Federated or Centralized? Jeffersonian or Hamiltonian? What’s your business preference?

Uh-oh, I am going to wade into the decades-old business dilemma of centralization versus federated business models. No, this has nothing to do with politics.  It is, though, a reaction to debates I am increasingly involved in during my own consulting work with companies aiming to unlock the strategic potential of digital- and AI-driven transformations. We had this debate often in the early days of emerging online capabilities, and now it’s spreading further.

Here's the basic question, and I am going to focus it specifically on personalization and customer experience: if we want to use AI to better empower, know, reach, show, and delight our customers (as in the five promises of personalization in my recent book), can “federated” lines of business, or functional teams, really make that happen? Does each business unit have enough data about a customer, enough it can offer to truly satisfy a customer’s need in a breakthrough way, enough tech support to deliver, enough risk management capability, etc. Or do CEO’s need to step back and put more weight at the enterprise level, more data integration, more design and coordination of customer experiences, more oversight, more training, etc.? Can entrepreneurial business units best optimize the growth of their customer relationships, or are those relationships really the enterprise’s assets? And when we enter a world where AI leads us towards integrating more data to better inform an AI engine, to better spot customer signals, and to manage the lifecycle of a customer journey, should company management models change?

Looking at some of the examples in our book, the answer seems to be yes, but carefully. Voya, an employee benefits company, realized that employees of the companies they served would be better off if they had an integrated picture of their retirement, health, and day-to-day financial management.  So they tore down many of the separations between their operating divisions to get that data integrated, offer one portal to consumers, and develop recommendation algorithms that are agnostic about which, if any, division benefits from the advice they give consumers to optimize their abilities to hit their goals.  Sephora, seeing that too much independent outreach from each of their retail category managers was overwhelming and confusing their customers, went through the hard work of integrating their databases, building orchestration capabilities to get the right offer for each customer, linking the data flow across all their channels so customers see the same promotions wherever they are, and capturing customer feedback that serves the whole business. 

In these cases, as well as many others, enterprise capabilities (and budgets) are getting rebalanced towards the center, to build these integrated capabilities. Managing them are central groups who aim to coordinate customer outreach and the nature of what the company offers at a given time to someone. That’s a tough role in companies where a central algorithm can affect a business unit’s ability to meet its goals. But taking a simple, “socialist” view that every model must be optimized for enterprise gain, not necessarily that of a specific business unit, misses the strategic nature of managing a business unit (or product) portfolio, and probably can’t consider every possible variable that would optimize corporate performance over the long-term. It can also negate the value of having tons of experiments going on to learn about the possibilities of new customer experiences and AI automation.  So we can’t just let a central AI run the show, and in reality, very few companies do.

Nonetheless, the personalized leaders we analyzed for our book are strengthening Enterprise-level roles and moving budgets, but they are also balancing this with new management models, such as monthly or quarterly resetting of priorities. Many are also positioning the leaders of these enterprise roles, such as Chief Analytics and Data Officers (CADOs), as “product managers,” whose “customers” are the business units. How much impact are they delivering for the BU’s?  What capabilities are needed for better adoption and impact? What new revenue streams and offerings could they unlock? The early days of “Chief Digital Officers” saw the same kinds of issues emerge, but now the stress is heightened as AI and powerful models can govern how the customers of an enterprise experience its services. 

So, while I have not seen the concept of business units or product P&L’s eroding in the name of higher growth through integrated personalization, I am seeing a rebalancing.  And like the debates on this that happen in politics, if the center to prove it can actually drive better value, and not just add overhead, then everyone can benefit.  But that’s usually a high bar, and likely to be one of the next big challenges companies face as they strategically pursue an AI-driven future. 

What are you seeing in your companies?  Everyone must be grappling with this…