveyed the Limbe and Blantyre markets after clearly defining what we wanted from our surveys. These surveys have been very useful in testing our assumptions regarding our customer archetypes.

When reporting on findings, we found it best to employ a scientific report style for readability and consistency. Having a clear introduction, hypothesis, method, results and conclusion just leads to a better flow of information and allows others to easily see your thought process and relevant learnings.

Little quirk: we found surveying to be far more successful during the morning than the afternoon. People were less willing to engage with us, and those who did were less open that when surveying in the morning. We’re not sure why this was the case, but might be important to note for teams surveying in the future.

USSD survey

We still have yet to verify the validity of the USSD (text based) surveys. We haven’t yet obtained the results of the USSD survey that we sent out so I’ll try to talk as generally as possible. USSD is currently limited due to a small limit of questions (we estimate only around 10 questions, with the user running the survey over 3 minutes). The scope of who is targeted in a USSD survey is currently an issue: at the moment, sending the survey to phone numbers not restricted by any variables we desire (say for example geographic location) means the survey may be reaching people who aren’t relevant to our needs. I still believe a USSD survey has great potential though, and a detailed write up can be found here:


Using email to contact stakeholders and sending them a list of questions to complete has had limited success. When conducting our comparative analysis, most email surveys that we sent out were never responded to. If at all possible avoid this method over face to face options.

Quirk: email vs door knocking

Door knocking has turned out to be rather successful we found. When a few stakeholders hadn’t replied to our emails, simply door knocking various offices in Blantyre has landed us interviews and information that we would not of been able to obtain over the phone or email. Can’t recommend enough!

If anyone has found any other points of note regarding data collection please share! I hope this will be useful to teams in the future.

edited on 5th September 2018, 23:09 by Justin Hakeem

Wade Tink Jul 27, 2018

Interesting observations. Would suggest that this is informative to all teams operating in Malawi in the future as opposed to just SoCon. Would potentially best sit in Ideas Box...

Reply 0

Henry Baldwin Jul 27, 2018

Thanks for the comment Tink. Would you recommend deleting and reposting into ideas?

Reply 0

Wade Tink Jul 27, 2018

Status label added: Work Update

Reply 0

Jordan McLoughlin Jul 27, 2018

Agree with face to face being an effective method of communication. It was definitely the best method of pitching the energy teams solar product and conveying the benefits.

For USSD surveys, we ran an experiment in relation to demand planning for our product. We tested locals to see if they could follow a series of simple instructions in order to log an EOI via text. This included listing their name, a guarantors name, the village they come from and several other text based details.

The insight we got was that a lot of people around Blantyre markets did not like open text answers, as it was time consuming to type, especially for those who did not own smart-phones (poorer demographic which many of our projects aim to target)

I appreciate that you may want to use this USSD for more qualitative insights but I think the number of responses you will receive will be a lot higher with yes/no or 1-10 based answers. It was also our understanding that these answers would be far easier to create in a USSD format, in comparison to open ended answers.

Reply 1

Henry Baldwin Jul 27, 2018

Hi Jordan, thanks for the input. I should of specified that the USSD survey we sent out was in forced choice. Definitely for a large scale survey having qualitative answers isn't a great idea due to the difficulty with analysis.

Reply 0