Project Everest

[Experiment Design]: MVP Prototype SoCon I Fiji December 2018

Lean Phase: MVP Prototyping/Solution 

Assumption: This experiment is to test the assumption that having access to a customer database can assist merchants with accurately measuring the risk of loan applicants attempting to obtain capital.

Time Box: 2 months

Success Metric:

  • Number of merchants interested in our database

Green Light- If our assumptions are valid if merchants are interested in utilising the information provided over the course of January and February and sign a MOU and go to villages to get personal data to put in the database to provide the merchants with a beta version.

Success point If 75% of merchants sign an MOU  and are interested in our alternative data source being tested to predict risk.

Orange Light- 40-50% of merchants express that they would be interested in a database where they can access the risk of a customer

Failure Point If under 30% of merchants sign a MOU 

Red Light- We will look at other methods to assist farmers in accessing capital and different ways of providing financial assistance.

Experiment build:

  1. Develop a universal information system across a series of farmer related merchants including Individual/personalised profile

  2. Digitise the merchant/farmer trusting relationship to build an online network of customer profile (similar to uber profile app) includes;

    1. Risk Assessment

    2. Transaction history

    3. Sync Mpaisa through the app or use cash

    4. TImeline for payments

    5. Tracker and text reminders for upcoming payments

  3. Offer Testing - Go to merchants to see if they are interested in our product in both Sigatoka and bigger wholesalers in Suva.

  4. Visit farms and villages to entice farmers to input their information into the database

  5. Extend 30 lines of credit and loans over a determined period of time to assess the risk and if it has been mitigated by the algorithm

  6. Determine if it has been a success or failure by the default rate and late payments.

edited on 1st January 2019, 22:01 by Jess Riley

Andrew Vild Dec 17, 2018

Hi Maddie,

Just some pointers.

Usually, you would try and limit one assumption and one metric per experiment.

I believe you do have this, however, you've listed 4 metrics and from my understanding, 3 of them are actually results that we would include in the experiment results. The only one that might needs its own experiment in addition to this is measuring the default rates and working out if there is a correlation between repayments and the alternative data sources that you use to determine risk.

Green light is the action if success point (measuring a single MAYBE two metrics) is met. Orange light is the action if you're below the success metric but above the failure metric. Red light is the action if you are below your failure point and the failure point is defined as a clear cut metric measurement, same as success point.


Green Light: Assumptions are proven valid, we will go into rolling out a beta version of the digital risk management system and collecting critical data on villagers and providing to merchants to make informed decisions

Success point: If 75% of merchants respond well to a test version of the digital database and agree to use it as a guide to offering further lines of credit to individuals listed on the database. 75% of individuals are prepared to put personal details onto the system. More than 80% of loans are repaid in a regular and timely manner. (I still think there are too many metrics trying to be measured here.

Orange Light: Would need to ensure we are capturing the reason that less than 75% of merchants would trust the database, is there too little detail, not enough individuals in the system, is it not yet proven? Would then need to address these issues or offer a guarantee and see if that changes the +25% that have said they wouldn't make decisions based on this

Failure Point: If under 30% of consumers are willing to add their personal details to the system, merchants would not utilise the information. If 40% of the loans default

Red Light: We will look to other methods to assist farmers in accessing capital and different ways of providing financial assistance.

Reply 0