HOLDAL Group

Elie Tanios

Operating Model Modernizer

Digital and AI programs are decided by process ownership, data foundations and workflow design.

None of those is an IT deliverable. Most transformation programs are governed as technology projects. One function owns them, the budget sits in one cost center and the steering committee reviews system milestones: configuration, testing, cutover. Administratively convenient, but analytically wrong.

The real scope is the operating model: which processes run end to end, who holds decision rights, what data each step depends on, which controls make the output auditable and how results are measured. Those are a business decision, which makes digital and AI transformation a shared obligation of finance, operations, commercial and supply chain. Software encodes the decisions it is given, and the ambiguity where they are unresolved, which is expensive to reverse once transactions post against it.

Process Ownership and Workflow

Each process needs an owner with authority over its design, not a coordinator who convenes meetings, and ownership runs from first trigger to final posting. The redesign is the substantive part: removing duplicate handoffs, moving validation to the point of data capture, dropping reconciliations that compensate for a gap earlier in the chain. Rebuilding an existing process in new software reproduces its defects at higher cost.

Data, Integration and Controls

Shared systems require shared definitions. If cost and margin are defined differently across business units, consolidated reporting aggregates the disagreement and management inherits it. Agree the definitions, assign stewardship for each master data domain and measure quality against stated rules. Cost and risk concentrate at the integration layer: which system is authoritative for each domain, what the interface contracts specify and how failed transmissions are caught. Controls belong in the design, since segregation of duties, posting restrictions and an audit trail are cheaper to build than to retrofit.

Adoption Follows Design and Competence

Adoption follows workable design and sufficient capability, not communication volume. Resistance is usually rational: people are asked to learn an unfamiliar process while carrying unchanged targets and closing the same period. Educate before training, because explaining what a process produces downstream prepares people for exceptions no procedure document covers. Develop instructors inside the functions, with allocated time and recognition and require line managers to work the new process themselves. A manager who tolerates the old method has announced that compliance is optional.

AI, at Tighter Tolerances

A license agreement changes nothing about how work is done. Data foundations become a precondition rather than a hygiene concern, because output reliability depends on sources that are current, documented and access controlled. Assistants placed over unchanged processes return marginal gains; value comes from restructuring work around what models handle reliably: extraction, classification and first pass review, while keeping human judgment where consequences are material. AI literacy belongs inside the business functions: what a model can and cannot establish, how to verify output and which data may not be submitted. A named individual stays answerable for any decision a model contributed to, and evaluation needs stated error tolerances by use case, with failure modes examined rather than aggregate scores.

Measure outcomes, not completion:

Cutover is a milestone, not a result. Outcome measures have to be defined before build and baselined before the change, or later claims of improvement cannot be substantiated: cycle time, manual touches per transaction, days to close, control exceptions. For AI, measure accepted output alongside assessed quality, because acceptance without verification is not a benefit.

In summary, a transformation concludes when accountability for process, data and outcomes sits with the business functions, and the technology organization operates the platform rather than carrying the change.