Here is the problem in one sentence. Your factory is not in your head.

They have three things: the physical sample, the tech pack, and your words.

If your words are vague, the gap gets filled with a guess, and the factory’s guess optimizes for what is fast on their machines, not for what you pictured.

A single sampling round is a real cost.

It is 10-20 days of calendar time, a courier bill in both directions, and the sewing time to make the corrected sample.

Multiply that by every comment the factory has to interpret, and vague feedback quietly becomes the biggest line item in early development.

This guide fixes that.

Two named devices do most of the work: The three-line comment, the priority ladder, and the checklist wire them into a repeatable process.

Nothing here needs software.

A tape measure, your tech pack, and a phone camera cover it.

01. Why “make it better” gets you nowhere:

Vague feedback fails in three specific ways.

Name them, and you’ll catch yourself before you send them.

It’s subjective:

“Looks cheap.” “Feels off.” “Not quite there.” These describe your feeling, not the garment.

The factory can’t sew a feeling.

Every subjective note is a comment the factory has to translate, and translation is where guesses enter.

It’s unlocated:

“The sleeves are wrong.”

Which measurement?

Which sleeve?

Cuff, bicep, or length?

An unlocated comment forces the factory to inspect the whole area and pick the change they think you meant.

Locate every issue to a single point, and the guessing stops.

It’s unranked:

You send twelve comments.

The factory has no idea which one blocks approval and which is a nice-to-have.

So they either treat all twelve as mandatory, slow, and expensive, or they stall and email you to ask.

Both cost a round.

Every vague comment is, at minimum, a round-trip email. At worst, it’s a wasted sample round.

This is not a soft cost.

Samples built to the wrong read need remaking, which burns material, labor, and calendar time before you ever reach production.

The founders who move fastest through development are not the ones with the best taste; they’re the ones whose feedback needs no interpretation.

For the full picture of how rounds stack up in a development timeline, see our guide to the sampling process, and for how these rounds map to your cash position, see the deposit-to-delivery gap.

Why the factory guesses the cheap way:

When a factory fills a gap in your feedback, it doesn’t guess randomly; it guesses toward whatever is fastest and safest on its own floor.

That isn’t bad faith; it’s how a production line survives.

A vague note leaves room for the low-effort read, and the low-effort read is the one you’ll get back.

Precise feedback removes the room.

It’s the only reliable way to make the correct outcome and also the easiest one for the factory to deliver.

02. Measure against the spec, point by point:

Before you write a single comment, measure. Do not eyeball.

Lay the sample flat on a clean surface, smooth out the wrinkles, and measure every point of measure in your tech pack.

Record the actual number next to the spec number.

This one habit removes more guesswork than anything else in this guide.

Know your tolerance:

Tolerance is the acceptable range above or below your spec, written with a +/- sign.

A chest of 20½” with a ±½” tolerance passes anywhere between 20″ and 21″.

Sampling tolerance is usually tighter than production tolerance.

You hold the sample to a stricter standard because it sets the pattern everything else copies.

Tolerances are not one number applied everywhere.

Larger measures of chest, hip, and body length carry a wider tolerance, often around ±½”. Smaller, fit-critical measures like neck drop, cuff opening, and placket carry a much tighter one.

The rule of thumb: the more a point affects fit, the smaller its tolerance.

If you don’t have a tolerance table yet, your patternmaker or factory can set one with you, and it belongs in the tech pack.

See how the spec sheet is built in our tech pack guide.

Point of measure Spec Sample tolerance Typical size
Chest (1″ below armhole) 20½” ±½” Larger / relaxed
Body length (HPS to hem) 28″ ±½” Larger / relaxed
Shoulder width 18″ ±¼” Tighter / fit-critical
Sleeve length 24″ ±¼” Tighter / fit-critical
Cuff opening 3½” ±⅛” Tightest

Don’t comment on what’s in tolerance:

This is the discipline most founders skip.

If a measurement sits inside tolerance, leave it alone even if it’s not the exact number.

Commenting on in-tolerance points trains the factory to treat your whole sheet as approximate, which is the opposite of what you want.

Save your comments for what’s actually out.

Use the factory’s own language:

When a measurement is out of tolerance, the industry phrase is “return to tolerance.”

It tells the factory exactly what you want without ambiguity.

Better still, your tech pack already numbers every point of measure.

Those POM codes are the shorthand the factory reads fastest.

Reference the code, not a description, and your comment lands on the right point every time.

When out-of-tolerance actually fits:

Sometimes a measurement is out of spec, but the garment fits beautifully on a body or fit model.

In that case, change the spec; don’t ask the factory to “fix” a sample that’s already right.

The tech pack serves the garment, not the other way around.

Update the number and note it so the next round measures against the truth.

Compare the corrected sample round over round:

When the corrected sample arrives, don’t only measure it against spec.

Measure it against the previous sample too.

That catches two things a spec-only check misses.

First, drift: a point that was in tolerance last round can move when the factory adjusts a nearby seam.

Second, over-correction: you asked for minus one inch and got minus one and a half.

Keep every round’s numbers in the same sheet, one column per round, so the pattern is visible at a glance.

If a must-fix from round one is still out in round two, that’s the headline of your next comment, not a new issue buried at position nine.

03. Write actionable, prioritized comments:

Now you have measured facts.

Turn each one into a comment the factory can act on without replying to you.

Two devices make this repeatable, but first, a diagnosis step.

First, separate a defect from a spec error:

Every issue is one of two things, and they don’t get the same comment.

A defect means the factory didn’t build to your tech pack; the spec said 20½”, and the sample came in at 22″.

That’s theirs to fix, and “return to tolerance” is the whole comment.

A spec error means the factory built exactly what you asked for, but what you asked for was a wrong placement, a measurement, or a construction note that was off in your pack. That’s yours.

The fix is to correct the tech pack, not to send it back as a factory mistake.

Blaming a factory for building your spec correctly is the fastest way to burn goodwill and slow the relationship down. Diagnose first, then write.

The Three-Line Comment:

Every comment you write gets exactly three lines. No more, no less.

This forces you to locate the issue, quantify it, and state the fix for the three things a factory needs, and a vague note is always missing.

Point: Where it is: the POM code or a named location.
Problem: What’s off, measured against the spec?
Target: The exact result you want.

Here’s the difference it makes on a single issue.

✗ Vague

“The sleeves feel too long, can you shorten them a bit?”

✓ Three-Line

Point: C12 sleeve length.
Problem: 25″, spec 24″, +1″ over, out of tolerance.
Target: Return to 24″ ±¼”.

The vague version starts a conversation.

The Three-Line version ends one.

The factory reads it, cuts to the number, and moves on.

Multiply that across twelve comments, and you’ve removed twelve possible misreads from the round.

It works for things you can’t measure with a tape, too.

A print, a seam, and a wash are the same three lines.

✗ Vague

“The print looks a bit off, can you sort it out?”

✓ Three-Line

Point: Front chest print placement.
Problem: Sits 1″ high; top edge is 6″ below HPS; spec is 7″.
Target: Drop to 7″ below HPS, centered on CF.

Notice the move: even a “soft” issue gets anchored to a measurement and a reference point.

Anything you can locate and quantify stops being a matter of opinion, which is exactly what makes it actionable.

The Priority Ladder:

Sort every comment into one of three tiers.

This is the piece that tells the factory what blocks approval, the single most common thing missing from founder feedback.

Must-fix

Blocks approval:

The sample is rejected until this is corrected.

Triggers a resample.

Reserve this tier for genuine deal-breakers: fit, construction, and wrong spec.

Should-fix

Correct next round:

Worth fixing, but not on its own a reason to reject and recut.

Bundle these recuts.

The next sample is in production.

Note

Log for production:

An observation, a preference, a thing to watch not a re-sample trigger.

Recorded so nothing is forgotten, but it doesn’t stall the round.

Without the ladder, the factory sees twelve equal comments and either overworks the sample or stops to ask which ones count.

With it, they know instantly: fix the three must-fixes, address the should-fixes if there’s time, and log the notes.

Approval gets faster because the decision is already made.

Four rules for the comment sheet:

  • Number every comment: The factory replies “done on 3, 7, 9,” and you both track progress against the same list.
  • One issue per comment: Two problems in one line means one of them gets missed.
  • Lead with fixes, not praise: A short “Overall, this is close” is fine, but don’t bury the must-fixes under compliments.
  • Give targets, not adjectives: “Cleaner” is an adjective. “Topstitch 1/16″ from edge, single needle” is a target.

04. Annotate the photos and the physical sample:

Words alone lose detail.

A comment that says “the print sits too high” is still open to interpretation until the factory sees exactly where you mean.

Annotation closes that gap, and it’s fast.

Photograph the sample properly:

Lay the sample flat against a plain background in daylight.

Put a ruler or tape measure in the frame so the scale is unmistakable.

Shoot the whole garment, then a close-up of every issue.

Bad lighting and busy backgrounds hide exactly the details you’re trying to point at.

Mark the image directly:

In the photo, add an arrow or circle at each issue, write the POM code beside it, and drop the three-line comment as a caption.

Any phone markup tool does this.

The factory now has the fact, the location, and the fix in one image, with no cross-referencing to a separate sheet.

Tag the physical sample:

Some problems don’t photograph.

Hand feel, seam pucker, a wash result, and a weight that’s off need the factory to hold the garment.

Pin a small colored tag at each issue on the sample itself, using the Priority Ladder colors: one color for Must-fix, one for Should-fix, and one for Note.

Then courier the tagged sample back alongside your annotated photos.

Send both.

Annotated photos give the factory a fast read they can start on immediately.

The tagged physical sample carries everything a photo can’t.

For the construction vocabulary that makes these annotations precise, see our decoration methods guide and the fabric and GSM guide.

05. The feedback template that speeds approval:

Put it all in one repeatable form.

It has the same structure every round, so the factory learns to read it fast, and nothing gets lost between rounds.

Here is the template.

# Point (POM/location) Priority Problem (actual vs spec) Target
1 C7 chest width Must-fix 21½”, spec 20½”, +1″ over Return to 20½” ± ½ ±½± ½”.
2 Left chest print Must-fix Sits 1″ high vs. placement Drop 1″, center on C-line.
3 C12 sleeve length Should-fix 25″, spec 24″, +1″ over Return to 24″ ±¼.”
4 Neck rib Note Hand slightly firm Watch in production.

Close every sheet with a verdict line so the factory knows the outcome at a glance:

Approved, approved with corrections, or re-sample required, plus the date you need the next round by.

That one line converts a list of comments into a decision.

Read the reply against your own list:

Because every comment is numbered, the factory can confirm against the same list: “Done on 1, 2, 4; 3 held for production.”

Before you approve the next sample, check it comment by comment against that reply.

Don’t approve on a general impression that it “looks better.”

Approve because each must-fix is measurably resolved.

The numbered list is what makes that check fast, and it’s what stops a resolved issue from quietly reopening two rounds later.

The full 12-point checklist:

The whole process, start to finish. Work down it every time a sample lands.

  1. Lay the sample flat and measure every point of measure in the tech pack.
  2. Record actual next to spec for each point; flag anything out of tolerance.
  3. Ignore everything inside tolerance; don’t comment on it.
  4. For each out-of-tolerance point, decide: fix the sample or change the spec because it fits.
  5. Write each issue as a three-line comment: point, problem, and target.
  6. Reference the POM code, not a description, for every location.
  7. Assign each comment a priority ladder tier: must-fix, should-fix, or note.
  8. Number every comment; one issue per line.
  9. Photograph the sample flat, in daylight, with a ruler in frame.
  10. Annotate the photos with arrows, POM codes, and three-line captions.
  11. Tag the physical sample by priority color for anything a photo can’t carry.
  12. Fill out the template, add the verdict line and next-round date, and send photos + sample.

From the manual

This is one chapter of The Complete Streetwear Manufacturing Guide

These chapters cover taking a streetwear brand from idea to shipped units, costing, sourcing, tech packs, QC, freight, and launch. All written from the Sialkot factory floor.

Open The Full Guide

FAQ

Two to three rounds are common for a new style: a first proto, a corrected sample, and often a pre-production sample.

Clear, spec-anchored feedback is the single biggest lever for keeping that number low.

Change the spec, not the sample.

If the garment fits well on a body or fit model, the tech pack number is what’s wrong.

Update it and note the change so the next round measures against the corrected spec.

No. If a point sits within its tolerance, leave it.

Commenting on in-tolerance points signals that your whole spec sheet is approximate, which encourages the factory to treat every number loosely.

Both. Annotated photos give the factory a fast read they can act on immediately.

The physical sample, tagged by priority, carries everything a photo can’t: hand-feel, pucker, weight, and wash result.

Feedback that’s unlocated and unranked.

When the factory can’t tell where an issue is or which comments block approval, it either overworks the sample or stops to ask and both cost a full round.

On the floor · Sialkot

Written by

Faizan Ahmad

Chief Apparel Technologist & Head of Manufacturing, Gibben Clothing · Sialkot, Pakistan

Faizan leads production at Gibben Clothing, a cut-and-sew streetwear manufacturer in Sialkot, with 8+ years turning raw yarn into retail-ready hoodies, tees, bottoms, jackets, tracksuits, and headwear. He doesn’t just write about clothing; he works the floor, so every guide here is grounded in real fabric behavior, QC standards, and production data from live runs.