There’s a specific, predictable gap between a written continuity plan and a working one — and every organization that’s run its first real drill knows the flavor: the phone number that changed, the step everyone assumed someone else owned, the workaround that depends on a system that’s also down, the “critical contact list” nobody could reach because it lived on the unreachable server. None of these appear when you read the plan; all of them appear the first time you walk it. Testing is how you find them at rehearsal prices instead of performance prices. And crucially — this is a different test from restoring backups: that proves the data comes back; this proves the business keeps moving while it does. Here’s the practical version.

The tabletop: 90 minutes, highest yield in the discipline

The workhorse of continuity testing is embarrassingly low-tech: the right people around a table, a scenario, and the plan followed literally, out loud. “It’s Tuesday, 6:45 a.m. The building has no power and the utility says maybe tomorrow. Go.” Then walk it, step by step: Who declares this a continuity event — is that person here, and did they know that? How do staff find out, given that email might be part of the problem? Where does the front desk’s phone line ring? Which workaround does invoicing use, and does it depend on something else in the blast radius?

Two rules make tabletops work. Follow the actual document — the exercise is testing the plan, not the room’s collective cleverness; every moment someone improvises past a gap, that gap is the finding, written down. And let the facilitator inject complications midway (“the person who declares emergencies is on a plane”; “a client calls asking if you’ve been hacked”) — real events don’t arrive one clean problem at a time, and the second-order stumbles are where the richest findings live. Ninety minutes, three or four scenarios a year rotating through your realistic set — power and weather, building inaccessible, key system down, key person unavailable, cyber incident — and the plan gets more real every quarter.

Targeted drills: test one muscle for real

Tabletops test decisions; drills test mechanics — small, contained, real:

  • The contact-tree drill. Actually send the emergency notification, off-hours, unannounced (within reason). Measure who confirmed within an hour. Every unreachable name is a finding that costs nothing today.
  • The work-from-elsewhere drill. One team, one scheduled day, working entirely by the plan’s displacement scenario — remote access, phones forwarded, the works. This drill got accidentally validated everywhere in 2020; the finding now is whether it still works after three years of system changes.
  • The workaround drill. Run one critical function its manual way for two hours — take orders on the paper process, look up client info without the system. The people doing the job find the plan’s fiction fast, and the fix (“the paper form is missing half the fields we need”) is a ten-minute edit.
  • The unreachable-document drill. Simplest of all: from a phone in the parking lot, can the three key people actually open the plan, the runbook, and the contact list? The number of plans that fail this test is the industry’s quietest statistic.

Findings are the product — treat them like it

A test that ends with “good session” produced nothing. The output is a findings list — each gap with an owner and a date — folded back into the documents within the week, while memory is fresh. Keep the log across tests: findings-per-exercise should trend down, repeat findings should trend to zero, and both numbers belong in the annual review alongside the metrics leadership already sees. One cultural rule protects the whole enterprise: findings are wins, publicly. The person who exposes the broken step in a drill just saved the company a real-day disaster; if drills become blame theater, year two’s drills will find nothing — and not because nothing’s there.

The one-year cycle that keeps it honest

Sustainable beats ambitious: quarterly — one 90-minute tabletop, rotating scenarios · twice a year — one targeted drill (alternate the types) · annually — the full restore fire-drill run jointly with a continuity tabletop, so the technical recovery and the business-keeps-moving tracks finally rehearse together, which is how they’ll actually run · after any real event or near-miss — a lessons pass, because reality is the one drill you don’t schedule. Total cost: a few hours a quarter. It’s the cheapest insurance in this library, and the only kind that improves every time you pay the premium.

The bottom line

A continuity plan is a theory; a tested one is a capability. Tabletops find the decision gaps, drills find the mechanical ones, findings get owners and deadlines, and the cycle repeats until the plan describes something your team has actually done. The goal was never a cleaner binder — it’s that on the real bad morning, somewhere in the room, someone thinks: we’ve basically done this before.