Sign Shops Bringing UV Flatbed Printing In-House: The Volume Where Outsourcing Stops Paying

The question of when a sign shop should buy a UV flatbed printer instead of outsourcing almost never gets answered with arithmetic. It gets answered with a feeling — usually the feeling you get in month four of a busy season, when the trade printer misses a Friday and you eat the install crew's day rate. That feeling is real, but it is not a business case, and it will not survive a conversation with whoever signs the cheque.

The arithmetic is not hard. It is one crossover equation with about seven inputs, and every one of those inputs can be pulled from documents already sitting in your office. What follows is that equation, written so you can drop your own numbers into it. The worked figures are placeholders — they exist to show the shape of the calculation, not to tell you what your answer is.

The two cost curves

Outsourcing is a pure variable cost. Every square metre you send out costs the same as the last one, so its cost curve is a straight line through the origin:

Outsourced monthly cost = V × P_out

where V is your monthly rigid print volume in m², and P_out is your blended outsourced price per m².

In-house is a fixed cost plus a much smaller variable cost. Its curve starts high and rises slowly:

In-house monthly cost = (A + L + S) + V × (I + M + R)

where:

• A = monthly amortisation of the landed machine cost
• L = incremental monthly labour
• S = monthly service, space, power and insurance
• I = ink cost per m²
• M = maintenance and consumables per m²
• R = reject and reprint cost per m²

Set them equal and solve for volume. The crossover is:

V* = (A + L + S) ÷ (P_out − I − M − R)

Above V*, in-house is cheaper. Below it, outsourcing is. The denominator is the contribution each square metre makes toward covering your fixed costs — think of it as gross margin per m² on work you were previously buying.

That is the whole model. The difficulty is not the algebra; it is that five of the seven inputs are routinely wrong on the first pass.

Getting P_out right: use invoices, not the rate card

Your trade printer's rate card is not your cost. Three things sit between them.

First, minimum charges. Most trade printers apply a job minimum. On a 0.4 m² job with a rate card of, say, $22/m², you are not paying $8.80 — you are paying the minimum, which might be four times that. If half your job book is small panels, your blended P_out is materially higher than the card rate, and that pushes your crossover volume down.

Second, freight and handling in both directions. Sheets go out, printed panels come back, someone loads and unloads both. If you courier, that is an invoice line. If you drive, it is a van hour.

Third, the reprint you paid for anyway. Trade printers usually stand behind their own defects, but they do not stand behind your file error found after delivery, and they rarely stand behind a colour match you approved on screen.

The correct method: take twelve months of trade-print invoices, total the amount paid, total the square metres bought, and divide. That single blended number is your P_out. It is usually 15–40% above the rate card, and it is the number the business case lives or dies on.

Building the in-house cost line

The variable side, split three ways

The most common modelling error is a single ink cost per square metre. Ink consumption on a UV flatbed is driven by coverage and by how many layers you lay down, and those vary enormously across a sign shop's job book.

Ask any machine builder for ink cost per m² in at least three classes, and ask for the coverage assumption behind each:

Machines that print white and varnish layers change what you can quote for — but they also change what a square metre costs you. If 30% of your volume is second-surface acrylic with a white flood, a blended I built only from CMYK figures will understate your variable cost badly.

Two more lines belong in the variable bucket and are usually forgotten:

• M, maintenance per m². Printhead purges, wipes, filters, capping-station parts, and the amortised cost of a head replacement. A head set is a step cost, not a smooth one, so spread its replacement price across the square metres a head set is expected to deliver — and ask the builder for both figures.
• R, rejects. In-house, a scrapped panel costs you the substrate as well as the ink. On rigid work the substrate is often the larger half. Take your reject rate as a percentage, multiply by (substrate cost + ink cost) per m², and put it in the model. A 4% reject rate on a $9/m² sheet is $0.36/m² before you count the reprint.

The fixed side, and why labour is a staircase

A is straightforward arithmetic on the wrong number if you use the quoted machine price. What belongs in A is the landed, installed and running cost divided by the months of useful life you will actually get.

For an imported machine, that means the ex-works price plus ocean or air freight, insurance, duty, customs brokerage, port and inland haulage, rigging into the building, any electrical work to get the right supply to the machine, the starter ink and spares package, and commissioning and training if those are quoted separately. Our landed cost calculation guide walks through the line items, and the hidden costs of sourcing from China covers the ones that appear after the invoice is paid. A flatbed in the 0.5–1.5 tonne class is a crated, crane-or-forklift-off-the-truck item, not a pallet you wheel through a fire door — plan the rigging before you plan the payback.

L is the input that most often breaks a spreadsheet, because incremental labour is a staircase, not a slope. At low volumes, an existing prepress or finishing operator absorbs the machine into their day and L is close to zero. Past some volume, that stops being true and you hire — at which point L jumps by a full loaded FTE and your crossover volume jumps with it. Model both steps. If your projected volume sits between them, you are in the zone where the decision is about workflow design rather than about the machine.

S is the quiet one: floor space at your actual rent, power, compressed air, extraction if required, machine insurance, and any annual service contract. Ask for the power draw and compressed-air demand in the quotation, not after delivery.

Running the crossover

Here is the equation with illustrative placeholders. Substitute your own numbers — these are chosen to demonstrate the method, not to describe any particular machine.

Assume I = $2.40/m², M = $1.10/m², R = $0.45/m². Variable total = $3.95/m².

Assume fixed costs in two scenarios: absorbed labour (A $1,000 + L $0 + S $250 = $1,250/month) and dedicated operator (A $1,000 + L $3,200 + S $250 = $4,450/month).

To translate into something you can check against your own job board: a 2440 × 1220 mm sheet is 2.977 m². At $22/m² blended and absorbed labour, 69 m²/month is about 23 full sheets a month. With a dedicated operator, 247 m²/month is about 83 sheets a month.

Now the three ways this table lies if you stop here.

It ignores the capacity ceiling. V* is only meaningful if your machine can actually produce V* in the hours you plan to run it. Ask the builder for throughput in m²/hour at the print mode you will actually sell — not the draft-mode headline — and multiply by realistic uptime. If V* sits above your practical capacity, the honest answer is that the machine does not pay back on cost substitution alone.

It ignores mix drift. The moment printing is in-house, the job mix changes: you quote work you used to decline, and you say yes to Thursday-afternoon reprints. That is usually good, but it means the I you modelled from last year's mix will drift upward if the new work is white-heavy.

It treats turnaround as free. It is not free, and it is not a cost line — it is a revenue line. A run of one costs the same per piece as a run of a thousand when there is no plate, film or screen to make, which is what makes same-day single-panel work sellable at all. Model that as incremental margin on jobs you cannot currently win, in a separate calculation, and do not smuggle it into the crossover to make the number look better.

Three details that change the answer

• The bed size determines your yield, not your capacity. A 2500 × 1300 mm table takes a 2440 × 1220 mm sheet with roughly 30 mm of margin on the long edges. That margin is the difference between running your existing stock list and trimming every sheet before it goes on the bed. If your stock is 3050 × 2030 mm, your real variable cost includes a cutting operation you have not modelled.
• Reject cost is asymmetric between the two options. Outsourced, a spoiled panel is usually the trade printer's substrate. In-house it is yours. Shops that model rejects as ink alone understate R by roughly the substrate share, which on rigid board is often two-thirds of the piece cost.
• A head replacement is a cliff in year three, not a line in year one. If your amortisation period is five years and the head set is rated for less than that, your A is wrong unless you have put the replacement into it. Ask for the printhead price list at quotation time, while you still have bargaining power, rather than at failure, when you have none.

What to ask before you commit

Work through this in order. Steps one to four cost nothing and settle most of the question.

1. Blend your own P_out. Twelve months of trade-print invoices, total dollars ÷ total m². Note separately what share came from jobs that hit a minimum charge.
2. Histogram your volume. Monthly m² for those twelve months, and the ratio of CMYK-only to white-underbase work. You need the mix, not just the total.
3. Compute V* at both labour steps. Then mark on the histogram how many of the last twelve months exceeded it. If the answer is "three, all in Q4", you are looking at a seasonal capacity problem, not a sourcing decision.
4. Get a landed cost, not a machine price. Ex-works plus freight, duty, brokerage, haulage, rigging, electrical, starter ink and spares, commissioning. Confirm Incoterms in writing — EXW, FOB and CIF put very different amounts of that list on your side of the line.
5. Ask the builder for the five numbers the quotation usually omits: m²/hour in production mode, ink consumption per m² by coverage band and ink class, printhead service life and replacement price, power and compressed-air demand, and the assumed operator headcount per machine.
6. Send your own substrates for a test print before the balance payment. Not the builder's demo file on the builder's board — your acrylic, your ACM, your artwork, with the white and varnish layers you actually sell. Machines that can be sampled the same day on standard models make this easy to ask for; a supplier who resists it is telling you something.
7. Write the acceptance test into the contract. Under 30% deposit and 70% before shipment, your remaining bargaining power sits entirely in that 70%. Define pass/fail on your own files before you release it.
8. Price the after-sales terms as part of the machine. Warranty length, spare-parts price list, remote support hours, and who pays for an engineer's travel. See our UV flatbed printer buyer's guide for how those terms are usually structured.

If you want a concrete counterparty to run this exercise against, Longrun Printing Machinery in Shenzhen builds UV flatbed machines — including a 2513 model with a 2500 × 1300 mm bed — sells from a single unit, and offers same-day sampling on standard models, which is exactly the kind of supplier you want to hand your own artwork and your own board to before the numbers above become a purchase order.