How to Mark Up a Drawing for the Next Tech

Why this matters

A drawing marked up badly is worse than one not marked at all. Circles with no meaning, a highlighted line nobody can date, an X that could mean "removed" or "problem here" or "start here" depending on who drew it. The next tech has to decide whether to trust marks of unknown authorship, and the safe assumption is not to, which throws away the work.

A markup is a document you are publishing. It needs an author, a date, a stated scope, and a mark vocabulary that means the same thing to a stranger. That is a specification, and it fits on the sheet.

The one rule that outranks the others

A markup never erases, covers, or corrects the original line. Draw beside it, strike through it with the original still readable, note next to it. Never white it out, never redraw over it, never delete it from a digital copy.

The reason is diagnostic, not archival. The difference between what was designed and what was installed is frequently the fault. A tech who can see both the original routing and your correction knows somebody changed something and can ask why. A tech who sees only your corrected version sees a clean drawing that appears to describe an original design, and will never think to ask.

The practical corollary: mark a copy, never the only copy. If the site holds one original print, photograph it whole before you touch it and mark a print or a digital copy. A digital markup layer that can be toggled off is the ideal, because the base sheet stays untouched by construction.

The header block every marked sheet carries

Put this in a clear corner before you make a single mark. Without it, the marks are anonymous and get discounted.

The scope field is what separates a usable markup from a decorative one. "Verified low-voltage control interconnect only, at the unit; did not open the branch panel and did not trace line-voltage feeders" tells the next tech exactly how far your credibility extends. Without it, a reader either over-trusts the sheet or discounts all of it.

The five mark classes

Colors are not portable. A shop's color convention lives in that shop, and a photocopy, a grayscale print, or a colorblind reader loses it entirely. Encode meaning in the annotation, and let color be a convenience on top. Every mark gets a short word or a symbol you define in a legend on the sheet.

Two of these are the ones that get skipped, and they carry most of the value.

"Verified as drawn" feels like it adds nothing, since the sheet already said so. It adds everything, because the next tech's real question is not what the drawing claims but which parts of it somebody has confirmed. A sheet on which some elements carry a check and initials and the rest carry nothing is enormously more useful than a clean sheet, and it took no extra visits to produce.

"Unverified" is the honest one. A mark that says you could not reach the device, or the tenant would not release the space, or you ran out of time, is the difference between a gap the next tech plans for and a gap they walk into. An unmarked element reads as untouched or as fine, depending on the reader's mood, which is the same as carrying no information at all.

The procedure

1. Photograph the base sheet whole, before marking. Including its title block and revision. If your markup is later questioned, this is what it was marked against.

2. Fill the header block first. Scope and method get written before you have found anything, which forces you to decide what you are actually going to verify rather than describing whatever you happened to notice.

3. Draw the legend on the sheet. Five lines, defining your five marks. It costs almost nothing and it makes the sheet self-contained, which matters because the sheet will be photographed and forwarded without you.

4. Walk the drawing element by element, in the drawing's own order. Not in the order you encounter things in the field. Following the drawing's sequence is what surfaces the elements that are missing from the field, and missing elements are invisible if you walk the field instead.

5. Prove dead before you verify anything that requires touching a conductor. Where verification means landing a meter inside an enclosure, open the disconnecting means, apply your own lock and tag, and test your meter on a known live source, then the conductors, then the meter again (NFPA 70E-2021, 120.5). The general-industry duty to de-energize and lock or tag before work on or near exposed energized parts is 29 CFR 1910.333(b)(2); the construction counterpart is 29 CFR 1926.417. If the verification cannot be done de-energized on this visit, that element gets an "unverified" mark and a reason, not a guess.

6. Mark stored-energy conditions as you find them. Where the drawing shows a device that holds pressure, a charged accumulator, a spring-loaded element or a suspended load, note the isolation and relief points beside it. That is written-down work under 29 CFR 1910.147, and a drawing is the natural place for the next person to meet it.

7. Write the "how you know" beside anything you added or corrected. Traced, measured, read from a label, or told by the site. A correction whose basis is "the maintenance guy said so" is still worth having, and it is worth much less than a traced one, so say which it is.

8. Do the cold read before you leave the property. Look at your own sheet as if you had never been there and ask what you would do first. If the answer requires knowing something only you know, write that thing on the sheet now, while the equipment is still open and you can check.

9. Get the marked sheet into the site file and onto the ticket the same day. A markup in a truck is not a document. A photograph of the marked sheet, flat, square-on, in good light, with the header block legible, goes into the site record before the day closes.

The filled-in artifact

A partial control sheet for a light commercial system, showing 9 devices in the low-voltage control loop. One visit, markup time about 25 minutes inside a 45 minute service call.

Header block as written:

• Base sheet: control interconnect sheet, revision B, dated on the sheet.
• Marked by: full name, mobile number.
• Date: date of visit.
• Scope verified: low-voltage control interconnect at the unit and at the wall control only. Branch panel not opened. Line-voltage feeders not traced.
• Method: meter continuity with the unit isolated, plus visual confirmation of device tags.

What the marks said, by class:

• Verified as drawn: 4 devices. Check mark and initials at each. All four matched the sheet in both location and terminal assignment.
• Deviation: 2 devices. One shown on the sheet in a location it is not in, with the actual location drawn beside it. One landed on a different terminal pair than drawn, with the actual pair written next to the drawn one and the drawn one struck through but readable.
• Abandoned: 1 device. Present, wired, out of service. Circled, labeled abandoned, with a note that its conductors remain landed at the control and are not isolated - which is the kind of detail that stops the next tech from assuming a dead leg is dead.
• Unverified: 2 devices. Both in a tenant space the building would not open during occupied hours. Question mark, and the words "tenant space, access only during scheduled shutdown."
• Additions: 2 devices present in the field and absent from the sheet, drawn in and labeled field, with a note that both were traced with a meter rather than taken from a label.

The arithmetic on the sheet closes, and it is worth checking out loud. The sheet showed 9 devices: 4 verified, 2 deviations, 1 abandoned, 2 unverified, which is 9. Of those 9, the abandoned one is out of service, leaving 8 of the drawn devices in service. Add the 2 field additions and 10 devices are in service against a sheet that describes 9, one of which is not running. A tech who counts devices in the field and counts symbols on the sheet and gets 10 against 9 has found something before touching a meter, and that count is the cheapest diagnostic move available on any drawing.

What the next visit did with it. The following call was a no-response fault. The tech opened with the two field additions, on the reasoning that anything not on the drawing was added by somebody solving a problem and is therefore both more recent and less likely to have been installed to the same standard. The fault was in one of them. That is not a rule that always holds, and it is a defensible first cut on a sheet whose header told him the low-voltage loop had been verified and the branch panel had not.

The failure mode, concretely. Had the same visit produced an unlabeled markup - three circles and a highlighted line, no header, no legend, no date - the next tech faces marks that might describe this revision or an earlier one, might mean "changed" or "check this", and might have been made by somebody who never opened the unit. The rational response is to ignore them and start over, which converts 25 minutes of good field work into zero. Anonymous marks do not degrade gracefully; they go straight to worthless.

References

• 29 CFR 1910.333(b)(2), general industry, and 29 CFR 1926.417, construction, on de-energizing and lockout or tagging of circuits before work on or near exposed energized parts
• NFPA 70E-2021, 120.5, on the live-dead-live proving sequence
• 29 CFR 1910.147, control of hazardous energy, for isolation and relief of stored mechanical, hydraulic and pneumatic energy
• See related: The As-Found Drawing and Why You Make One; The Site Documentation File SOP