A recent post I wrote on LinkedIn hit a nerve. Talking about information/semantic rich projects failing due to the lack of organizational commitment is not new, for me or others in the field. Unfortunately, the reasons for these failures are often hidden to the organizations and are often pinned on bad technology, complexity, or a hard to measure ROI. While each of these may be front and center in the project failure, they are usually only indicators of organizations skipping steps. Major capabilities like search, personalization, AI, analytics, knowledge management are not solved with technology, a new experience, or new processes alone. They are solved when the organization is able to truly embrace them as a long term initiative rather than point in time software purchase.
In the past I have focused on building the foundations for an information ecosystem; governance, user understanding, tooling, modeling, organizational alignment, etc. These foundations are essential and help organizations move forward in a sustainable and durable way. However, building, maintaining and managing these foundations requires organizational alignment around the entire set of goals, processes, and activities necessary to manage semantics. This alignment is the semantic mindset and it needs to permeate an organization.
What we mean by “Mindset”
The idea of an “xyz mindset” is not new for organizations. It implies that something is important enough to directly impact everyone in the organization, that everyone in the organization has their role to play, and that this direction is provided by (and modeled by) the executive team. For example, a well-managed security program at a large enterprise requires much more than making people tap in with a badge to enter a building. Managing security requires executive support, funding, staffing, technology, training, fostering the right culture, metrics, and ongoing development and evolution of the security protocols. No individual piece of software, hardware, policy change, or training is going to address the systemic security needs of an organization. To do all this successfully requires a mindset shared by the entire organization.
The same can be said for enterprise semantics. They also require executive support, funding, staffing, technology, training, fostering the right culture, metrics, and ongoing development and evolution. The notion of a “mindset” sounds amorphous, but like the security mindset, there are a number of concrete areas that an organization can focus on.
Building it out
The following list is definitely not exhaustive, but it provides a good overview of the expansiveness of a true semantic program. Organizations that are going to successfully implement a knowledge graph, personalization, enterprise analytics, content management and orchestration, or any generative AI initiative need to address the following:
Organization wide models
These models provide understanding, modeling, and instantiation of the primary business information (products, customers, analytics, employees, etc) across the organization as well as a common language across the organization.
Well documented and supported models allow for system integrations, analytics, workflows, etc. exist, that are transparent, and are able to adapt to new business needs. Getting organizational agreement to these models is often far more difficult than people may expect.
Strategy to drive the model and the program
Any (all) semantic dependent projects require executive guidance and oversight on organizational goals and how they are going to be implemented and measured. Top line strategies need to address the capabilities and limitations of the organization’s ability to manage, resource, and instantiate semantics. Executive oversight is also required to make sure that cross-business unit alignment is addressed.
Dedicated resources to manage and maintain the models
Resources are essential to provide oversight, governance, maintenance, and evolution of the conceptual and instantiated models. This work needs to be part of people’s job description and have visibility up to the executive level. In particular there needs to be a person/role who can evangelize the semantic model throughout the organization along with people who can represent the models at a conceptual and technical level.
User Understanding
This cannot be said enough, it is essential to understand the goals, mental models, and expectations of both internal and external users and reflect them in the semantic structures like ontologies, taxonomies, metadata, etc.
Funding and Resources
The need for funding (along with the funding itself) is understood and is tied to the value of the initiatives the semantics are supporting rather than the semantics themselves. some examples include:
- Onboarding new business units, systems, or capabilities.
- Governance and maintenance.
- A well-maintained technology stack for managing the semantics.
- Documentation and training
Engineering Adoption
It is essential that the engineering teams are familiar with semantic models / structures and can develop with associated technologies and languages like RDF databases can be queried via SPARQL .Wit this expertise the engineering team will be able provide insights and guidance to the business teams during the modeling process and develop the solutions in a holistic way that supports the long term strategy.
Training and Documentation
All semantic structures, their governance processes, and changes need to be documented and available across the organization. Documentation needs to include ontology, taxonomy, metadata definitions, examples, usage guidelines, etc.
Appropriate training for the usage, management, and integration of semantics needs to be available for new business users, technical users, and general users in various forms and needs to evolve as the semantic models, the users, and the capabilities using the models change.
Conclusion
Just like the security mindset, the semantic mindset needs to permeate the organization. Your search, analytics, personalization, knowledge management, etc projects are not “one and done” projects, instead they require support across roles, business units, and the organizational hierarchy. Unfortunately, there are no shortcuts, there is no free lunch, nor is AI is not going to solve this.
Like the security example, semantics should be seen as a foundation in the organization, a foundation that supports essential capabilities. Just like no one asks for the return on investment (ROI) of a badge swiper or cameras in the parking lot, people are not asking for the ROI for the semantics. The ROI discussion is held at the capability level, not the semantics level. Thus, the question is not, “what is the ROI for adding synonyms to our subject taxonomy?”. Instead the question is, “what is the ROI of a fully functional search capability?”
