All models are wrong... but some are useful!

In this model, analyst groups are associated with business units and corporate functions. There is no corporate reporting or consolidated structure for analytics

Analysts are close to the business, so they can be very responsive. They are also likely to learn the nuances of the business very quickly

It is difficult to set analytics priorities across the organization. It is also difficult to develop staff, and take advantage of specialized skills sets across the organization. There is almost always a lack of communication between analytics groups in this model.

This type of organization is best fit for a diversified corporation where the multiple businesses have very little in common. Reading between the lines, it's probably not the best option for a hotel or casino company, although many are set up this way currently due to organic or grassroots growth.

In this model, analysts are all part of one corporate organization. They may be assigned to business units or functional areas, but they report to the corporate unit, and the corporate unit sets strategy and priorities

It is easy to invest in specialized resources because they can be deployed for key projects across the organization. Supporting high priority projects is also facilitated in this model, as it is easy to reprioritize and deploy analysts wherever they are needed. Analysts can build a community and gain new skills, it is generally easier to recruit talent, because there is a demonstrated corporate commitment and an established community to join

Distance can be created between analysts and the business problems if the analysts are all located at corporate

It is easiest to deploy this model if there is already an organizational precedent for shared services at corporate. If there isn't an existing awareness of the value of analytics, and the benefits of working with this group, the group may not have enough "work" from the field to sustain operations Further, it may be difficult to work with lines of business who feel as though analytics are being "done to them" instead of controlled by them

in this model, there is one major analytics group in the organization that reports into a business function that is the primary consumer of analytics. This unit may also act as consultants for the rest of the organization. In hotels, the revenue management function may behave this way today. For casinos, it could be the casino marketing group

Analytics are deployed against key business initiatives, and the analysts stay really close to the business, so they can be highly responsive and build local knowledge. However, they are also available to support projects in other areas as they arise.

Opportunities to apply analytics in other functional areas may be missed with this laser focus on one functional area.

For organizations that are just starting on their analytics journey, this model could be good. It provides focused attention on high-valued business initiatives, but there's some flexibility to solve other problems. Additional focused groups could be added as new projects are identified, or the group could be redeployed

Analysts work for a central organization, and business units "hire" analysts to work on their projects. This is a bit different than the centralized version because here the analytical priorities are set by the lines of business as opposed to centrally.

Key analytic resources are positioned to solve problems across the organization. The right fit skill sets can be deployed against key projects. Analysts who are deployed to lines of business can build close relationships with decision makers

If enterprise focus and prioritization activities are weak, analysts may not be deployed on the projects that deliver the most value, but instead become focused on the projects for whatever line of business executives yell the loudest.

The analytics organization must understand the value of the work provided, and be able to set priorities. It will be crucial to market and sell to internal clients to keep building the project pipeline

This model is a collection of decentralized analytics groups that report to business units, but have formal ways to collaborate. This could be a steering committee or a chartered enterprise governance committee, but there is an organization that ties the groups together

This model has an immediate enterprise view, with coordination on priorities, initiatives, resource deployment, and analyst development

Committees may lack clout in the organization. It could be difficult to establish standards, set priorities, and share resources for corporate analytics initiatives

This model works well in large and complex organizations where business units share some, but not all, things in common

Decentralized analyst groups are embedded in lines of business, but are also members of a central coordinating structure that builds a community of analysts at the enterprise level

This structure builds a community of analysts that can share experiences and best practices, and creates opportunities for learning and development

This is a less formal arrangement than the federated model, so this group, while good for best practice sharing, rarely has the power to assess corporate analytics needs, prioritize projects, and manage analyst career paths.

Organizations can adopt this model as a first step on the path toward a centralized, consulting. or federated model. It is useful for organizations that want to promote community learning and development, but don't yet have a mandate for any corporate level initiatives

Our industry is at a crucial inflection point where analytical talent must keep pace with technology (and vice versa) to be successful

the enabling technologies for analytical automation are introduced department by department. Users become accustomed to incorporating system results into their decision making. As the analytical culture grows in these departments, the enterprise begins to prepare for the organizational and cultural changes required for more holistic and synchronized decision making

is about manual information sharing. This is probably the phase where the cross-departmental Iteam responsible for data governance is taking a strong role in ensuring a clean, credible, and unified version of of the truth in the database. Departments begin to manually incorporate one another's data into decision making.

The phase is where cross-departmental data sources are included in the automated systems, changing analytical results

After routine decisions are automated and users are operating on a single source of the truth for synchronized decision making, it is time to think about _______. Organizations can think about new data sources to incorporate, or new initiatives like real-time decisioning or location-based marketing. which requires different access to data and analytics. This phase might bring the organization right back to the establish phase if new technologies or new departments need to get involved, but with an established process like this, the innovations are much more likely to succeed