You have picked home decor. Now you need someone to actually make it, and this is the category where that job is both easier and more dangerous than founders expect.
Easier, because the clusters will run 12 to 50 pieces a design for you. That is the lowest real minimum in physical D2C, and it means a first collection can cost less than one apparel size run. More dangerous, because a 30-piece order goes to a workshop with eight people and a kiln, not a factory with a QC line. You are not buying a product here. You are buying a batch. Two runs of the same vase from the same workshop will differ, and your real job as the buyer is to decide, in writing, how much difference you will accept. If the category call itself is still open, settle it first in how to start a home decor brand in India.
Six clusters cover almost everything a decor brand sells: Moradabad for brass and metal, Jodhpur and Saharanpur for wood, Khurja for ceramics, Firozabad for glass, Jaipur for block-print textiles and blue pottery, Channapatna for lacquered wood. Real minimums are 12 to 50 pieces a design at artisan and wholesale level, 100 plus at export-quality workshops, and higher wherever the unit of production is a kiln load, a furnace run or a new die. Lead times run 2 to 6 weeks and stretch badly in the two months before Diwali. Sampling is the cheapest in D2C, often ₹150 to ₹800 a piece off existing shapes, but the freight on samples costs more than the samples, so batch them one consignment per cluster. The category's core risk is not price, it is batch-to-batch consistency: colour, glaze, grain and finish drift between runs, and the only defence is a physical sealed sample plus a written tolerance. Packing is a sourcing decision here, not a logistics one, because a workshop packs for a truck to a wholesaler and not for six manual handoffs in a courier network. There is no single decor licence and no mandatory BIS for general decor, but GST moves with the HSN and with how the piece was made, so get the HSN on the quotation before you order.
The cluster map: what each place is genuinely good at
Pick by material first, not by who replies to your message fastest. Each cluster is built around one process, and it gets expensive the moment you ask it to do something else. Moradabad is Pital Nagri, the brass city, where the craft employs thousands, and Channapatna's lacquered turned wood has carried a GI tag since 2006. These are not hobby towns. They already make export-grade goods for buyers in Germany, France and the Gulf.
| Cluster | Genuinely good at | Where it disappoints | Real minimum per design | Lead time |
|---|---|---|---|---|
| Moradabad, UP | Brass, copper and aluminium: urlis, vases, planters, diyas, wall art, candle stands. The deepest metal-finishing bench in the country | Lacquer and plating quality swings hugely between job shops; raw brass price moves weekly, so quotes go stale | 25 to 50 on existing shapes; 100 to 500 at export units; a new spinning die or mould is a separate cost | 2 to 4 weeks |
| Jodhpur, Rajasthan | Solid and reclaimed wood: trays, mirror frames, shelves, boxes, small furniture. Export packing habits already in place | Seasoning. Under-dried mango or sheesham cracks in a Delhi winter three months after it ships | 12 to 50 for small decor at a workshop; 50 to 200 at export units | 3 to 6 weeks |
| Saharanpur, UP | Hand-carved wood, fretwork and brass inlay at prices no machine shop can match | Carving is hand work, so no two pieces match exactly; deep relief traps dust and photographs badly | 12 to 50 artisan; 100 plus for export-quality repeat runs | 3 to 5 weeks |
| Khurja, UP | Ceramics: vases, planters, mugs, serveware, tiles, with a wide glaze library already on the shelf | A new glaze is a kiln trial, not a sample; shade drifts from one kiln load to the next | 50 a design on existing glazes, because a kiln load is the unit; 100 to 300 for a custom glaze | 3 to 5 weeks |
| Firozabad, UP | Glass: vases, tealight holders, bottles, hanging lights, beads. Real furnace capacity | The furnace sets the minimum, not your order. A custom mould is real money, and this is the most fragile thing you will ever ship | 100 to 500 on existing moulds; higher with a new mould; small lots only through wholesalers | 3 to 6 weeks |
| Jaipur, Rajasthan | Hand block print (Sanganer, Bagru) and blue pottery: cushion covers, runners, quilts, decorative plates | Block print bleeds and fades if washing and fixing are skipped; blue pottery is brittle and yields are low | Textiles 50 to 100 pieces or fabric by the metre on existing blocks; blue pottery 25 to 50 | 2 to 5 weeks |
| Channapatna, Karnataka | Lacquered turned wood in vegetable dye: bowls, coasters, boxes, lamp bases. Genuinely distinctive and GI-protected | Thin artisan base, so capacity is limited and festive lead times stretch; the natural-dye palette is narrow | 25 to 50 at a workshop; 100 plus through cooperatives | 3 to 6 weeks |
Read the minimum column as a process fact, not a negotiating position. In brass, wood and block print the unit of production is a person, so 25 pieces is genuinely possible and anyone insisting on 500 is quoting you as an exporter. In ceramics and glass the unit of production is a kiln load or a furnace run, which is why Khurja starts near 50 and Firozabad rarely goes under 100. You cannot negotiate a kiln into being smaller. You can negotiate a shared kiln load across three of your designs in one glaze, and that is the conversation worth having. The script for it is in MOQ negotiation with suppliers. Until a design has proved itself, order shallow across more designs rather than deep on one, for the reasons set out in inventory management for D2C brands.
One planning note that costs founders a season. Add 2 to 4 weeks to every quoted lead time for orders placed in the eight to ten weeks before Diwali. The cluster is not lying to you in August. It is quoting its normal speed and then taking on three times its normal book.
Workshop or trading house: four questions that sort it in one call
Every cluster has more traders than makers. A trading house is fine when you are buying finished stock off a shelf, and a good one carries range you would need ten workshops to match. It is expensive when you think you are getting workshop pricing on a custom run and you are actually paying a layer to forward your drawing. The vocabulary for the difference sits in white label vs private label vs OEM.
- "Who does the finishing, and in which room?" A maker names the process and the place: buffing wheel, lacquer booth, glaze dipping line, kiln number. A trader says "we have our own unit" and cannot tell you where the polishing happens.
- "Show me this same design from two different runs." This is the question that matters most in decor. A workshop that keeps reference pieces will produce them in a minute. One that has never been asked will change the subject.
- "What HSN code will you invoice this under?" A real unit answers with a number. A trader guesses. That number protects you later, for reasons in the GST section below.
- "How do you pack for a courier, not for a truck?" Almost nobody has a good answer. The ones who ask what you need are the ones to work with.
Then ask for a GST tax invoice from the very first order. Plenty of cluster counters will still do the deal in cash. It is cheaper on the day and it costs you the input credit, the paper trail behind any quality claim, and a clean answer when a marketplace audits your sourcing. Seller-side detail is in GST for ecommerce sellers in India, and the general shortlist-to-sample method is in how to find manufacturers and suppliers in India. What follows is what changes in decor specifically.
Batch consistency is the risk that actually costs you money
Ask a decor founder what went wrong and it is almost never the first order. It is the third. The first 40 vases were beautiful, the reorder came back a shade lighter, and now the photo on your listing does not match what ships. In a handmade category that drift is normal, not fraud. What is not normal is having no written definition of acceptable. Three controls fix it, and none of them costs money.
- A physical sealed sample. Not a photograph. Make or buy two identical reference pieces, seal one at the workshop signed and dated by both sides, keep the other on your shelf. Every future run gets judged against that piece. Screens lie about colour, and a WhatsApp photo is worse than useless.
- A tolerance with actual numbers in it. Colour within one shade of the sealed piece in daylight. Dimensions within 3%, because wood and clay shrink and that is physics, not sloppiness. Weight within 5% on metal, since that is where a supplier quietly saves cost. No more than 2% of the lot carrying visible glaze pinholes or lacquer runs.
- A staged dispatch. On any order above roughly 100 pieces, ask for photos of the first 20 finished pieces held against the sealed sample before the rest is made. It costs the workshop ten minutes and it is the only moment when a drifting run is still cheap to fix.
Supplier Scorecard™: score every shortlisted workshop out of 10 on five lines before you discuss price, then let price break the tie. In home decor the five are: can they name their own finishing process, can they show you two runs of the same design, will they put an HSN code on the quotation, will they pack to your spec or let you supply the packing, and what happens to an off-spec lot. A workshop that is ₹20 a piece cheaper and scores 3 on remedy costs you more than the difference on your first bad batch.
Sampling is cheap here. The freight on samples is not.
This is the category's structural advantage and most founders waste it. In skincare a sample means a formulation plus a stability study. In apparel it means a full cut-make-trim setup for one piece. In decor a sample is often just a piece off the shelf.
| What you are sampling | What a sample actually is | Rough cost | What it proves |
|---|---|---|---|
| Brass or metal, existing shape | One to three pieces at trade price | ₹150 to ₹800 a piece | Weight in the hand, buffing, lacquer coverage, whether the photo flattered it |
| Brass, new shape | A hand-raised or spun one-off, before any die is cut | ₹1,500 to ₹5,000 | Proportion, and whether the shape is makeable at your target price |
| Ceramics, existing glaze | Pieces off the shelf | ₹150 to ₹600 a piece | Glaze evenness, wall thickness, base finish, hairline cracks on a coin tap |
| Ceramics, new glaze | A kiln trial, not a piece | ₹3,000 to ₹8,000 or a minimum trial batch | Whether your colour is achievable at all, and repeatable across loads |
| Wood | A one-off from an existing pattern | ₹400 to ₹2,000 | Joinery, seasoning, finish under the stain, warp after two weeks indoors |
| Block-print textile | One to two metres printed on existing blocks | ₹300 to ₹1,200 | Colour fastness after three washes, print register, bleed at the edges |
| New print block | A carved wooden block, and one design usually needs two to four | ₹800 to ₹3,000 a block | Whether the motif holds up at production scale |
Now the part nobody plans for. The samples are cheap and the freight is not, because couriers bill on volumetric weight, length by breadth by height in centimetres divided by 5000, whenever that beats the scale reading. One ₹250 vase in a protective box can carry ₹300 to ₹500 of freight on its own. Eight vendors sending eight separate parcels is the fastest way to spend ₹3,000 learning nothing. So sample smart: one consignment per cluster, 8 to 12 mixed pieces in one box, sent once you have finished collecting. Ask each workshop to hold your pieces until the consignment is ready, and pay them for the storage if they hesitate. The full billing math is in volumetric weight and shipping costs.
Then test the sample the way a customer receives it, not the way you received it. Repack one piece in your intended retail box, hand it to a courier and have it delivered to a friend three states away. That single experiment tells you more about your packing spec than any supplier assurance, and industry packaging data puts damage in transit near 11% of unit loads across Indian ecommerce. You want to find your share of that number for the price of one vase, not across 200 customers.
Put it in writing before the advance
Seven lines on one page, signed before any money moves. Design reference with photo and code, plus the sealed-sample number. Material and finish carrying an actual spec: brass gauge, wood species and moisture band, glaze code, fabric GSM. Dimensions with a tolerance, plus or minus 3%. Colour judged against the sealed sample in daylight, with the defect ceiling stated as a percentage of the lot. Packing: who supplies the retail box and inserts, and the exact outer carton size. Quantity and delivery date, with the remedy for late or off-spec written as rework at their cost. Payment staged against a pre-dispatch photo check, never against the calendar. On a 300-piece ceramic order, a missing colour tolerance is the difference between a clean lot and 60 pieces you can only sell as seconds.
Advances here typically run 30 to 50%, and one rule keeps you safe: never let the advance exceed what the supplier has actually bought or made for you. Fifty percent on a brass order is honest, because the unit buys metal on day one and brass moves with the market. The same 50% on a block-print run where the fabric is already sitting in the workshop is you funding somebody else's working capital. Ask what the advance buys. A real supplier will tell you, and will stage the balance against a pre-dispatch photo check without a fight.
In my supply chain years the cheapest control we ever put in was a retained sample. Not a spec sheet, not an audit, a physical object in a sealed bag with two signatures on it. It works because it removes the argument. When a batch drifts, nobody debates whose memory of the colour is right. You put the two pieces side by side on a windowsill and the answer is obvious in three seconds. Decor founders skip this because the first order arrives beautiful and the relationship feels warm. The retained sample is not distrust, it is the thing that lets a warm relationship survive its first bad kiln load. Suppliers rarely resent it. They quote better against a defined standard, because a vague order is a risk they have to pad. A retained sample costs nothing and settles every dispute. On a 50-piece artisan run worth ₹11,250 the sealed object in the bag is the only version of the truth both sides accept.
Packing is a sourcing decision, not a logistics one
This is what separates decor from every other category. Your workshop packs for a truck to a wholesaler: newspaper, jute, a loose carton, thirty pieces together. Your customer's parcel goes through a courier hub with five or six manual handoffs and a sorter belt. Those are different problems, and the second one is the reason breakage, not returns, is this category's loss driver. Decide at sourcing time who packs the retail unit.
- The workshop packs it, to your spec, at your cost. Realistic in Jodhpur and at export units in Moradabad and Khurja, because they already pack for overseas buyers. Send them your box spec and your inserts, and inspect the first shipment obsessively.
- You pack it. The workshop ships bulk, you repack in your own space. Cheaper per unit, slower, and it needs a table and a person. Most brands under ₹3 lakh a month do this, and should.
- A 3PL packs it. Only works if you supply the exact insert and box and audit it monthly, because a fulfilment picker will not hand-wrap your ceramic on a busy day.
Budget 8 to 15% of the selling price for fragile packing. On a ₹999 vase that is ₹80 to ₹150 of double-wall box, corrugated or foam insert, wrap and void fill. That is not a rounding error and it is not optional. It is what buys the damage rate down. One thing to specify tightly: outer box size. A box 4 cm larger on every side is billed weight you pay on every order for the life of that SKU, because the courier charges for the air. What actually survives an Indian courier network is in ecommerce shipping and packaging protection, and the brand side of the box is in product packaging design.
The QC that catches what actually comes back
Inspect at receipt, not at dispatch, and photograph every defect against the sealed sample the same day. A claim raised on day one with photographs is a rework. The same claim on day twenty is a discount request you will probably lose. Take 10% of the lot, or the square root of it if the lot is large, and check four things.
- Finish. On metal: lacquer runs, buffing swirl, and unlacquered patches that will go green in a Mumbai monsoon. On wood: sanding marks showing through the stain, and whether the underside was finished at all.
- Glaze. Run a fingernail across the surface for pinholes and crawling. Tap the piece with a coin. A clear ring means intact, a dull thud means a hairline crack you cannot see. Check the unglazed foot for grit that will scratch a customer's table.
- Joinery. Test wood joints under thumb pressure, not eyesight. Anything with a hollow base, a screwed-on stand or a glued handle gets loaded to twice its intended use before you accept the lot.
- Structural integrity as shipped. Take one piece, pack it exactly as a customer would receive it, and drop it a metre onto a hard floor. If it cracks, your packing spec is wrong, and you found out for the price of one vase.
Paying the advance against a photograph instead of a sealed sample. A founder orders 300 ceramic pieces from one Khurja workshop for a festive drop, 120 vases, 100 planters and 80 mugs, at ₹210 average ex-factory. That is ₹63,000 of goods, and she pays 60% up front, ₹37,800. The purchase order says "sky blue matte, same as sample photo". The lot arrives across three kiln loads. The last load is two shades lighter and about 8% of the pieces carry glaze pinholes. The workshop points out, correctly, that a photo is not a spec and that kiln variation is normal. She has already paid 60% and needs stock for the drop. So 24 pinholed pieces get scrapped and 60 off-shade pieces get listed as seconds at 40% off. At a ₹649 average retail the 24 scrapped pieces bin ₹5,040 of stock at cost and kill ₹15,576 of revenue she can never book, and the 60 seconds give away ₹260 of margin each, another ₹15,600. Call it ₹31,000 of revenue and margin gone from a lot that cost ₹63,000 to buy, plus the drop date. Two sealed reference pieces and one line of colour tolerance would have cost nothing.
GST moves with the HSN, and with how the piece was made
There is no single home decor GST rate, and any page that gives you one is guessing. In the September 2025 rate rationalisation a long list of handcrafted lines came down from 12% to 5%, and the government's own briefing describes ceramics, glassware and handicrafts moving to the 5% slab with effect from 22 September 2025. Other decor lines sit at 18%. Decorative and non-LED lighting is the common example.
The trap is that the concession attaches to how a thing is made, not only to what it is. GST law carries an actual definition of handicraft goods, set out in the explanation to Notification 21/2018-Central Tax (Rate): goods predominantly made by hand even where some tools or machinery are used, carrying ornamentation or similar work of a substantial nature, with aesthetic, artistic, ethnic or cultural features that make them amply different from mechanically produced goods of similar utility. Two vases of the same shape from the same town, one thrown and hand-painted, one slip-cast and machine-decorated, can land in different slabs.
So do three things. Get the HSN code onto the quotation, in writing, before you order. Keep the supplier's tax invoice showing that HSN. Have your CA confirm the rate for your exact product after GST 2.0 rather than copying a number off somebody's listing. Classification is one of the most common triggers for a GST notice in this sector, and your defence is the invoice trail, plus an advance ruling where the call is genuinely borderline.
Certification: almost nothing, until it plugs in
No mandatory BIS standard governs general home decor. Vases, trays, wall art, cushion covers and wooden pieces need no licence and no ISI mark, which makes this one of the lightest compliance categories in D2C. Two exceptions matter at sourcing time. The moment a product plugs in you are in a different regime, because LED lamps and luminaires are notified under the BIS Compulsory Registration Scheme, so adding a decorative lamp to the collection is a compliance decision and not just another SKU. And anything you market as a toy for under-14s falls under the Toys QCO, which can catch Channapatna-style pieces depending on how you describe them. Everything sold as a packaged retail unit still carries Rule 6 of the Legal Metrology (Packaged Commodities) Rules, 2011, on the pack and on the listing: name and address of the manufacturer and packer, the common name of the article, net quantity or piece count, month and year of packing, MRP inclusive of all taxes, and a consumer complaint contact. If a Khurja workshop makes it and you pack it in Gurgaon, both names go on the label. Sort that out with the workshop before the first order, not after a marketplace flags the listing.
Execution checklist
- Pick the cluster by material before you contact anyone. Quotes across materials are not comparable.
- Run the four questions on the phone: who does the finishing and where, show me two runs of one design, what HSN will you invoice, how do you pack for a courier.
- Get the HSN code and a GST tax invoice from the first order, including at cluster counters where cash is normal.
- Make two identical reference pieces, seal one at the workshop with both signatures, keep the other on your shelf.
- Write the tolerance in numbers: colour against the sealed piece in daylight, dimensions within 3%, metal weight within 5%, defect ceiling 2% of the lot.
- Batch your sampling: one consignment per cluster, 8 to 12 pieces, not one parcel per vendor.
- Decide who packs the retail unit before you order, and budget 8 to 15% of selling price for fragile packing.
- Specify the outer box size, not just the contents. Every spare centimetre is billed weight for the life of the SKU.
- Stage the advance at 30 to 50% against what the supplier has actually bought, and tie the balance to a pre-dispatch photo check.
- Inspect 10% at receipt on finish, glaze, joinery and a one-metre drop of a fully packed piece, and photograph defects the same day.
- Add 2 to 4 weeks to every quoted lead time for orders placed in the ten weeks before Diwali.
What your first decor run actually costs
A 50-piece artisan run of a ₹699 item at a 32% landed cost, the low end of the workshop MOQ band.
| Stage | What it costs |
|---|---|
| Sampling and finish approval | ₹2,000 to ₹6,000 |
| Production, 50 pieces at ₹225 | ₹11,250 |
| Fragile-goods packaging at ₹85 a piece | ₹4,250 |
| Freight, volumetric rather than dead weight | ₹3,500 |
| Total cash out | ₹25,000 |
₹25,000 for a 50-piece test, against ₹34,950 if it clears at ₹699. Decor is the cheapest category on this site to test and the most expensive to ship, so the packaging line is not optional.
Your next action
Today, do the thing that costs nothing and changes every conversation after it. Write your one-page purchase order template, the seven lines above, and send it as your opening message instead of asking "what is your MOQ". Send the same one-pager to five workshops in one cluster: same reference photo, same quantity at two levels, same four questions. Inside 48 hours you will have five comparable quotes, and the two who came back with an HSN code and a real answer on packing will have sorted themselves from the rest. That comparison, not a search result, is how you find your maker. The sourcing frameworks referenced here come from Ravikant Tyagi's operating system for founders doing exactly this.
If you'd like the complete execution system, calculators, SOPs, templates and operating frameworks behind this process, continue inside D2C Acquisition.Lab.
