Everyone who has talked with me about CRM and marketing automation architecture has heard my bizarre lecture about the water cycle in nature and how it is related to CRM architecture. The water cycle in nature is a process or system where the water flows from one reservoir to another, for example, from rivers to oceans, from oceans to clouds, and from clouds back to ground. I’ve simplified the idea by thinking that in the CRM architecture, there is only one direction of the flow.
When I’ve been reading machine learning and AI literature in the evenings to get better relaxation after using my brain the whole day, I think the water cycle analogy might be too complex and doesn’t fit the AI hype we have now. I should rethink the analogy and explain it in terms that AI experts can also understand.
Everyone who has worked with AI knows the gradient descent algorithm. For those who are not AI experts, the gradient descent algorithm is an optimization algorithm that can help to find the local minimum. Presented with non-mathematical terms, it means that if you are in the mountains, pee a lot, and your feet get wet and warm, you are in the local minimum. If not, someone else gets a surprise.
The data should always flow towards the local minimum in the CRM and MA architecture; an architect should always have a clear view of how data flows in your system (in the ideal situation, this is also documented somewhere). Last, you should have some mental models to support your reasoning.
