Home » Uncategorized

Why Every CEO Needs a Chief Analytics Officer

Two former CAOs Make a Compelling Case for the CAO to Report to the CEO to build out an AI and Data Science Accountability COE.

Anthony Branda, MBA, Ph.D.,  Kevin Kramer, Ph.D.

  • Company CEOs today are being bombarded by C-suite executives with urgent requests for significant funding for Data Science and anything that can be labeled Artificial Intelligence (AI):
  • The CMO wants better segmentation, dynamic predictive scoring of leads, offer optimization, better insights from the buzz of social media, and so on – to acquire new customers, retain existing ones, increase market share, enhance brand awareness, etc.
  • The CIO/CTO wants to replace legacy infrastructure, build ever larger data repositories and ‘modernize’ core platforms with more data, more automation, more standardization and more documentation.
  • The Chief Digital Officer wants more funding for Chatbots, SEO, adoptions of a broad range of tool sets from Google Analytics to tagging and tracking of all digital assets.
  • The CFO wants AI to process real-time signals to identify what makes money, what loses money and, to dynamically re-allocate budgets and investments to optimize net revenue.

CEOs must act now to fund significant increases in AI spending, lest as once was said of ‘Big Data’ – to avoid becoming extinct!

Maybe it’s time for a chill pill?

Let’s step back for a moment. What problem are we really trying to solve? And what are the premises upon which we answer this question? It seems to us that the hype and frenetic rush to embrace any and all things AI is based on three themes in play today in boardrooms:

  1. Corporate America has finally recognized that data/information are their single most important asset (always has been, but that’s a storyline for another Blog). Today there is more data, petabytes more, than ever before.
  2. Data-driven insights/evidence-based decision making should be the basis for guiding business actions.
  3. For those with the wherewithal to move quickly, Data Science and AI will enable machines automatically to mine data to generate optimal processes and outcomes, lower costs, and confer significant competitive advantage.

The challenge then, as we see it, is:

  1. how can the various tools of data science/AI be employed meaningfully to tap into this mother lode of data to improve business processes, enhance customer satisfaction, and increase bottom line revenue
  2. who should be the ‘owners’ of the use and deployment of these tools and to whom should they report in order to maximize enterprise value.

We believe the root answer is organizational – namely, should Data Science/AI continue to be a decentralized set of activities as it is mostly practiced today? Or an enterprise C-level mission, reporting directly to the CEO?

We are strong advocates of the latter, an enterprise wide analytics function, headed by a Chief Analytics Officer (CAO). The CAO is a direct advisor to the CEO, serving as a fact-based communicator of the ‘truth.’ No more reporting to the CMO, the CIO, the COO or some other Administrative function.

This point of view challenges the existing corporate norm where analytics is decentralized, with each function and P&L owner having its own pod of data scientists, just as it has other SMEs, i.e., financial analysts, business analysts, HR, etc.

The shortcomings of decentralized analytics are manifold: Analytic SMEs are often ‘orphans’ – just one or a few folks in a line area operating in isolation; few opportunities to engage other analytic SMEs to discuss and share best-demonstrated practices or learn new skill-sets; no real career pathing which translates into high turn-over; little exposure beyond their own domain knowledge limiting the full range of how their recommendations may impact other business units; little development of enterprise-wide institutional history, let alone creation of broadly agreed upon KPIs.

In our prior staff positions leading Business Intelligence functions and now as consultants we have repeatedly been brought in to help fledgling or stymied corporate analytics functions get pointed in the right direction. The common themes we often are asked to confront arise from this decentralized approach:

  1. The CAO is not really a C-level executive, is subordinate to another C-level executive; has to adhere to the goals of that division and not those that best benefit the enterprise
  2. CMOs, CIOs, other C-suite executives have self-interest to control the narrative around the performance of their respective area – hence, they are threatened by the CAO function that measures performance and return on analytic investment
  3. How to gain traction when analytics are decentralized – little standardization across tool sets, modeling approaches, operational definitions for KPIs, data quality issues, data silos, etc.
  4. Absence of rigorous testing protocols (experimental design), measurement periods, determination of causality as opposed to correlation, agreement on success metrics, and the like.
  5. Because of decentralization the hiring of the CAO has backfired on the organization ; many have recruited the wrong leader with an overemphasis on the technical side(while important) and not on the business strategy. See Tom Davenport’s Competing on Analytics Book. The CAO must possess a knowledge of the business, executive communication skills combined methodologies. Analytics execs needs to be collaborators to take insights to action by collaborating with other enterprise stakeholders. Analytics isn’t an end it’s a beginning and there must be accountability for the analytics to work.
  6. The CAO is the same role as the Chief Data Officer as the data helps drive the analytics. You need accurate and timely information/data to have the best analytics, so we believe the roles should not be separate. (More on this in future blog posts.)

It is long past due for corporations to embrace an enterprise-wide solution for business intelligence and establish a CAO as a direct report to the CEO. Pockets of analytic activity are by definition, sub optimal and often impede rather than accelerate the successful deployment of analytics throughout the enterprise. The success of an enterprise approach has been well documented by Tom Davenport in his book, competing on Analytics. The time has come for the CAO to really earn his or her C stripe.

We draw from our own experience at a large consumer oriented financial institution which built a 100+ person enterprise Business Intelligence function that supported 16 business units in the US and internationally – marketing, risk management, collections, cross-sell, up-sell, back-office optimization and the like. While the BI function was separate organizationally from the units it served, it absolutely was the case that data scientists worked on a daily basis at the business sites, attending meetings, helping set the agenda, and supporting through analytics, the achievement of each areas respective business goals.

By having an enterprise BI function, it

  • enables data scientists to share best practices and learn from one another
  • aligned on best approaches and tools for modeling, segmentation, forecasting, and experimentation;
  • built ‘single’ source(s) of ‘truth’ and 360o views of prospects and customers;
  • minimized making the same mistake twice
  • facilitated a relentless focus on test and refine, fast fail, rapid prototyping
  • facilitated rotations across business units to expand the analyst’s awareness of the enterprise rather than one silo;
  • provided career pathing, reducing staff turnover. Having a job family for analytics is key!

All of this is best achievable when the CAO is clearly supported by the CEO and that the CEO fully embraces data-driven insights as the standard for business management.

If you are a CEO entrusting your CMO. CIO/CTO, strategy lead, or some other C-executive executive charged with executing an enterprise analytics strategy, please reach out to us. We strongly believe, based on a lot of empirical evidence (right, we too are data driven) that an enterprise-wide analytics mind-set and infrastructure is the best approach to optimizing the huge potential value to be achieved from data science and AI. We can help define the blueprint and the journey to get there.

Tony Branda and Kevin Kramer.

[email protected]
[email protected]