Identify Business Opportunity With AI
I. Identify Customer Needs
1. A Product Co-creation
In order to identify the value provided to the customer, it is necessary to understand and address their needs. This topic has often been addressed in multiple research papers, and as such there are several theoretical frameworks such as Payne et al. (2008) with a value co-creation framework. The framework by Payne et al. contains 3 main ideas :
First, co-creation can help companies bring the customer’s point of view to the forefront and improve the initial process of identifying customer needs and wants.
Secondly, in this research paper, they theorise and give examples for the encounter process. They correspond to the different ways in which the company can achieve this co-creation. The framework suggests three broad forms of encounters:
- Communication encounters: Creation of content in order to connect with customers, promote and create dialogue (e.g., advertisements, brochures, internet landing pages and documentation).
- Usage encounters: Understand customer practices in the use of a product or service by studying its interactions.
- Service encounters: Study customer interactions with customer service personnel or service applications (e.g., via a contact center).
Managing value creation by co-creation includes setting goals for both customer and supplier, and evaluating whether current encounters are achieving these goals.
Finally, value co-creation demands a change in the dominant logic for marketing from “making, selling and servicing” to “listening, customizing and co-creating”. The prototyping method is favoured for the realisation of this co-creation.
2. Prototyping and Getting Feedback
This prototyping process is one of the foundations of the Lean ideology that is proposed in the article by Nguyen-Duc and Abrahamsson (2020) for AI project management. The aim of this framework is to minimise the risk of making a product that does not meet customer's expectations. To do this, companies can perform continuous experimentation with a minimum viable product (MVP).
The goal is to create a continuous validation loop by iterating between customer needs, business needs, and the product.
Although Payne et al. describe its use as suitable for major changes in an industry, its use is not necessarily generalizable. This framework has only been tested for service companies and for a few manufacturers.
A good approach to creating MVPs for AI consists of three key points:
- Minimum Viable Model: At the beginning, create a simplified model with a correct prediction and don’t spend to much time trying to have the best one. It is not necessary to make complex neural networks, a linear regression or a random forest can already bring a lot of value quickly.
- Minimum Viable Platform: You don’t need to have a big infrastructure to store and process data. Creating a data lake at the beginning of a project can be tedious, using ready-to-use Platform as a Service (PaaS) or Software as a Service (SaaS) solutions is much easier.
- Minimum Viable Data Product: At the beginning of a project, you may not have enough data to train your models. However, you can still test your idea using similar solutions that do not require this data. With A/B testing, the goal is to see if the consumer is interested in your idea or not.
The purpose of these techniques is to highlight opportunities, identify failure points, improve service enhancement, re-engineer processes, and support differentiation. On the other hand, the use of prototyping is fully justified for identifying user value, as it allows very a rapid validation of consumer interest. This technique makes even more sense as many companies do not have any use of AI within their firm.
II. Identify Artificial Intelligence Use Opportunity
To understand how companies create value, it is important to first study how companies determine opportunities for using AI and how this materialises. To get some insight, I interviewed multiple Chief Technical Officer and Chief Data Officer of companies that are using IA. The method unanimously used is the establishment of exchange between the business and IT parts of companies.
1. Annual Meeting With Business Leaders
A method proposed and used by some, especially in large companies, is to periodically schedule discussions between business decision-makers and IT representatives related to data. This allows the opportunity to create an exchange on the development vision of each branch of the company while trying to identify how Big Data and AI can enable them to achieve their goals.
To have the best innovative process, it is important for everyone to have a good understanding of what AI is. For example, to structure this exchange, some use the Design Thinking approach based on business data. Customers or internal decision-makers talk about the data at their disposal, their medium and long-term objectives, while explaining the difficulties they encounter on a daily basis. On the basis of this exchange, the Data Scientists identify a list of use cases which are then prioritised. The use cases with the highest feasibility and added value are then carried out.
2. Spontaneous Request From Business Leaders
Most companies operate by allowing the various business representatives to express their needs spontaneously without waiting for a particular event. This allows them to have greater flexibility in their innovation processes, but it is not necessarily the best way to proceed.
The problem with this method is mainly due to a misunderstanding of AI by business leaders. On the one hand, the image given by the media makes people think that this technology can “magically” solve all problems, so some people have unrealistic expectations. On the other hand, due to the lack of understanding, some people do not see its potential use in their business. This problem is especially present in companies that are still at the beginning of their digital transition, so it is necessary to explain and train the decision makers of all companies to improve their understanding in order to allow them to propose viable projects.
3. Spontaneous Ideas From Data Scientists
Another approach is for data scientists to propose their ideas for using AI. This approach is based on the analysis of data and processes implemented in the daily activity of the company.
This technique is not widely used by companies, as it is considered by some to be inefficient and relatively expensive. The problem with this approach is that the data scientist does not know what the real business problems are, so expensive algorithms are created that do not meet any needs. In order to avoid useless projects, the other methods described above are preferable because they are based on discussions between the IT and business parts of the companies.
4. Discussion Between Companies
In addition to having discussions within the company, business leaders are taking inspiration from outside of it. For example, there are discussions between companies in different industries to understand and see what can be done with AI. This provides inspiration on potential uses of the technology and gives rise to thinking about how to use it in their business.
Some business leaders schedule weekly meetings with other business leaders from other companies. For example, a CDO from the banking industry may discuss and be inspired by a transportation company. Observing other industries gives them other perspectives that don’t limit them to ideas specific to their field.
5. Use Well Know Use Cases
Another common practice is to use AI for issues where its use is recognised, such as customer attrition or "Churn". The same is true for banks to improve their detection of banking transactions to avoid money laundering or terrorist financing. Other types of classic use such as these can be found on the internet and are also a basis for reflection for companies.
If the company is carrying out a first AI project, it is advisable to carry out a project on a theme where its use is recognized. This reduces the probability of project failure, especially if the development teams do not have much experience or if it is the first time that the company’s data is being used. Because the topic is known to be suitable for solving with AI, many resources will be available on the internet, which can help with development. Starting a project that is too ambitious from the beginning will increase the probability of its failure and therefore it will be more difficult to convince internally to realize new projects using AI.
6. Try New Research Paper
The last method is based on the study of the latest research in the field of AI. This new technology has many researchers who are demonstrating the possibilities of new uses through their work. Every day new research papers appear with sometimes new uses and new approaches being proposed. Thanks to these advances in research some companies can offer new services.
III. Conclusion
All these approaches are suitable for the development of proofs of concept (POC) within companies. These POCs are an opportunity to quickly test ideas by demonstrating their feasibility using simple and inexpensive techniques. Based on the results, it is then possible to determine the potential uses and investments related to the project.
Furthermore, it is important to specify that even if it is justified to use AI to solve a problem encountered, it is not necessarily useful to integrate a solution. Indeed, it depends mainly on the technological maturity of the company. For some, the use of AI is necessary because it meets a deep need with no other possible solution. However, for most, AI is just an unnecessary asset to the viability of the business. For companies just beginning their digital transition, using this technology may not be the best idea from the start, as it requires a lot of upfront investment in data, infrastructure and people.
Compared to the results found in the different research papers dealing with this topic, companies mainly use the co-creation method presented by Payne et al. Moreover, this is partly in line with the approach proposed by Nguyen-Duc and Abrahamsson (2020) which consists in using the Lean ideology to understand the user need. This method presented in the book Lean Startup by Eric Ries is based on the practice of continuous improvement based on the understanding of users’ demands and feedback. However, through the interviews, new methods were presented that were not defined in the literature review. Among others, it is possible to find opportunities to use AI through discussions between companies, the identification of classic cases or the use of research papers.
Nguyen-Duc, A., & Abrahamsson, P. (2020). Continuous experimentation on artificial intelligence software : A research agenda. Proceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, 1513‑1516. https://doi.org/10.1145/3368089.3417039
Payne, A. F., Storbacka, K., & Frow, P. (2008). Managing the co-creation of value. Journal of the Academy of Marketing Science, 36(1), 83‑96. https://doi.org/10.1007/s11747-007-0070-0
