Over the last year, many published articles have continued to build on the suggestion that monetizing the ecosystem is what the future winners in perspective industries will do. I set forth the idea to “Influence what you can and monetize all you can” in your expanded ecosystem of digital services.
I have strongly suggested that one key strategy for success is to create customer-specific bundles with interoperability. What exactly is that? It is a set of processes that can make recommendations and allow the customer to “interact” with the bundle. In doing so, they receive a benefit and more value.
Developing an AI strategy to do this is important as humans can only scale to create personalized bundles based on inferred business rules pulled from massive amounts of data. Additionally, only some companies have all the data needed to pull those rules or the permissions to access such data.
When creating that initial personalized bundle, reviewing the end-state model that is best for your customers and long-term customer engagement is critical. What is “interoperability,” and why is it so important?
As part of the IBM AI Certification Course, students are introduced to a similar model.
The key point is that interactive customer engagement creates the best experience and a positive “cognitive” engagement. As such, creating the ideal bundle involves an ideal suggestion based on data and an interactive element to refine the bundle. By engaging the customer, they feel they have a personalized experience and are more likely to buy and stay in the program.
It may be correct that an AI-generated bundle with AI-generated pricing/payments is compelling. Many developers may seek to accomplish this, expecting the best results in the marketplace. While it may cost less to implement AI this way narrowly, it will cost you customers and sales because the engagement model lacks the human element of decision-making.
The tendency to create a narrow AI engagement model to perform a single bundling operation may be an excellent place to start. However, it is unlikely that your developers will have the data set required to “train” the AI Bot to perform what is needed.
The AI model required for interactive sessions is much more complicated. One challenge is to find a company with the data required to train the AI Bot on bundling use cases and, most importantly, the interactive nature of the shopping process. If you get past this, you must add the conversational element. Adding the human dimension is critical to the ideal long-term customer experience.
To explain this in detail, let’s discuss the last time I updated my golf clubs. I am an avid golfer and quite particular about my bag. I retool my clubs every 3 to 5 years. My most recent experience was an ideal example of how AI and the “human” shopping experience must come together.
There is massive investment in AI for golf clubs. Most AI is “narrow AI (ANI),” while the data sets are enormous and the outcomes are accurate, they lack the human element. In the base case, massive amounts of data are used by machine learning algorithms to predict the outcomes. In golf, that is, launch angle, backspin, and side spin misses..(plus a lot more, but enough said to illustrate the point). I can be directed to the right club by taking a few data points.
In my particular case, I did not like the recommendation. I prefer a club designed for higher swing speeds than I produce. I also prefer firmer chord grips, high bounce on my lofted wedges, and lower bounce on my gap for square-to-square shots. This is a perfect example of how the human brain differs from machine learning. My mind added layers of context and experiences.
What is described here is the preferred scenario for building an ecosystem through bundling. It involves both machine learning and large Language Models (LLM) working together with the customer to derive the best bundle for their needs. Machine learning is an excellent place to start, but it can never fulfill the context that a human can offer. Making that interactive builds customer-specific scenarios and, in effect, trust.
Relating this to our cognitive response model, we can outline how to create the best customer experience using predictive analytics (suggestion engine) and LLM to converse with the customer to understand their unique needs in human terms.
The updated model
Relating this to my example of selecting golf clubs by this into Gemini Advanced:
“What type of golf club should I buy for 100mph driver swing speed? I tend to hook the ball under pressure or late in the round. I prefer chord grips because of the firmer feel at impact and in humid conditions here in Singapore.”
The output from Google Gemini Advanced is what you might expect. It was high-level and lacked the data required to provide exact specifications. However, it did understand what I was asking, the tone, and the need to engage. The conversational element needs to be added; LLM can translate that and refine the data set for my particular case. That does not exist today, but it may represent a path forward for owning your ecosystem.
While this does not all exist today, beginning with the end in mind may help establish how to advance your bundling in your ecosystem.
