At each of its stages, an information system project leaves documents in which the preparation of change can or cannot be read: the scoping note, the acceptance testing defect log, the deployment plan, the arrangements for the first days. Change management in an IT project, that is, preparing users to work with the new system, can therefore be examined in these documents, without waiting for the usage figures that follow go-live.
The signs are grouped according to when they appear. At scoping, users are described without having been met. In acceptance testing, impacts emerge that nobody had noted. As deployment approaches, the announced change agents have neither a name nor time. At go-live, teams are left alone with the tool. Each sign admits several readings, to be checked before acting, and what remains possible depends on the time left.
At scoping, users described without having been met
The first sign is read in the scoping note. Users appear there under a collective name (“the case handlers”, “the network”), with no headcount, no site, no management. No one from the business took part in the workshops. For change, the schedule carries only a “training” line placed between acceptance testing and go-live, and the budget does not identify this item separately. Nobody has asked the human resources department and the lawyers whether the project calls for informing or consulting staff representatives.
One possible reading: the project was scoped as the delivery of a tool, and adoption is taken for granted. The sign may also reflect a simple order of work, with the team planning to come to it once the solution has been chosen. The check consists of asking who, in the project organisation, is in charge of preparing users. Usage entrusts this task to the MOA, according to the split recalled in our entry MOA, MOE, AMOA; but a person still has to hold it.
At this stage, we propose a light response: a list of the populations affected and an initial impact analysis, carried out with people on the ground before the design freezes choices. These are the first two steps of the method set out in our article on the change management plan.
Acceptance testing reveals impacts that nobody had noted
Acceptance testing may be the first time users handle the tool. Some of the gaps they report are neither defects nor forgotten requirements: the application does what the specifications ask, and it is the work that changes. An approval moves from one department to another. A check that an employee used to carry out from memory is entrusted to a business rule.
These gaps can be recognised by the fate the defect log reserves for them: classified as “compliant with specifications” or “out of scope”, then closed. They may indicate that the impact analysis did not take place, or that it followed the tool’s functions without going through the business units. They may also cover a business case genuinely missing from the specifications, which then falls under a change request, according to the distinction drawn in our article on user acceptance testing.
We propose starting with a sort. Gaps closed as compliant are reviewed with a business manager, who says for each one whether the work changes, and for whom. Those that affect an entire population feed an impact matrix, even a late one. The heaviest are escalated to the steering committee, which may decide on a phased deployment or a revised date. The others provide the content of the forthcoming sessions: testing has shown what will need explaining.
Change agents with neither name nor time as deployment approaches
The deployment plan mentions a “network of correspondents” or “key users”. The sign is read in what is missing around these words: no named list, no breakdown by site, no time freed up by management. A variant: the list exists, but brings together the people who carried the acceptance testing, without relief from their usual tasks.
At this stage too, front-line managers receive the announcement at the same time as their teams. Sessions are scheduled, and their content follows the application’s menus instead of work situations.
These signs may reflect a change left to the project team alone, without commitment from the user departments. They may also be due to workload: when other workstreams already occupy the same people, management may no longer have change agents to release. If the sign recurs from one project to the next in the same organisation, the lead to check is the common project framework, which may ask nothing on this point at its milestones. A project management office can then include, among the documents required at each phase review, the list of populations affected and the list of change agents, which makes their absence visible in committee; designing this framework is the subject of our page PMO and project governance.
As long as the date allows, we propose that the business departments appoint the change agents and reserve time for them, in writing, and that managers be informed before their teams. The choice of audiences, timing and format of sessions is covered in our guide on user training. When the date no longer allows it, we propose two options to the steering committee: postpone go-live, or deploy first on a site where change agents exist.
IT project change management caught up after a go-live without support
The last sign is observed on go-live day and in the weeks that follow. The usual help desk receives requests with no reinforcement or instructions. Nobody from the project is present in the teams. Usage questions and tool defects arrive through the same channel, unsorted. The old tool remains open, with no shutdown date, and the project team has already been reassigned.
Such a go-live may stem from the project’s terms of reference, whose end was set at the day of go-live: nobody was then asked to follow up the first weeks. It may also come from a late trade-off, which removed the planned arrangements from the budget. The minutes of the committee that approved go-live can help decide between these two readings.
A catch-up can still be organised: change agents appointed after the fact, a drop-in support desk, a record of what the teams do instead of what was planned, a shutdown date for the old tool announced by management. When workarounds persist despite this support, the subject changes in nature, and our article on resistance to change offers a diagnosis by causes. Human resources tools add their own constraints, described in our article on HRIS change management.
A single sign admits several explanations. When the scoping note, the testing log and the deployment plan are all silent about users, a common explanation becomes plausible, also to be checked: adoption was entrusted to nobody, or was entrusted without resources. Recovery then goes through the complete set of arrangements (impact analysis, stakeholder mobilisation, communication, training, adoption measurement) presented on our service page on change management. Signs of drift in schedule, cost or scope call for another reading, offered in our article why IT projects fail.
The review comes down to four documents, from the scoping note to the go-live arrangements, to be reopened in the order of the project. The first one that says nothing about users indicates the point to take up again.