Data products and the data mesh importance has emerged as a consequence of a key factor:
the increase in the demand for Artificial Intelligence in the past 5 years. I still remember the commercial meetings in my first AI company.
We looked like aliens getting in the meeting room with potential clients, nobody understood a single word of what we were talking about, and certainly the interest was far from being a reality a nice-to-have, a commodity nobody would put a budget on.
Now it is a new flourishing industry and companies, understanding the value of AI and data products, start to fetch providers and to put resources into it.
The final goal of AI apps is to create actionable insights for teams. To do so the tools themselves must be actionable, meaning, they need to allow their users to create actions whose performance they can review and show a clear uplift in the result.
But most AI projects are still understood and carried out as techie projects managed by techies. From this comes one of the biggest problems with data products, in this case AI products: we forget who the final user is - typical AI needs are in the marketing & sales departments. The outcomes of AI projects have to be understood & actioned by marketing and sales professionals fast. A useful AI app is about bringing the predictions for the right marketing and sales tools to be actioned so that the KPIs are improved in days. And what is more important, the typical use cases can be shortlisted:
- Lead scoring AI app. So that sales teams know what potential clients to address first;
- Demand forecasting AI App. So that you can match your offer (whether it is a product or a service) to the demand you will have;
- Retention & churn forecasting AI App. So that you can increase the lifetime value of your clients as a key metric in the era of retention;
- Anomaly detection AI App. So that you can know instantly when there is any key business metric in an unusual value to act on time.
Why do we currently have hundreds and thousands of companies working independently with very expensive data scientists to count on a very similar if not the same solution? The future of the sector depends on proposing solutions to these universal problems that are easy to set up, to use and to actionate for any company. While, of course, more specific problems could be solved in-house, it is not optimal to develop common solutions independently and the economic forces will make this reality clear in the following years.
To put it in a plain way: we don’t need every single company to develop their own churn forecasting.
Download this post in PDF to have it whenever you want