Interested in how AI is revolutionizing small business support? Keen on diving deeper into AI’s role in global development? Check out our Bizzy online journey. Our series of webinars and articles will delve deeper into how AI is becoming a strong ally in combating poverty.
ONOW recently joined YBI's High-Flyer program that aims to support high-growth potential entrepreneurs throughout their journey - from connecting them to the right resources, to building useful skills, gaining access to capital, or simply improving our tracking on whether our interventions are actually useful. Over the coming months, stay tuned as we share our approach and learnings in these important areas for high-flying MSMEs in SE Asia!
A Scalable and Reproducible Approach
One of ONOW's primary hopes is to create scalable and reproducible processes as we seek to provide a coach for every business. IPA recently released their Best Bets for emerging opportunities for impact at scale, and listed "consulting services to support small and medium-sized businesses" as one of the fourteen best bets for the world. But if we want to create a process that truly scales this idea, we need data to fuel this endeavor. But MSMEs don't have time for lengthy surveys without getting something immediate in return for their time. So our approach starts with a two-sided product that seeks to provide confidence and insight to MSMEs, their coaches, and potential capital providers.
We're calling it the "High-Flyer Score", which combines personal details from a business owner with externally-curated data. This is then combined and analyzed to provide MSMEs with a simple, but detailed report on what it means for their unique situation. And if business owners have questions about anything, we're prepared to combine this tech with touch from real-life coaches to provide nuance and detail.
But spotting businesses with high-flying potential is a complex problem with many factors at play. We need data to scale our approach, but we also need simplicity to distill this into an actionable and valuable output that not only helps us identify these businesses, but provides them with clear action steps and confidence on how to proceed.
It's clear that predicting entrepreneurial success is a complex task. YBI's previous research in the High-Flyer program states factors such as "educated males living in an urban area" as being more likely to succeed in their test population. Or having an existing business with an ongoing track record of success being key to growth probability. However, these metrics were all retrospective, and are likely proximate factors and not the ultimate factors that lay underneath the surface.
At ONOW, we're trying to build upon this approach by leveraging existing research on ultimate factors of business success, and using them to create a quantitative and (hopefully) predictive model. In addition, we're taking motivation from existing modeling techniques in credit scoring and predictive investment security selection, and adjusting them to our target MSME audience. This leads us to a factor-based methodology that decomposes the complexity into defined factors that are combined to create a unified output. While imperfect from a modeling perspective, we find that this approach allows us to proceed with limited and imperfect data, and provides a high degree of explainability that is crucial to the process. We want actionable output, and not simply academic-only results that are difficult to implement.
The spectrum above shows a range of factors that all interact to influence a given business' probability of success and growth. Our program goes beyond traditional standardized training as we look to coach businesses across this spectrum. Some areas are obviously within an individual's control - such as skill training, or financial success. Others are mixed - like psychometrics, access to financing, or the nearby competitive landscape. And finally, others are highly influential, but simply outside of anyone's control to change in the short-term - such as the political stability or infrastructure of their environment.
By creating these discrete categories, we're able to provide clear thoughts and advice on each individual category, while also building a total profile that combines all the factors into a final output and what it means for the business owner's chances of success, and what they can do to influence that moving forward. This provides business intelligence and clarity that MSMEs desperately need in the challenging environments where they operate.
High-Flyer Scoring: Packaging is Important
With such a complex environment with multiple factors to consider, it can easily become overwhelming for MSMEs and their coaches to discern what to do and what it all means. This is where predictive modeling really shines, as we create weights for the factors, and distill everything down to a single number with an easy-to-interpret meaning of a score out of 100 - very similar to a grade you may receive in school. This familiarity is important to the user experience of both the MSME and the coach in the loop, and increases the usability of what can otherwise be a complex and academic modeling output.
To provide a simplified example of the approach with a hypothetical business owner in a made-up location:
Score (of 100)
The business owner's current political environment scored in the 20th percentile due to recent instability around the election.
Coming out of COVID, the inflationary environment continues to be elevated far above normal levels. This has particular importance for the business as it relies heavily on imported goods for their production process. It also introduces more risk as the business looks to expand its operations.
The business owner has a very high score in openness to new experiences, self-efficacy, and risk tolerance. These areas have been shown to highly correlate to possible entrepreneurial success.
Financial Tracking of Existing Business
The business owner has kept detailed financial records for the last two years. This gives confidence in the ability to accurately manage their finances and attention-to-detail as they look to expand.
This business owner has many great personal attributes that are highly correlated to ability to grow successfully. However, they are working in an exceedingly challenging environment due to political instability and high inflation rates. They may be able to overcome these challenges with their personal and business traits, but we recommend caution in the short-term as the environment is in a very challenging cycle.
This report can be given to the business and the coach, and helps each party identify areas of strength, weakness, and what it all means. The components shown above are the output of qualitative or quantitative modeling processes. For example, the political stability score of 20/100 is based on a predictive model that looks at the current environment compared to past scenarios and potential future paths before producing a score of 20. This would represent a very challenging environment akin to a 20th percentile scenario, but is also outside of the MSME's control. The well-known phrase of 'a rising tide lifts all boats' would be working in reverse in this case, with the political tide out to sea, and businesses suffering as a result.
Meanwhile, the weighting (30% for the political example) is the product of a separate modeling process. We can start with simple heuristics from existing research materials, and as we gain experience with ultimate outcomes for business owners, we can retroactively look at the factor scores that preceded the outcome. This allows us to continually tune and/or fine-tune the model for different contexts and environments. And as the dataset grows, we can deploy increasingly sophisticated modeling techniques.
We're just getting started, and as we experiment with more business owners, we're likely to find important details about what information we need to collect from them directly, and opportunities for simplification through things like multicollinearity or other layers of ultimate causes that we're missing in this first iteration. We're seeking to find a simple, but not simplistic, causal model that can give us a fuzzy picture of a single business owner's potential to grow and become a high-flying business that has an oversized impact on their employees and community in the process. We don't know exactly what we'll find, but it's an exciting journey to see what comes from it all.
Make sure to subscribe and stay tuned for all the updates as we progress on the path. And let us know if you have thoughts or ideas on strategies, factors, or other areas that we might be missing here! We want to create a community of like-minded thinkers on this subject that believe we can find better ways to help MSMEs around the world, and would welcome any input you have.