Getting Size Runs Right on Large Staff Orders

Building the right size run for staff uniforms is the difference between a rollout that lands clean and a rollout that generates three weeks of swaps. On an order of several hundred pieces spread across sites in Miami, Hialeah, Doral and Fort Lauderdale, the garments are the easy part. The hard part is knowing, before you commit, how many Smalls and how many 3XL you actually need, and then holding enough of the right sizes so the person who starts on Monday is not waiting. This is the method we use with large South Florida employers: how to collect sizes without chasing anyone, why what people tell you is often wrong, how to turn raw answers into a curve you can order against, and what to keep on the shelf afterward.

Why size runs break at scale

At thirty people you can eyeball it. At four hundred across five locations you cannot, and the failure is always the same shape: the order was built from the replies that came back easily, and the replies that came back easily came from the people sitting closest to a computer.

The response rate problem

Send a form to a large workforce and you will get a partial answer. Office and administrative staff respond quickly. Route drivers, floor crews, night shifts and anyone whose only channel is a supervisor who forwards nothing will not respond at all. If you order against that data you have effectively measured your front office and handed the result to your warehouse, and the sizes that run heavier in field roles arrive short.

The silent default

When the data is thin, something fills the gap. Usually it is either a flat spread, equal counts of every size, or a guess weighted toward Large. Both are expensive in opposite directions. A flat spread strands the sizes almost nobody wears and shorts the two or three that most people wear, which is precisely the shortage you feel in week one. A guess weighted to Large leaves you reissuing the small end one piece at a time.

Partial data is worse than late data

A curve built on sixty percent coverage looks complete. It has a number in every cell. That false completeness is why so many programs commit early and then discover the gap only after the boxes are opened, when the quantities are already decorated and the window to adjust has closed. Treat coverage as a number you report, not a detail you assume.

Collecting sizes without chasing

The goal is one pass, one deadline, and no follow up campaign. That is achievable if you stop treating size collection as a survey and treat it as a roster reconciliation. You are not asking an open question. You are filling a known number of rows.

1. Export the roster first. Pull active headcount from payroll or your HR system, broken out by location and by role, before you ask anyone anything. Now you know the denominator. Every later decision depends on knowing you need four hundred and twelve rows, not roughly four hundred.
2. Assign the ask to a person at each site. Give every location one named owner, hand them their own roster segment, and make the deliverable a completed list rather than forwarded emails. A supervisor who has to return a list with no blanks will walk the floor. An inbox will not.
3. Name the actual garment. Tell people which blank they are sizing for and show the garment, not a size chart in the abstract. A person who sizes for a relaxed cotton tee and receives a closer cut will tell you the order was wrong even when the label matches.
4. Reduce it to one question with a short list. One field, one list of adult sizes, no free text. Free text produces entries like medium slash large and tall, which someone then has to interpret, which is how interpretation errors enter a locked file.
5. Publish the default before the deadline. State in the request that anyone who has not answered by the stated date will be issued a specified size. This converts non response from a data hole into a known, documented decision, and it raises response rates more than any reminder.
6. Put a fit set at each site. One piece of each adult size, in the exact blank you are ordering, hung where people can try it. This single move removes most of the swap volume on large programs. If you are not sure about a blank, order a sample before you launch the program.
7. Lock the file and keep it. The finished list is an asset, not a one time input. It becomes the basis of your buffer, your reorders and next year's curve, so store it with the garment, color and decoration spec attached so later runs are reorders that match what you already issued.

Why self reported sizes skew

People do not report their size. They report a mix of fit preference, habit, and whatever the last brand they bought happened to call that fit. Three forces account for nearly all of it.

Fit preference is not a size

Ask for a size and you get an answer about how a person likes clothes to sit. Some people size up by one for room, others size down because they do not want a boxy shirt. Both answers are correct about preference and both are wrong as data. If you are issuing a worn every day uniform, say so in the ask, because that reframing moves answers.

Sizes drift between blanks

A Large is not a fixed quantity. A heavier, boxier cotton tee and a closer, longer retail cut do not measure the same across the chest even with the same label, which is why people sometimes wear different labels in different brands and are not confused at all. Settle the blank before you collect, not after, and work through the tradeoffs in our guide to choosing blanks for company apparel.

The asking context changes the answer

Sizes collected in front of a supervisor, on a clipboard, in a group, skew differently than sizes submitted privately. Nobody is being dishonest. It is simply a question people would rather answer quietly. Allow a private channel back to one named coordinator and the edge sizes come in accurately instead of being guessed later.

Building a size curve

A size curve is just your roster expressed as counts by size, with a documented rule for every row you could not fill. Build it in layers and keep each layer visible, because when something is short you need to know which layer was wrong.

Separate the core band from the edge band

In most workforces a small number of adjacent sizes absorb the majority of people, and the remaining sizes are held by a countable number of individuals. Order the core band in volume, because that is where quantity pricing works for you. Order the edge band by name, because a size worn by six people should be ordered as six pieces with six names beside them, not as a rounded estimate that leaves one person uncovered or four pieces in a box.

Run the curve per location, then combine

A Doral distribution site and a Brickell office will not share a curve, and averaging them produces a shape that fits neither. Keep each location's counts separate for allocation, then combine only for the purchase quantity so you still get the volume tier. Multi location programs have their own failure modes, which we cover in the guide to running a company uniform program.

Split the curve by garment, not just by headcount

Role mix changes which garments each person gets, and garments do not share a curve. A polo for embroidery is often offered in separate cuts, a cotton tee for printing is not, and a hat is a different conversation entirely. Build one curve per garment line so the counts add up to the same headcount without quietly double counting anyone.

Buffer stock for new hires

Buffer is the layer most programs skip, and it is the layer that determines whether the program feels organised in month four. It is not a second order placed in a hurry. It is a planned quantity of undecided pieces held for people you have not hired yet.

What actually drives the number

Four inputs set your buffer, and all four are numbers you already have: how many people start in a normal month, how much turnover each role carries, how many sites need to be covered independently, and how long you are willing for a new starter to wait. A property management group in Aventura hiring steadily needs a different buffer than a professional services firm that adds a few people a quarter, even at identical headcount.

Hold the core, order the edge

Keep buffer only in your core band, where the next hire is statistically likely to land, plus a light layer one size either side. Do not try to hold the full range at every location, because edge sizes sitting in five separate closets is how dead stock happens. Hold edge sizes centrally, or simply reorder them as named hires appear. Because there is no minimum on bulk and repeat orders, a single 3XL is an order you can actually place, which is the whole reason the hold the core strategy works.

Write down what buffer means for your approvals

Buffer only functions if someone is allowed to draw it down and someone is required to replace it. Name both. Set a trigger quantity per size per site, and make the top up a standing action rather than a new approval cycle each time. Teams that tie their buffer into a repeatable onboarding handoff stop having the same conversation every month.

What a wrong curve costs you

The cost is rarely the garments. It is everything that attaches to them.

• Day one shortages. A new site opens and the people most visible to customers are the ones out of uniform, because the heaviest used sizes went first.
• Dead stock. Decorated pieces in sizes nobody wears cannot be returned, redistributed or sold. They are a line item that never clears.
• Swap labour. Every exchange consumes a supervisor's time, a coordinator's time, and a shipment between sites. Dozens of those add up to more than the order error.
• Brand drift. The usual fix for a shortage is a fast top up on whatever blank is available, in whichever decoration method is quickest. Now your Fort Lauderdale crew is in a visibly different shirt than Hialeah, and the inconsistency outlives the shortage.
• Lost pricing. Quantity is what moves cost per piece down. Splitting one run into a main order plus three corrective orders gives up that leverage for no benefit.

Garment and print decisions

Decoration method does not change your curve, but it changes what you can hold undecorated and how a logo reads across the size range.

Match the method to the fabric, not to the order

Our decoration is all in house. Kornit DTG printing, on industrial Kornit Atlas MAX PLUS direct to garment printers, handles cotton and cotton rich blanks with water based ink that stays soft and does not crack or peel the way heavy plastic transfers do. DTF transfers cover polyester, blends and performance fabrics, which is where most field and athletic style staff apparel lives. Custom embroidery covers polos, hats, bags, jackets, aprons and beanies. If you are weighing method against garment, start with why DTG suits business apparel.

Decide whether the logo scales with the garment

Standard placements are front 12 by 16 inches, back 12 by 16 inches, left chest 5 by 5 inches and top back 12 by 4 inches. A fixed left chest mark sits differently on a Small than on a 3XL, and on a large staff order that difference is visible when people stand together. Decide deliberately whether the mark holds one size across the run or steps with the garment, and record the decision in the spec. The print placement reference and our placement guide both lay out the measurements.

Keep the spec with the curve

Garment, color, decoration method, placement, artwork file and curve belong in one record. When you reorder eight months later for a new Doral site, that record is what makes the new stock sit beside the old stock without anyone noticing a difference. Store it once and every later order becomes a quantity decision instead of a research project.