What does drawing review actually cover?

Confirming, before release, that the drawing matches design intent and that the part can be made, bought, and inspected. It is often mistaken for a dimensional proofread. In practice it also covers manufacturability, procurement, measurability, and conformance to internal standards, customer requirements, and regulation. Done properly it uses a fixed set of checks so two reviewers reach the same conclusion, and it leaves a record.

Who should review a drawing, and when?

Immediately before release, by someone other than the person who drew it. Self-checking alone leaves misses. The usual split is three stages — designer self-check, independent review, and approval — each looking at something different. The self-check closes out dimensional consistency; the independent review asks whether it can actually be made and bought.

How many items should a check list have?

Too many and it becomes a ritual; too few and things slip. Around 30 fits on one page and runs in about ten minutes. What matters is not the count but whether your own past failures are on the list. Do not adopt someone else's list wholesale — add items from your own corrective action reports.

Should drawing reviews be recorded?

Yes. Customer quality audits and ISO 9001 assessments can ask for evidence that design verification took place. And when a problem surfaces later, knowing how far the review went shortens the investigation. A form with reviewer, date, finding, disposition, and approval doubles as that evidence.

A 30-Point Drawing Check List: Stop Misses With Process

Hello, this is Ryuta Hamamoto from TIMEWELL.

Whenever drawing review comes up, I hear the same sentence. "Our senior guy spots it in five minutes."

I believe it. The problem is that there is one of him, and nobody has written down what he is looking at. A junior engineer releases a drawing three times, gets pulled up on the same thing three times, and then misses it again on the next part. That is not an attention problem. It is the absence of a method.

This piece lays out 30 checks in five groups: dimensions, tolerance, manufacturability, procurement, and compliance and records. There is a sheet at the end you can fill in.

The short version:

Drawing review is not proofreading dimensions. It asks whether the part can be made, bought, and inspected
The designer's own check is not enough. Separate self-check, independent review, and approval
Around 30 items fits on a page and runs in ten minutes
Do not adopt someone else's list. Add items from your own corrective action reports
Records become audit evidence, and they speed up root cause work later

There are five groups, not one

Drawing review is usually pictured as hunting for missing dimensions and typos. That is one group. There are four more behind it.

Group 1. Dimensions and consistency. Does it hold together as a drawing? Do dimensions close, are any duplicated, do the views agree?

Group 2. Tolerance and precision. Is that precision actually needed? How will it be measured?

Group 3. Manufacturability. Can that geometry be produced on your equipment, or your supplier's?

Group 4. Procurement. Are the material and bought-out parts specified in a form you can get when you need them?

Group 5. Compliance and records. Internal standards, customer requirements, regulation. And does a record survive?

A review that only covers group 1 is a check of the drawing, not a check of the job. You can release a drawing that is flawless on paper but impossible to machine, specifies material nobody stocks, or carries a tolerance nobody can measure.

The 30

Add and subtract to suit your shop.

Dimensions and consistency (1–8)

Do dimensions close? Does anything fail to add up in a chain?
Is the same dimension given in more than one place? (Duplicates are how revisions get missed)
Do the views agree — front, plan, and side consistent with each other?
Are section cut lines and arrow directions correct?
Are datums stated? Is it unambiguous where a measurement starts?
Does the stated scale match what is drawn?
Do quantity, material, and part name agree between the drawing and the BOM?
Are the drawing number and revision correct? Is an old revision in circulation?

Tolerance and precision (9–14)

Are tolerances specified? Where general tolerance is relied on, is that appropriate?
Is the tolerance genuinely required? Any precision specified out of habit?
Is the geometric tolerancing sensible given the function?
Does the surface finish callout agree with the intended process?
How will that tolerance be measured? Does the gauge exist, in house or at the supplier?
Is inspection frequency defined — 100% or sampling?

Manufacturability (15–21)

Can the tool reach? Any unreasonable deep holes, tight corners, or back-side features?
Do corner radii match available tool diameters? Has anyone called out R0?
Will thin sections deflect during machining?
Can the part be held? Is there a surface for the chuck or fixture?
Is the number of setups reasonable? Is five-sided machining being demanded?
Where welding is specified, can the electrode or torch physically get in?
Is distortion and dimensional change from heat treatment or plating accounted for?

Procurement (22–26)

Is the specified material available in standard sizes? Has the lead time on special material been checked?
Are plate thickness and bar diameter standard stock sizes?
Is the part number for a bought-out item current? Has it been discontinued?
Where material is customer-supplied, are the handover terms on the drawing or in a note?
Do quantity and lot size suit how the material is cut?

Compliance and records (27–30)

Does it conform to internal and customer-specified standards?
Does it use material subject to regulation — restricted substances, export-controlled items?
Are reviewer, date, finding, and disposition recorded?
Was it released after approval?

Three stages, three viewpoints

One person running 30 checks in one pass usually turns into a ritual. Split it into stages and change what each stage looks for.

Stage 1: designer self-check. Mostly groups 1 and 2 — dimensional consistency, tolerance callouts, agreement with the BOM. You can close these yourself, and passing them on unresolved wastes the reviewer's time.

Stage 2: independent review. Mostly groups 3 and 4. Can it be made, can it be bought? These are the parts the designer cannot see. Ideally the reviewer knows production or purchasing; another designer will do if that is not possible. The value is that it is not the author.

Stage 3: approval. Group 5, plus the overall call, which brings cost and schedule into it.

Put all three on one form, each with a name and a date, and you can tell afterwards how far the review actually got.

Do not use someone else's list

This is the part I most want to land.

There are drawing check lists all over the internet, including the 30 above. Used as-is, they do not work. The reason is simple: they do not contain the failures your shop has actually had.

There is a reliable way to build one that does. Add items from your corrective action reports. A customer reports a defect, or you scrap and remake something internally. Each time, ask whether the review could have caught it. If it could have, add a line.

Keep that up for two years and the list becomes yours. If ten of the thirty are specific to your shop, the list is now an asset. Nobody else having the same document matters.

Conversely, a list that has not grown in years is probably not being used. Static is the symptom.

A veteran's eye can be made explicit

Back to the opening line.

The reason the veteran spots it in five minutes is that he has been burned by a similar drawing before. That hole cannot be reached. That material has an unpredictable lead time. That tolerance has no measurement method. All of it is accumulated experience.

And that experience lives in past drawings and the findings raised against them. If reviews were recorded, you have a database of which geometry drew which objection. If they were not, the whole thing leaves when he does.

What our enterprise AI, ZEROCK, does here is make that accumulation searchable: find past drawings similar to the one on screen and surface the findings they attracted. The AI is not reviewing the drawing. It is standing in for the part of the veteran's memory that can be indexed.

Same caveat as ever: you need to be recording first. With no records there is nothing to retrieve. Start by running the sheet and letting the records pile up.

There is more on the succession angle in skills transfer in manufacturing and on the filing side in drawing management, PLM and PDM.

The sheet

There is a drawing check sheet covering these 30 items with columns for reviewer, date, finding, disposition, and approval. The three stages have separate blocks, so it drops straight into an existing workflow, and it presents as design verification evidence in a quality audit.

It is built on the assumption that you will add your own rows. The download link is at the end of this article.

In summary

Drawing review is not a dimensional proofread. It asks whether the part can be made, bought, and inspected
Five groups, 30 items: dimensions, tolerance, manufacturability, procurement, compliance and records
Split self-check, independent review, and approval, and change the viewpoint at each
Do not adopt someone else's list. Add your own items from corrective action reports
A list that has not grown in years is not being used
Records give you audit evidence, and they turn a veteran's eye into something explicit

If you want to standardise drawing review, or make past findings retrievable, get in touch.