As soon as a group brings together several entities, its digital transformation involves several decision-makers, several information systems and several budgets. Subsidiaries, business lines, countries and business divisions have, to varying degrees, their own teams, applications and priorities. The digital transformation of large enterprises inherits this organisation before it has chosen a single technology.
Scale therefore adds questions that a single-entity company does not encounter, or encounters in a simpler form: at what level each decision is taken, what is built once for everyone, what legacy systems allow, where to find room in an already full portfolio, who decides between two divisions. A group executive committee benefits from answering them before opening the workstreams. Left unanswered, each one settles itself, entity by entity.
At what level is the digital transformation of large enterprises decided?
A group has at least three levels of decision: the group, the entity (subsidiary, business line or country) and the project. The difficulty comes from decisions whose level is written down nowhere. The group may then reserve all choices for itself, and the entities wait for answers on subjects they know better than it does. Conversely, each entity can decide for itself, and the group ends up with several solutions for the same need.
We propose a criterion: a decision is taken at the lowest level that bears all its consequences. The choice of an exchange format, of customer master data or of a security rule constrains the other entities, so it goes up to the group. The order of a subsidiary’s workstreams or the ergonomics of a local tool commit only its own results, so they stay with it.
This division is written on one page, which sorts decisions by type (choice of a shared foundation, launch of a project above a set amount, derogation from an architecture rule) and indicates for each the body that decides and its response time. In a matrix organisation, a team can receive its priorities both from its entity and from a group functional division: the page says which prevails for the transformation’s work.
Should services be shared, or should each entity choose?
The question arises for each object, and the answer depends on where the value arises. Some objects are valuable because they are shared: the identity of customers and employees, reference data, security rules, technical foundations that are costly to build twice. Others are valuable for their closeness to the field: a market’s customer journey, a process that local regulation or the business makes different from one entity to another.
A few checks rule out the shared service that no entity wants:
- the need is genuinely the same from one entity to another, beyond the word that designates it;
- a team is named to operate the service over time, with funding that does not depend on a project’s budget;
- the entities’ change requests have a known channel and order of priority.
A single tool imposed on processes that differ can become covered in entity-specific configurations, to the point of wiping out the savings expected from pooling.
There is an intermediate path: shared rules and local implementations. The group sets the data formats, the interfaces and the security requirements, and each entity chooses the tool that complies with them. This path requires someone to write these rules, keep them up to date and respond quickly to projects that request a derogation, as described on our page on IT architecture.
Do legacy systems set the pace of the transformation?
Legacy systems here means the old core applications (management, production, billing) that new services need in order to read data or record a transaction. In a group, acquisitions may have left several of them for the same function. A digital journey shared by three subsidiaries must then communicate with three systems, each with its own maintenance team and release calendar.
The committee can ask, for each objective of the transformation, for the list of systems it depends on and, for each one, what it can do: expose its data through an interface, absorb a change within a known time. The age of the system says little on this point. The useful criteria are the availability of skills, the end-of-support date announced by the vendor and the cost of each change, which the notion of technical debt helps to express.
For each system, the decision can be reduced to four families: keep it as it is, open it up through interfaces, isolate it behind an intermediate layer, replace it. Our page on application modernisation details the strategies. The order of the work, for its part, is decided at group level, because a transformation calendar written without that of the legacy systems can send several projects, at the same time, to the same maintenance team.
Where do you find capacity when the portfolio is already full?
A group can have as many project portfolios as it has entities, plus its own. Each can be balanced taken on its own. The transformation adds initiatives to them that draw on resources no one sees in full: the architects and the data or security experts that entities share, and the business teams, called upon for workshops, acceptance testing and training by projects that do not know about each other.
The committee’s question therefore concerns what is shared. We propose consolidating a view limited to two objects: the scarce skills common to several entities, and the business populations affected by several projects over the same period. The entities’ portfolios remain their own.
We add one rule: a group initiative arrives in an entity with an indication of what it postpones or replaces there, validated by the entity’s head. Without this indication, the initiative is added to an already full workload plan, and the surplus is paid for in delays. Reading capacity by skill, choosing priority criteria and the functioning of the arbitration committee are detailed on our page on project portfolio management.
Which arbiter for two divisions that are each right?
The disagreements of a group transformation can set equally legitimate objectives against each other. Take three positions. A subsidiary’s sales division defends its launch date. The group’s security division defends a shared rule. The finance division defends the convergence of tools. Each is right within its mandate, and none has authority over the other two.
Arbitration first requires an arbiter placed above the parties: the sponsor of the transformation on the group executive committee, or the chief executive when the sponsor heads one of the divisions concerned. It then requires an order of priority between objectives, written before the first conflict: for example compliance, then convergence, then the local calendar, each group setting its own. Finally, it requires a format: a page that sets out the positions, reviewed by those who hold them, the options, what each costs each entity and the expected response date.
Preparing these requests, keeping track of dependencies between entities and following up the implementation of decisions is the work of programme management. When these mechanisms exist and the expected effects do not arrive, the question changes: our article on diagnosing a transformation that is stalling starts from the signals observed to trace back to their cause.
These questions can be tried out on an initiative already under way that affects at least two entities. Those that remain unanswered provide the agenda for the next group executive committee.