A café owner I know closed for three days after a pipe burst above the kitchen. She did almost everything right, and the one thing she did wrong — leaving Google Maps saying OPEN while the door was locked — generated more angry one-star reviews than the closure itself.
Here is the reframe: a closure is two documents at once. A customer-facing notice (what's closed, until when, how to reach you, how you'll confirm the reopen) and an operational record (who was told, when, through which channel — the folder your landlord, insurer, and licensor will ask about later). Most businesses write neither; a few write the first and skip the second; almost nobody thinks about the second until they need it.
The seven lines every closure notice contains
- What is closed — the site, the door, the service. Not "we" in the abstract.
- Dates and times, both ends. Closed from 2pm Tuesday, planned reopen 7am Friday. A planned reopen you later move costs nothing; an "until further notice" vacuum costs you the customer. If the reopen date genuinely can't be estimated, give the date of the next update instead.
- The reason, in one honest sentence. "Burst pipe in the kitchen" beats "unforeseen circumstances."
- What still works. Online orders? Email answered? Gift cards? Half of lost closure revenue is customers who assumed everything stopped.
- The contact channel that WILL be answered — one named owner, stated response time.
- The make-good for scheduled customers. Every booking, delivery, or pickup in the window gets contacted individually. This line commits you to it.
- How you'll confirm the reopen. This sentence is why your reopening post gets read.
The channel order (and the slot everyone forgets)
The order follows how customers discover you're closed: at the door, on Maps, or by phone.
- The door/window sign — printed, dated, ten minutes.
- Website banner + homepage line.
- Google Business Profile hours. Customers check Maps before they drive. An OPEN pin on a locked door is the angriest review you'll ever read — and this is the single most-forgotten channel on the list.
- Voicemail greeting — re-recorded, same seven lines.
- Email autoresponder.
- Social: one pinned post. Post once, pin it, stop. Ten closure-day posts read as chaos; one pinned post reads as control.
- Scheduled customers, individually — the difference between "they closed" and "they called me."
Match the response to the hours, not the drama
- Under 24 hours: door sign, voicemail, Google hours, individual calls to anyone scheduled in the window.
- 1–7 days: the full notice on every channel, autoresponder on, and the landlord/insurer notification if the lease or policy requires it (many want notice within 24–48 hours — check your lease before the closure, in peacetime).
- Over 7 days: everything above, plus a stated weekly update cadence, written staff scheduling, and suppliers paused in writing so stock doesn't arrive at a dark store.
The five closure traps
- "Until further notice" with no date of any kind — reads as permanent or evasive.
- Silent voicemail — the default carrier greeting over a locked door reads as sloppiness.
- Maps says OPEN while the door is locked — the classic driver of furious reviews.
- Announcing the reopen before it's verified — the same-day "sorry, still closed" retraction is the second incident.
- Letting booked customers learn from the sign — the Thursday 2pm booking gets a personal message, not a drive past a dark storefront.
The verified-reopen checklist (run it BEFORE the reopening post)
Utilities holding (fridge back under 5°C, not "the compressor hums"); safety-critical equipment tested end-to-end (a POS transaction that actually settles); first-shift staff confirmed by name; stock checked and emergency orders placed; contact channels reverted before the reopen post goes up (voicemail, autoresponder, banner, Google hours); and the make-good ready. Then the reopening notice: verified reopen, thanks, what changed for the better, the make-good, and the closure notice kept visible for 48 hours with a redirect line.
The worked example: 14-seat café, 3-day closure, 62 bookings
Tuesday 9:12am, the pipe lets go. Water main killed by 9:20. Door sign printed and up by 9:40 — closed from 2pm, planned reopen Friday 7am, one honest reason, what still works, the answered email, the make-good promise. Google hours closed through Thursday by 10:15, voicemail 10:30, autoresponder 10:40, one pinned post 10:50. Then the bookings list from the till, worked top to bottom: 62 bookings in the window, all 62 contacted personally — 41 moved, 17 taken as credits, 4 cancelled. Landlord emailed 11:15 (the lease wanted "prompt" notice), insurer line logged with a timestamp, and the repair invoice + plumber's report + the closure notice itself went into one folder — which is exactly what the business-interruption claim later asks for.
Friday 6:40am: the verified-reopen checklist runs — fridge at 3.8°C holding, POS transaction end-to-end, alarm armed and disarmed, nine staff confirmed by name, channels reverted — and only then the 8:05am reopening post with the "first coffee's on us" make-good. Four of the 17 no-shows redeem credits that weekend. The metric the owner now tracks monthly: door-sign-to-all-channels was 1h53m that Tuesday. Target: under 60 minutes. The thing that ate the margin was the Google hours update nobody remembered until a customer called about it.
The four numbers to review after every closure
Hours from door sign to last channel updated (under 60 minutes); scheduled-customer contact rate (100%, every time); verify-before-announce (100% of reopens verified first, zero retraction posts); and day-1 channel accuracy — voicemail, Google hours, and banner all saying the same thing (3 of 3; one stale channel is the classic half-closure).
The full page — the seven lines, the channel order, the duration tiers, the traps, the verified-reopen checklist, and the copy-paste one-page notice template — lives here: Temporary closure notice template for small businesses. Pair it with the outage email templates (the online-outage half), the business continuity plan template (the decision to close, made in peacetime), and the internet outage runbook (when it's the connection, not the doors).
If you want the whole continuity stack on paper, the HIVE80lab Ops Starter Kit covers incident response for small teams ($14), and the Ops Mega Bundle collects all five ops kits in one download ($49).
Top comments (0)