Intended audience: DataKind Volunteers
Discovering a project starts with problem identification. As you seek to understand the partner and its pain points, consider using the organization’s pre-mapped logic models based on their theory of change. Work together to draft problem statement(s). In doing this, you can consider asking:
Organization overview and success metrics
- How is success measured in the organization? A good understanding of the organization’s ultimate success will help you understand what to learn/optimize to get there.
- What trade-offs would the organization be willing to make to achieve those metrics of success? For example, do we care more about false positives or false negatives? Do we care more about speed or accuracy?
- We researched your organization this past week, and learned you have _____ programs. Can you go through the impact that each of them has? If you understand how an organization’s activities fit into their theory of change, then you’ll have a good idea of how to design a project for seamless integration and maximum impact. Try to get a detailed understanding of a “day in the life” of a member of their staff.
- Who are the key organizational players and stakeholders? Are they involved in the project, and how are their views taken into account?
- What are the organization’s key objectives or focus areas for the year? Are they aligned with the problem you’re talking about solving?
You will probe the organization’s goals and success metrics in more detail during the impact map creation step.
Bottlenecks and challenges
Copy the folder structure template and all the templates inside it to create your own google folder for your project - it takes less than a minute! Simply download the folder zip file, unzip, and then re-upload to the correct place within DataKind’s google folder structure. Here are step by step instructions:
- What does it look like when things are working well? And what prevents your organization from operating like that all the time?
- What bottlenecks do your programs, activities, and processes have? As a data scientist, listen for minimize/maximize optimization challenges.
- What are the biggest challenges your organization faces in achieving your mission? Listen for pieces that data science could address.
- What are the pain points of the organization? If working with a consortium - what are the shared pain points across the organizations? Who experiences these pain points in the organization(s) (e.g. a specific program, a specific role/position, a specific geography)? Over what timeframe is the pain experienced? What does this pain point prevent the team from achieving? / How does this pain point limit success/organizational ability to create change?
- What else keeps you up at night?
Data
- What data has the organization collected? Does the organization know of any other data sources that will help them meet their goal?
- How is data currently being accessed and used? What inputs inform decision-making?
See this article for additional questions and details on how to assess an organization’s data maturity.
Solution brainstorming and impact
- Let’s set aside data concerns for a minute and assume that we have all the well formatted data that we need… if we could wave a magic wand and create any solution that you want, what would it look like? Who would use it? This exercise is aimed at getting the org to think about what levers would be most useful in their decision making. It provides a good exercise to visually imagine what the ideal goals of the project could be. Despite telling them to forget about data for a minute, while talking through this answer the organization will likely begin to realize data would be useful for this magical dashboard (this is a good thing…).
- What would be the final deliverable of the project idea, and how would it be used? Is it a static report? A dashboard? An algorithm to be implemented into a web app?
- What is the analytic timeframe for the potential project? To what extent is the organization interested in understanding the past (retrospective analysis), present (real-time analysis), or future (predictive analysis)?
- What would it look like to have addressed the bottleneck or challenge you previously mentioned? How would you describe a fixed solution to your pain point? What would an ideally run system look like? Since there might be multiple solution options, and we don’t yet know what would match their goals and infrastructure.
- Let’s enumerate the people who will be impacted by the solution: Who will be impacted? Are there any potentially negative consequences for them? How are you accountable to them?
- What are pilot or intermediate approaches to creating these solutions? If there aren’t any, what is a non-ML baseline that we can implement to know if our solution is worthwhile.
- What other organizations or entities could benefit from this proposed solution?
Implementation
- Are implementation and adoption subject to specific environmental conditions? For example, will the final product run on a mobile phone? Will it be adopted in a place with no electricity? A conflict zone? Any other constraints? What level of usage or accuracy needs to be established?
- How long would you estimate it might take for the solution to become outdated? Will ongoing maintenance be necessary? Does the partner organization have technical staff capable of managing the solution? What skills does the person responsible for maintenance have and how much are they willing to learn?
- When does a solution need to be in place?
- Who is the user (at the organization)? What amount of adoption support might be required? If 100% of its intended users adopt the solution, what impact will it have?
- What does a non-result look like? If you have a negative finding, what learnings will you get? This is helpful to dig into points of ambiguity and manage expectations.
- Can we expect executive support for this work? A project like this can involve considerable effort on the partner organization’s end. Who can commit the resources for that, and are they excited about this project?
Data tools
Finally, if the solution pathway proposed is a data tool, here are some questions to consider asking alongside stakeholder mapping and needs assessment:
- If we are proposing building a tool as a solution, are the technologies and tools in the proposed solution viable and affordable for this organization?
- How do we know that data tools solve for the stated pain points?
- What are the end-users’ levels of comfort with data? What is their level of trust of data, and what is that trust based on?
- What are the known gaps in the data and how do you solve for the gaps? Proxies?
- What collaborating partners would need to be on board for data tools?
- How fast would the solution be outdated? What would it take to sustain it, and does that maintenance capacity exist within the organization?
- What is the worst that could happen if we fail? If we succeed?
DEI Due Diligence
Within at least one discovery call, it will be essential to ask questions as part of checking for DataKind value alignment, specifically focused on Diversity, Equity, and Inclusion (DEI). Make sure to review the questions and means of verification in this Playbook article on DataKind Value Alignment. Most elements of DataKind’s DEI due diligence framework are best verified through a combination of discovery calls and research. Gathering information on DEI within the potential partner organization during a discovery call is essential because it is impossible to confidently know how DEI is incorporated into an organization’s work solely based on their public messaging. Choose questions to ask based on what you are and are not able to find in your online research.
Contributer(s): Caitlin Augustin, Caroline Charrow, Jack Craft, Erin Antcliffe, Matthew Harris, Arina Igumenshcheva, Ben Lebovitz, Sebastien Ouellet, Seward Lee, Srivalya Elluru, Dulcie Vousden, Rachel Wells
Contact us
If you would like to learn more about us, partner with us, or get in touch, email us at community@datakind.org
Subscribe