Same-Day Calibration Go-Live: An Hour-by-Hour Plan

A clean calibration sheet can go live in one day if it passes three go/no-go gates. Here is the hour-by-hour plan, and the point at which to stop.

October 12, 2026 9 min read 1,710 words

"We’d love to get off the spreadsheet. We just can’t afford for it to turn into a project."

I hear some version of this from almost every quality engineer who owns a calibration workbook. It’s a reasonable thing to worry about. The workbook works, more or less. A careful person built it. It cost nothing, and it has held up through several audits. Taking a team away from the line and the lab for a three-month software rollout is hard to justify against a file that hasn’t failed you yet.

Two things are true here. A clean calibration sheet can be imported and running in a real system in one working day. A messy one can’t, and pushing it through anyway is how a calm migration becomes a bad audit. This article gives you the day hour by hour, with three go/no-go gates. Their job is to tell you, early and cheaply, which of those two sheets you have.

What "same day" honestly means

Be precise about the claim before you plan around it. Published implementation guidance in this field usually separates three things: one to two days to import, one to two days to configure, and two to four weeks to full operation, often with a parallel run alongside. That’s realistic. Teams have to get used to new habits, and the first round of reminders has to actually reach people.

"Same day" covers the first step, and part of the second. By the end of the day, every in-scope instrument is registered, with its interval, its last calibration, a due date calculated by the system, its location and its owner. Reminders are on. Overdue blocking is configured. The old sheet is frozen. Retraining every operator in one afternoon is not part of it. Neither is importing ten years of history, rewriting your calibration procedures, or getting through the next audit unaided.

That narrower goal is still worth having. From the next morning, overdue status no longer depends on someone remembering to open a file.

The day, hour by hour

07:30 — Freeze the sheet

Tell everyone who edits the workbook that it is read-only until the end of the day. Save a dated copy, and work from that copy only. If you can’t name everyone who edits the file, write that down. It’s a finding in its own right, and the main argument for doing this at all.

07:45 — Gate 1: triage (go / no-go)

Spend 45 minutes on the sheet before you import anything. Most "the import lost some gauges" stories come from Excel behaviour that people don’t see, not from the destination system. Check for these six things:

| Check | Why it matters | No-go if… |
|—|—|—|
| Merged cells (location, department blocks) | Only the top-left cell holds a value. The other rows in the block export blank | More than a handful of blocks, and nobody knows which location they belonged to |
| Blank rows inside the data | A blank row splits the range, so AutoFilter and many exports see only the first block | You can’t confirm the total row count against a physical count |
| Active filters, hidden rows, hidden sheets | Hidden instruments are skipped without warning | A hidden sheet holds a second population, such as "retired?" or "plant 2" |
| Status stored as formatting | Red, yellow and green are fills, not data. They vanish on export | There is no status column you can rebuild from the dates |
| Due dates from =A2+365 | This drifts across leap years: 1 March 2024 + 365 is 28 February 2025 | Typed-over and formula due dates are mixed, and nobody knows which is which |
| History in a year-per-column layout | "2022 / 2023 / 2024" columns don’t map onto a calibration record | You need the history to import today, rather than archiving it |

If none of these hit their no-go condition, carry on. If two or more do, stop here and book a cleanup day instead. That isn’t a failure. You found out at 08:30, not after the import, and the table above is now your cleanup checklist.

08:30 — Clean

Unmerge the merged cells and fill down. Delete the blank rows. Clear every filter, unhide everything, and reconcile the row count against what you know physically sits in the crib, the lab and the line. Replace any status that lives in formatting with an explicit value, or delete it, because the new system will calculate status from dates and intervals. Where a due date was typed over, keep the last calibration date and the interval, and let the system work out the next due date. That is the whole point.

10:00 — Map to the minimum column set

Map the six fields you need to go live first. Leave everything else for later.

Leave out the "Next Due" column on purpose. If you import the date the sheet calculated, you import whatever errors that formula had.

11:00 — Gate 2: pilot import

Import 10–20% of the rows first, covering every instrument type and every location. Most published migration guides recommend a pilot of this size, and it earns its place. If the pilot turns up mapping errors, such as a wrong interval unit or locations that don’t resolve, fix the mapping and run the pilot again. If it turns up data errors you can’t fix by lunchtime, it’s a no-go. Keep the triage notes and reschedule.

12:30 — Full import

Import the remaining rows, then check the total instrument count straight away. The number in the system should match the number you reconciled at 08:30 exactly, with no "roughly".

13:30 — Gate 3: validate a sample

This step turns a fast import into a defensible one. ISO/IEC 17025 clause 7.11 requires systems that record and retrieve data to be validated for functionality before they are introduced. The same discipline makes sense under ISO 9001 7.1.5, IATF 16949 7.1.5.2, AS9100 7.1.5 and ISO 13485 7.6. Each of those expects you to show that measuring equipment is calibrated at specified intervals, and that you can prove it.

Pick about 25 instruments: some due soon, some just calibrated, one of every type, and every leap-year case you found. Then work through each of them:

Compare the system’s due date with your own manual calculation from the last calibration date and the interval.
Check that the location and owner match what is physically true.
Open the attached certificate and confirm it belongs to that serial number.

If every due date matches and every certificate matches its serial, that’s a go. Record the sample, the results and who checked them. That record is your validation evidence. It’s also probably more evidence than the spreadsheet ever had, since the spreadsheet was almost certainly never validated itself.

15:00 — Switch on the controls

Turn on reminders. Configure overdue blocking: a hard block, where an overdue instrument can’t be checked out, or a soft warning. This is a setting per tenant, so decide it with production, not on your own. Apply your calibration procedure templates by instrument type, and override them for individual units where needed. Check-out and permanent allocation start recording possession from today, so the next "who had torque wrench TW-114 on Tuesday?" has an answer.

16:00 — Archive the sheet

Save the workbook read-only, give it a dated name, and keep it in a controlled location as the historical record. Announce the change in one sentence: "From today, calibration status lives in PrüfDesk. The sheet is archive only." If the old file stays editable, people will keep using it.

A worked example

Here is an illustrative case. The workbook covers 180 instruments. At triage, three problems show up. A blank row at line 112 means the filtered view only ever showed 111 instruments, so 69 were missing from it. Two merged location blocks span 30 rows between them, and the technician remembers which cell each block belonged to. Fourteen due dates use +365, and two of them have already drifted across the 2024 leap day.

Every one of those is fixable before 10:00, so it’s a go. The pilot of 30 instruments shows that one gauge type’s interval was stored in weeks and the rest in months. The mapping is fixed and the pilot run again by 11:40. In the 25-instrument sample, every due date matches the manual calculation. By 16:00 there are 180 instruments live, and the two drifted gauges now show their correct due dates.

The same team with an 1,800-row workbook, three plants, merged blocks nobody can account for and history spread across year columns would get a no-go at 08:30. That’s the right outcome for them.

"Isn’t one day too fast for a regulated environment?"

It’s too fast without the gates. With them, the speed comes from removing the waiting, not from skipping the checks. Each of the three gates produces evidence: a reconciled count, a clean pilot, and a validated sample. A three-month rollout without that evidence is weaker than one day with it.

History is the other common worry. A same-day go-live keeps it, in two places: the latest calibration record and certificate for each instrument go into the system, and everything older stays in the read-only archive. You can bring the deeper as-found/as-left history across later, in a planned way, without holding up the day the reminders start working.

What it costs

PrüfDesk prices by number of instruments, with unlimited users on every plan: Starter €49/month for up to 250 assets, Professional €129/month for up to 1,000, Enterprise €299/month with no asset limit. Because users are unlimited, every operator and technician can log in, which is the only way check-out records and possession history stay complete.

If you’d like a second pair of eyes on Gate 1 before you plan the day, talk to us about your import. Bring the workbook as it is. The triage table above is where we’d start too.