The second-roster problem
Almost every digital business card deployment starts the same way: someone exports employee data from the HR system, tidies it in a spreadsheet, and imports it. It works. Everyone has a card. The project is declared done.
Then the organisation does what organisations do. Someone is hired and does not appear. Someone is promoted and their card says the old title. Someone changes their mobile number and the card keeps the old one. Someone leaves and their card stays live, because taking it down was not on anyone's offboarding checklist. Three months later the card data and the directory disagree in a few dozen places, and nobody knows which ones.
The root cause is not carelessness. It is that a manual import creates a second roster with no owner and no update mechanism. The fix is not more discipline — it is removing the second roster. When cards are generated from the directory, there is exactly one place employee data lives and exactly one team accountable for it, and that team is already doing the work.
What "sync" should mean, and what it often means instead
Vendors use "directory sync" for a wide range of behaviours, and the differences matter a great deal at renewal time.
- A one-time import. Data is read once at setup. Everything after that is manual. Frequently marketed as directory sync; it is directory import.
- A scheduled re-import. The whole directory is re-read on a cadence. Better, but changes lag by up to the interval, and departures stay live until the next run.
- Event-driven provisioning. The identity provider pushes changes as they happen, over SCIM. Joiners, title changes, and — critically — deactivations propagate without waiting for a batch window.
- Two-way sync. Changes made in the card platform write back to the directory. Rarely what you want: it puts a marketing tool in the position of editing HR records, and the governance conversation that follows is not worth the convenience.
When a CSV really is the right answer
Directory integration is not universally correct. If your HR system has no SCIM support, if the organisation is small enough that changes are rare and visible, or if the card programme covers a fixed population like a conference delegation, a mapped CSV import is simpler, cheaper, and entirely adequate.
The honest test is whether anyone owns re-running it. A CSV import with a named owner and a calendar reminder is a real process. A CSV import that happened once during rollout is a decaying dataset with a plausible-looking start date. If nobody owns the refresh, connect the directory instead — that is the whole argument.