Why IT-OT Integration Projects Fail Before The Technology Is Selected

Ask why an IT-OT integration programme underdelivered, and you will often get a technology-focused answer: the historian didn't scale, the middleware choked on protocol diversity, or the analytics layer never reached the shop floor. Those answers are convenient because they are auditable – but they are usually downstream of the real problem. By the time a tech shortlist is drafted, the outcome of the project has already been shaped by organizational conditions. These are factors that impact the potential for industrial agility, which depends on a cycle – plan, sense, decide and act – running fast enough to absorb volatility in energy, supply, labour and demand. IT-OT integration is the connective tissue of that cycle.

Three conditions decide whether IT-OT integration holds:

• Are IT and OT optimizing against the same risk model?
  OT owns safety, uptime and process integrity, while IT owns confidentiality, patch cadence and enterprise agility. Both positions are valid and aligned with functional requirements, yet each carries a different definition of an acceptable failure. For IT teams, an unpatched system is the exposure. For OT, the patch is the exposure, because validating a change on a running process line costs production hours – or worse, may cause a safety event. The same asymmetry runs through remote access, where a vendor connection that IT treats as a routine support ticket is, on the plant side, a live path into a safety system. Left unreconciled, these differences resurface as a proxy war over segmentation and third-party access, and every fresh disruption reopens the same argument instead of triggering a response, which is where agility is lost. The tell is escalation: when the issue reaches an executive because no one below them holds authority over both sides, the problem is governance.

• Is there an organization-wide shared asset ontology?
  A pump is a tag in the historian, a line item in the CMMS and a node on a P&ID, often with nothing linking the three. Connectors deployed before context move disconnected data faster, but they do not make the information usable. The 2025 Verdantix Green Quadrant on industrial AI analytics software found that leading vendors convert messy OT, IT and engineering signals into a single navigable asset model at ingestion, whether by knowledge graph, as Cognite does, or through near-real-time validation and cleansing, as C3 AI does. Norwegian hydropower producer Skagerak Kraft shows the buy-side work involved: before scaling anything, a cross-domain group met weekly for close to a year to agree what the business meant by its own core objects, then built an enterprise data model drawing from standards such as the Common Information Model. That is an ontology decision, not a procurement one, and it is what allows a decision to be made in minutes rather than after a week of reconciliation.

• Is the business case anchored to a system, or just a decision?
  Transformation programme funding is often approved on the promise that a new platform will enable faster, better decisions, but those benefits are usually defined in broad terms, making it harder to demonstrate clear business impact and value. Contrast this to the continuous pharmaceutical process improvements documented by Seeq, where a manufacturer's quality engineers built a continuously updating dashboard to monitor KPIs, flag deviations and calculate when manual operations are due. This recovered roughly 30 minutes per engineer per day previously lost to aggregating data. In specific terms, industrial software buyers are looking to shorten the journey between ‘something looks wrong’ and ‘here is the action we are taking’, and as such are seeking AI capabilities that sit inside daily processes to deliver this. Agents deployed by Cognite, IFS and hyperscalers such as Microsoft can compress that distance, but a foundation of proper process and data ownership is non-negotiable to gain value from adoption.

So, why do IT-OT integration projects fail?

IT-OT integration projects fail before technology selection because the questions that decide the outcome are overlooked. The big questions are: who has authority when IT and OT disagree? Do you have a robust asset ontology? Which core decisions need to get faster? Arrive at tool evaluation with those answers and it becomes a precise exercise, testing platforms against the problems in need of solving to unlock true industrial agility.

For more on how industrial firms are improving decision-making under sustained volatility, check out our research on industrial agility and the Verdantix Dislocation Index.