Project Everest

Post Not Ready

[Idea/Concept Not Complete]: FarmEd Fiji - USSD Data Collection for FarmEd Market Viability Data - July 2018

Isaac Crawford
Isaac Crawford | Jul 23, 2018 | in FarmEd - Fiji

The Problem


Too much of the past data collected by previous FarmEd groups has been unquantifiable and unanalyzable. This is due the fact that the metrics for simple questions such as ‘Do you have access to a mobile phone’ change monthly. With responses ranging from – ‘yes, my brother has a smartphone’, or ‘yes, I own a Nokia’. Whereas the answers can be kept discrete by having the following three options to the previous question:

-       Yes, my own phone.

-       Yes, somebody else’s phone.

-       No.


Due to the varied range of responses recorded, previous data collected has never had weight in terms of making assumptions on the market size of FarmEd. The key market size metrics being; smartphone access and usage, internet access and usage, and English ability (based off the assumption the MVP will operate in English).


The ultimate problem being, even if a team collected discrete and quantifiable data it would then only be of a small sample size between 70 to 100 surveys in rural areas. This creates difficulty to do with making assumptions on the market as there is not enough population data to make legitimate conclusions from.



Proposed Solution


By using a USSD data survey, not only will FarmEd Fiji be able to collect wide reaching data, it will also legitimize the data that is collected in the face-to-face surveying conducted by teams.

The survey would ask questions regarding mobile phone/internet access and English language ability. This USSD Survey was planned to allow collection of data across all of Fiji which could corroborate the same data as was collected in face-to-face surveys in the Sigatoka Valley region. This would hence allow PEV to demonstrate access to the Fijian farming market during VC pitches for the FarmEd app.


Please find attached below a sample of what the USSD script would look like if being put into practice.


Moving Forward

Ultimately, the team had to pivot away from surveying using a USSD this month due to costing limitations. However, a strong relationship with, and a detailed competitor analysis on both Vodafone Fiji and Digicel has been created, meaning execution on this data collection method is more than viable in the future.


The benefit of the USSD is that FarmEd would be able to target specific rural regions to gather data from. This would help significantly with the scaling of the application as the USSD could give insight into information on mobile phone/ internet access and usage, and English ability. Thus, enabling an informed decision to be made on whether roll out in this area is viable.


Below are attached some useful documents. The first link being a sample of the USSD script, The second a detailed competitor analysis of both Vodafone and Digicel with regards to what they were offering in terms of USSD development. Lastly, a simple two-page document outlining the USSD process undertaken by FarmEd Tech & MVP Testing this month.

edited on 6th September 2018, 01:09 by Justin Hakeem

Zoe Cahill Jul 25, 2018

Status labels added: Proposed Experiment, Experiment Results

Reply 2

Olivia King Jul 25, 2018

When deciding on the survey, eliminating any subjectivity on our part and theirs for English language ability was the most difficult part is deciding how to maintain maximum objectivity. A scale had to be employed, and we tried to leave it as closed as possible. We used factors like education and self judged scale. It was hard to fully establish quantifiably english ability across different surveys. But using a comprehensive rubric, the risk of deviation can be somewhat mitigated

Reply 2

Wade Tink Jul 27, 2018

Status label added: Idea/Concept Not Complete

Status labels removed: Proposed Experiment, Experiment Results

Reply 0

Wade Tink Jul 27, 2018

This looks like it could potentially be a proposed experiment and it would massively help the data requirements for FarmEd. Could you outlay the component parts of the experiment to be undertaken, especially what you are seeking to validate and the metrics around it?
You can do this by editing the post. Thanks.

Reply 1

Rose Martin Jul 31, 2018

Following your insights for USSD here in Fiji, what do you see as the optimal solution in the short-term for proving market validity? i.e. If you can't test your assumptions on a mass scale using USSD due to cost, what do you see as the next best alternative?
I think insight into this, or a post on this, would be useful for all projects encountering issues with USSD!

Reply 0