Why do some companies succeed while others fail?
This challenging question is at the very basis of my research interests and agenda as a Doctoral Candidate, at Cardiff Business School, previously a masters student at Cambridge University
My areas of research include
What determines a company's or organization's ability to perform? People, Technology, Money, Competition, Macro-Economics and many more come to mind. But, do they really matter and how can we prove this?
What does good or poor performance even mean? How do you measure it? Is it stock price for a public firm? Market Share, Margins or Talent Satisfaction? These are some reasonable KPIs.
Which of these factors matter more than others, and are they the same across all types of organizations, industries and geographies? Does a company in Florida stand a better chance of winning vs. one in Nebraska?
Technology Adoption and Impact
How important are Advanced Analytical Technologies (big-data, AL, ML) to the performance of organizations in the Financial Services industry?
How can we diagnose the technological needs of an organization and predict the impact that certain technologies will have on their performance?
My MAIN Research Question Therefore is:
Can the performance of a single organization be predicted using ML-based predictive diagnostics?
Sub Research Questions include
Does the Technology stack of an organization impact its performance? (and to what degree)?
Does the Talent Pool of an organization (at various levels) impact its performance? (and to what degree)?
Does a firm's Intellectual Property Portfolio impact its performance? (and to what degree)?
Do a firm's partnerships impact its performance? (and to what degree)?
How do other, additional factors, impact performance?
Can a predictive performance model allow for timely intervention and optimization (ie. change the course of events)?
Qualitative factors can and must be quantified to allow for empirical analysis
A wide array of KPIs can be used as determinants of performance
A sufficiently large and representative sample of data can be used for analysis and allow for general model development/application across broader subjects
Interviews can be used to augment and validate key aspects of the research, for example the underlying factors
This data is comprised largely of publicly available information
Suitable Machine Learning methods can be applied to meet the analytical objectives of this research
Such a large area of research clearly requires scoping to make it far more manageable. I have decided to scope the organizational type to public companies in the Financial Services industry, a domain I have worked in now for close to two decades. These two constraints greatly reduce the population to a more reasonable scope, and should allow me to establish a strong basis upon which other domains can be explored.
Factor and KPI Research, Quantification for Analysis
The research methodology I have selected starts with the development of measures for analysis of a wide array of factors which I hypothesize have an impact on company performance. This is derived from academic and industry literature as well as my personal experience of 25 years as an industry executive. Alongside this, I am developing the KPIs, the outcomes which are indicative of success and failure, against which the factors can be evaluated.
Many of the factors have traditionally been very hard to quantify and measure. As an example, how do you quantify the talent of an organization? The literature is very sparse in this realm, and is referenced as such. So, how does one measure the unmeasurable? This is part of the work I am undertaking, initial results of which can be seen on the Talent Analytics page.
The data sources for this research are comprised, first of all, of publicly available information.
Brightdata.com through their BrightData.org research support program, is providing the necessary public information to support the analysis.
Other data includes public company performance information from Wharton Research Data Services.
Data Analysis is being performed using a variety of tools, including those from Qlik.com, Dataiku.com, Google Cloud Platform and more. The analytical techniques being used vary based on the needs, and include various forms of regressions, Natural Language Processing (NLP), k-means and KNN for clustering, Random Forest and many other supervised and unsupervised methods.
As a practicing academic, my interests lie in the application of my research to organizations and their ecosystem. Some key use cases resulting directly from my research include:
Sales - Scientifically predicting the benefit for an organization to adopt, utilize and benefit from various technology products or services is the ever-missing proof that leaders need in order to make and ground their strategic investment decisions. Naturally, this speaks not only to CIOs, CTOs, CPOs and others within an organization, but also to vendors selling to these parties.
Talent and Training - Evaluating the current skillset, capabilities and behavior at individual team member, all the way up to aggregate firm levels. This allows internal leaders to evaluate their teams, capacities and ability to meet various project priorities, as well as allow talent firms to seek to augment as needed.
Investment - A deeper view into current and predicted firm performance offers buy-side investors an edge in their investment decisions.