Custom work doesn't fail because fabricators are careless. It fails because the spec left a decision open, and someone filled the gap with a reasonable guess that wasn't yours.
When you specify a stock lounge chair, the manufacturer has already made a thousand decisions for you: frame construction, seat depth, foam density, stitch pattern. You pick a SKU and a fabric, and the SKU carries the rest. A custom banquette has no SKU. Every one of those decisions is either written down by you or made by someone else — usually the fabricator, sometimes the installer, occasionally the owner's rep on a site walk.
On a commercial renovation, the stakes multiply. A guestroom vanity isn't one piece; it's forty. A late remake doesn't just cost a reorder; it can push an opening date. This article covers how to spec, document and communicate custom work so it survives every handoff — without boxing in the creative decisions that make it custom in the first place.
If you're looking for a cloud-based collaborative ai-powered specification management platform that will help you implement these processes with ease, book a demo here.
The gaps tend to show up at four points between concept and install.
Most of these aren't craft failures. They're information failures. The piece gets built correctly — to the wrong information.
And the cost of fixing it climbs at every stage. A wrong dimension caught at spec costs five minutes. At shop drawing, it's a day and a redline. After fabrication, it's a remake, a new lead time and a call with an owner whose opening date just slipped. At install, add an idle crew and a GC sending you their delay costs.
That opening date is where commercial work gets expensive. On a 150-key hotel at $200 a night and 70% occupancy, every week of delay is roughly $150,000 in room revenue that never comes back.
Here's a composite story. The details are changed, but every part of it has happened to someone.
A design firm is renovating the lobby bar of a 150-key hotel, and the reopening date is already announced. The star of the room is a curved, upholstered banquette wrapping the bay window: channel-tufted back, slim base, deep green velvet. On the concept board, it looks great.
Concept to spec. The spec says "channel-tufted back, 18-inch seat, green velvet." It doesn't say how many channels, how wide they are, or whether they follow the curve. It doesn't say whether that 18 inches is measured with the cushion compressed.
The shop drawing. The upholsterer draws seven wide channels, because that's their standard. The designer was picturing eleven narrow ones. But they're on a job site, looking at it on a phone, and the outline looks right. So they approve it.
A revision. The brand team decides the green reads too dark in morning light and picks a lighter one. The designer emails the upholsterer, and the fabric order is updated. But the welt cord comes from a different vendor, and nobody tells them. Three weeks before reopening, the banquette arrives in light green with dark green trim.
The invoice. The curve needed a form the shop didn't have. The fabric was calculated for a straight run, so it took two more yards. A curved frame doesn't ship flat, so freight went up. The owner approved $14,800. The final bill was $19,600.
Whose fault was that? Not really the upholsterer's. They built exactly what the documents said and made reasonable guesses where the documents were silent. Every problem goes back to a decision that was never written down, or written down in one place and not the others.
The worry we hear most from designers: "If I spec every detail, I lose the flexibility that makes custom work custom." The opposite is true. A precise spec protects your creative decisions by making it clear which ones are yours — and it tells the fabricator exactly where they're free to use their craft.
For every custom item, each attribute belongs in one of three buckets:
If an attribute isn't in one of those buckets, it's a gap, and gaps get filled by somebody else.
Here's a line you've seen on plenty of FF&E schedules: "Custom vanity, king suite typical, walnut, matte finish, floating, 60 inches, double sink." Here's the same vanity written with intent, tolerance and discretion:
| Attribute | Spec |
|---|---|
| Overall | 60" W × 21" D × 20" H box; top of counter at 34", ±1/8" |
| Species and cut | Rift-cut walnut veneer; grain vertical and continuous across both doors |
| Finish | Stain to approved sample #S-04, dated and initialed; sheen 10°, ±5 |
| Reveals | 1/8" between all doors and drawers |
| Sink cutouts | Per plumbing fixture spec sheet; centered on each basin bay |
| Construction | Fabricator's standard; show joinery, drawer box and wall mount on shop drawings |
| Field verify | Final width in the model room after drywall; hold the production run until signed |
Every number has a unit. The finish points to a physical, dated sample instead of a word like "matte." And the last line names a hold point, so nobody builds forty of them before the wall is final.
| Instead of | Write |
|---|---|
| "Slim," "chunky," "standard" | A dimension with a unit and a tolerance |
| "Matte," "natural," "warm" | A sample ID, a sheen value, and who approved it |
| "To match" | To match what, where, and under what light |
| "Similar to" | The attributes that must match, and those that may differ |
| "TBD" | Who decides, by what date, before which milestone |
Finish disputes are won or lost on samples, and a sample only helps if everyone is looking at the same one.
Stone deserves extra care because it's expensive and irreplaceable. For a reception desk top, a bar die or a run of vanity tops, document the slab by lot number or tag, the vein direction drawn on the template, the edge profile as a dimensioned section, the thickness and allowed seam locations, and the cutouts with fixture cut sheets attached.
The short version: a word describes a feeling, and a number describes a thing. Put feelings in your concept presentation. Put numbers in your spec.
Every custom spec has three readers.
The owner and brand read it to confirm you met their intent.
The fabricator reads it to price and build.
The installer reads it to put it in the right place.
Before a spec goes out, reread it as each of them. If the owner wouldn't recognize their vision, add the intent. If the fabricator would have to guess, add a number. If the installer would have to call you, add the field condition.
And keep the evidence attached to the item itself: the elevation, the approved sample photo with its ID, the fixture cut sheets and the field measurements. If the fabricator has to ask for it, it's already late.
A great spec is necessary but not sufficient. It will be read by an estimator, redrawn by a drafter, cut by a machinist, finished in a spray booth and installed by a crew that's never seen your mood board. Every one of those is a translation, and translations drift. Three controls keep them honest.
The submittal is where you find out whether the fabricator understood you. For every custom item:
Timing varies by item. Many custom shops won't draw until the order is placed, while contractor-supplied architectural items are often submitted before it. Know which you're dealing with, and put it on the schedule.
When you review, you're not really hunting for mistakes. Most of what's off on a shop drawing isn't wrong from the fabricator's point of view; it's their default. Your job is to find every place their default and your intent don't match. Watch for dimensions taken from a different edge than the one you care about, small deviations in reveals or radii, cutouts placed for drafting convenience, and missing hold notes. Pay special attention to silence: if the drawing doesn't show something, it usually means "our standard," and their standard may not be yours.
A few habits make review faster. Go in the same order every time: dimensions, materials and finish, details, hardware and cutouts, then notes and holds. Use one color for "must change" and another for "question." And write the spec reference next to every redline, so the drafter can check it without calling you.
The most dangerous document on a custom project is the second-most-recent version. It looks right, it's in someone's inbox, and it's wrong. You should be able to answer three questions about any custom item in under a minute:
If any answer means searching email, the process lives in people's heads. Put a revision number and date on every spec and drawing, keep a short change log on the item itself, and ask the fabricator to confirm receipt in writing.
The banquette's welt problem was a revision notice problem. A notice that works covers six things:
| Field | Example |
|---|---|
| Item and revision | Banquette B-1, Revision 3, dated October 2 |
| What changed | Body fabric from Color 14 to Color 09; self welt, same dye lot |
| What did not change | Frame, channel layout and seat height unchanged from Rev 2 |
| Who needs to act | Upholsterer, fabric vendor and welt supplier |
| Cost and schedule impact | None expected; confirm within two business days |
| Acknowledgment | Confirm receipt of Rev 3 and remove Rev 2 from the floor |
Two lines do most of the work. "What did not change" stops people from reopening settled decisions. And naming everyone who needs to act is what would have saved the welt — the upholsterer was never going to forward your email to a supplier they didn't know was involved.
Hold points are milestones where work stops until something specific is signed off. On commercial renovations, the common ones are:
Write them into the spec itself. They protect the fabricator as much as they protect you — which is why it's worth walking your fabricator through submittals, holds and revisions at kickoff. Good shops welcome it. It keeps them from being blamed for gaps in the documents, and it means fewer remakes eating into their margin too.
One last note on communication: not everyone needs every document. The fabricator needs the spec, drawings and samples, but not your pricing. The owner and brand need the approved selections and renderings, but not your trade cost. The GC and installer need drawings and hold-point sign-offs. When everyone works from one source but sees only their part, fewer people are tempted to work from a stale PDF.
Owners remember the first number you show them. It's the number in their budget and their pro forma, and every dollar added after approval spends trust, even when the addition is legitimate. The aim is to show the real number the first time.
Consider a custom corridor pendant quoted at $2,000 each — the fabricator's price for the piece. By the time it's installed, the real cost is closer to $5,000 each: overseas freight, crating, import duties, a finish upcharge and an electrician's time for a non-standard canopy. Each cost lived in a different place, so no one saw the total until the invoices arrived. Across thirty fixtures, that's $90,000 the owner never budgeted.
The fix is to price from the parts up. For every custom item, you should be able to see:
When those roll up into one price before approval, the conversation shifts from "why did this go up?" to "here's what it takes to build this."
Custom projects change. What damages trust is change without a record. Describe the change and its reason, show the cost and schedule impact before work proceeds, get the owner's written approval, and update the spec, drawings and fabricator the same day.
Here's what that looked like for the banquette's fabric change, in a five-sentence note to the owner:
"You've asked to change the body fabric from Color 14 to Color 09. The fabric is the same price per yard and carries the same flammability and abrasion ratings. Because the curve needs two additional yards regardless, there's no cost change from the revised quote. The mill shows a two-week lead time on Color 09, which moves delivery from November 6 to November 20 — still ahead of the December 1 FF&E install, so the reopening date holds. Please approve by Friday so we can hold the dye lot."
The owner knows what changed, what it costs, whether it touches the opening date and what you need from them. Total time: about fifteen minutes.
The mistake to avoid is presenting an allowance as if it were a fixed price. If a number is still moving, label it.
Internally, know your margin before you present. Someone on your team should be able to say what the item costs, what the markup is and what's left after freight and install. Custom pieces are where firms earn their reputation — and quietly lose money. Track actual against quoted cost on every custom item, and review the gaps when the project wraps.
Finally, share selections, not spreadsheets. A clean document with the image, approved finish, key dimensions and price reads as professional service. A forty-column spreadsheet reads as a problem the owner now has to solve.
Everything above works on paper or in a spreadsheet, and plenty of good firms run it that way. What a platform adds is keeping the standard, the spec, the materials and the owner-facing document connected, so nobody works from a stale copy.
That's the idea behind SpeCurate, Fohlio's specification platform. Product types carry attributes, requirements and instructions your firm can edit, so the same questions get asked on every project. Fabricated items get their own structure, so a channel count or a mock-up hold point has a place to live instead of disappearing into a description field. And the same data produces a branded presentation for the owner and a spec report for the fabricator, so each person gets the document they need.
Your custom work deserves better than a forty-column spreadsheet.