If OneDrive is where individual work lives, SharePoint is the company’s shared brain: the client files, the templates, the operations documents, the department libraries — plus, invisibly, the storage layer underneath every Team anyone has ever created. That shared nature changes the risk profile: a OneDrive problem hurts one person’s work; a SharePoint problem hurts everyone’s, at once. Here’s where SharePoint specifically goes wrong, and the protections that fit.
Risk one: permission sprawl — the shared brain forgets who it told
SharePoint’s permission model is powerful and, over years of real use, entropic. Sites beget subsites, libraries get “unique permissions” for one urgent exception, Teams creation spawns sites with their own membership, and individual files get shared past their container’s rules. The compound result: nobody can crisply answer who can see the client financials library — and that answer is what data classification promised to control. The discipline: keep permissions coarse — grant at the site level through groups, resist per-file and per-folder exceptions (each one is a future mystery) · structure sensitive content into its own sites, so “who can see this?” is answerable at the container level, with external sharing disabled where the tier demands it · and review access twice a year on the sensitive sites — membership drifts, departures leave residue, and the review is an hour that keeps the map true.
Risk two: deletion and corruption at shared scale
SharePoint’s native recovery is genuinely decent for the everyday: a two-stage recycle bin (93 days total) and versioning on by default, which quietly rescues more mistakes than any other feature — the overwritten proposal, the mangled spreadsheet, restored to Tuesday’s version in three clicks (teach your power users this; it’s the highest-value SharePoint trick that exists). What the native layer can’t do is the now-familiar list: recover what’s discovered after the window, reassemble a gutted site — structure, metadata, permissions — rather than items, survive a compromised admin, or roll a library back wholesale after ransomware syncs through it via somebody’s synced library. Independent backup covering SharePoint — full-fidelity, site-level restorable, tested against a real site occasionally — is the layer that answers all four, and it’s the same product already covering the rest of the tenant.
Risk three: site lifecycle — the brain grows but never prunes
Every project, initiative, and ad-hoc Team creates a site; almost nothing deletes one. Years in, the tenant carries hundreds: active ones, dormant ones, and — the risk — orphaned ones whose owners left, holding data nobody governs, sharing links nobody remembers, permissions nobody reviews. Stale sites aren’t clutter; they’re unwatched data with the lights off. The right-sized lifecycle: every site has a named owner (and loses “orphan” status the moment offboarding reassigns it) · an annual pass flags sites untouched in a year — archive (read-only, retained per policy) or delete, decided rather than deferred · and site creation itself gets the light governance treatment, because lifecycle problems are cheapest to prevent at birth.
Risk four: the platform settings underneath
SharePoint inherits the tenant’s posture, so the settings review is doing much of the work: sharing defaults, app consent (a granted Sites.Read.All is the whole brain, readable by API), MFA and sign-in policy on every account that can reach it, and audit logging retained long enough to answer “who accessed that library, and when?” — the question every data incident eventually asks.
The bottom line
SharePoint concentrates the company’s shared knowledge, so its risks are concentration risks: permissions that sprawl, deletions that hit everyone, sites that outlive their owners. Keep access coarse and reviewed, let versioning catch the small stuff and independent backup catch the big stuff, give sites a lifecycle with names attached, and let the tenant hardening underneath do its job. The shared brain deserves at least the care of any single desk — it is, after all, most of them combined.