A manufacturing workshop employing thirty staff had grown by acquisition as well as growth, and its digital estate told the story: an ageing on-site server holding the original file share, two cloud apps adopted by different departments in different years, and job documentation scattered across supervisors' personal drives. Finding a drawing or a costing sheet depended on knowing who had worked on the job. The server's warranty had expired, its backup had never been tested, and the owner's description of the arrangement — "held together by habit" — was accurate.
The problem
The risks were concrete. A single server failure would have taken the business's documentation with it, and the backup's recoverability had never been verified. Quoting was slow because historical job files took an average of twenty minutes to locate. Site supervisors could not access documents off-site, so printed copies circulated and went stale. And every new starter inherited the confusion, multiplying it rather than resolving it.
The obstacles
The owner's central fear was downtime: production stops for nobody, and a migration that interrupted the workshop was unacceptable. The second obstacle was classification — eleven years of files, some commercially sensitive (customer drawings, costing formulas), had to be sorted by sensitivity before any move, and no one had time. The third was habit: supervisors had built personal folder systems they relied on, and would abandon any structure that felt imposed.
What we did
We began with an inventory and sensitivity classification of the estate — what existed, where, who owned it, and what could be archived or discarded under the business's own retention rules. The target design was a single cloud document platform with a permissions model mirroring the workshop's real structure: production, estimating, finance and management, each seeing what they needed and nothing more. Migration ran in four phases over six weeks, department by department, with each phase rehearsed on a copy first and a rollback available; cutover happened outside production hours, and the old server remained read-only for a full stabilisation month before decommissioning. Supervisor folder habits were not fought but mapped — their structures were rebuilt as views within the shared model, so nothing familiar was lost.
The result
The migration completed with zero unplanned downtime, and the tested backup regime replaced the untested one. Historical job files now retrieve in seconds through a search that understands content, not just filenames; supervisors work from any site on current documents; and new starters receive a documented, permissioned environment on day one. The expired-warranty server and its unverified backup are gone, along with the quiet anxiety attached to them. The owner's verdict: the business now has one version of the truth, and it is backed up properly.
Engagement type: Defined Project (inventory, classification, migration, permissions), with an optional maintenance retainer.
The non-technical half that decided success
Technically, this migration was unremarkable — and that was the achievement. What determined the outcome were three non-technical decisions. First, the classification came before the cloud: eleven years of files were sorted by sensitivity and value before any platform was chosen, which prevented the most common failure of SME migrations — moving everything, including what should have been retired, at maximum cost. Second, supervisor habits were treated as requirements: the personal folder structures that felt chaotic from outside actually encoded useful working logic, so they were rebuilt as saved views in the shared system rather than abolished. Adoption followed almost without argument. Third, downtime was refused as an option from day one, which forced the phased, rehearsed, rollback-ready design — and that discipline is why a workshop that cannot stop moving never noticed the change.
For any business carrying an ageing server, the transferable point is this: the migration is a project about people and knowledge first, storage second. The inventory and classification exercise that preceded the technical move took two weeks, cost a fraction of the migration, and would have been worthwhile even if the business had decided not to move at all — because for the first time, it knew what it had.
An estate "held together by habit"?
An inventory and classification exercise is the safe first move — and it is quick.