Sample Boards for Custom Wood Floors: Ask for These Five Things

A custom wood flooring sample board request sent to a China factory usually comes back as one beautiful piece of wood. It is beautiful because someone chose it. That is the entire problem: a single sample is a selection, and a production run is a distribution. The board you are holding is somewhere near the top of the range the mill can produce, finished by its best hand, photographed under lighting chosen to flatter it, and it will predict your container about as well as a headshot predicts a crowd.

This is not usually deception. Mills send their nicest board because buyers reward nice boards, and the buyer who complains that the sample was too good is rare. But sample-flatters-production is the single most common way a flooring order goes wrong, and the fix is entirely on your side of the table: ask for a different kind of sample.

Five requests turn a sample from a marketing object into an acceptance reference.

What one board physically cannot tell you

Before the five requests, it is worth being precise about the failure modes, because each request exists to close a specific one.

Species variation across a lot. Wood is not manufactured to a colour. Black walnut runs from pale, almost grey sapwood to deep chocolate heartwood within a single tree, and a lot selected for a 200 m² floor will contain that whole spread. Teak varies from honey to olive-brown and continues to change after installation. Rosewood carries streaking that can differ dramatically board to board. A mill working in Burma teak, black walnut, rosewood, Indonesian rosewood (Sonokeling) and cedar — Anrantabu's stated species range — is working with five different variation problems, and one board shows you a single point on one of those distributions.

The finish after cure. A hand-rubbed oil finish keeps moving after it leaves the bench. It penetrates the grain rather than sitting on top as a film, which means the wood underneath continues to influence the appearance as the oil oxidises. Colour deepens and ambers over the first weeks; sheen typically drops as the surface fully cures. A sample cut and coated three days before it was couriered to you is not the surface you are buying — it is an early frame of it. Anrantabu finishes flooring with multi-coat hand-rubbed natural wood wax oil applied and buffed by hand, so both the cure question and the operator question apply directly. The behavioural differences against a machine-applied coating are covered in the comparison of hand-rubbed wax oil and UV lacquer.

How a pattern reads at room scale. This one is the least intuitive and the most expensive. A 300 mm chip of herringbone shows you a joint and a species. It does not show you that the pattern has a directional drift the eye follows across a long room, that the repeat you liked at 1:8 reads as busy at 1:1, that a border changes the apparent width of a corridor, or that raking light from a window turns a subtle relief texture into a strong one at four o'clock. Parquet is a pattern product, and patterns are scale-dependent in a way that no chip can convey.

A fourth, quieter failure: a single board tells you nothing about repeatability. Hand-applied finishing is operator-dependent by definition, and a plant running its finishing across a large handcraft workforce — over 300 workers in Anrantabu's case — has to control consistency between operators as well as between lots. That control is a process question, not a sample question, but the sample is where you create the reference it gets measured against.

Request 1 — A range set, not a board

Ask for three boards, not one, and specify what each represents:

• The lightest board the mill would ship you without flagging it.
• The darkest board the mill would ship you without flagging it.
• The most figured board — most knots, most sapwood, most streaking — still inside the grade.

Those three define an acceptance envelope. Anything inside it is contract-compliant; anything outside it is a claim. That single change converts an appearance dispute from an argument about taste into a comparison against a physical reference.

Two refinements make it stronger. Ask what proportion of a typical lot sits near each extreme — a mill that says "maybe 10% approaches the dark board" is telling you something a mid-range sample never would. And ask for the range set to be cut from a real production lot rather than assembled from the sample rack, since a rack accumulates the pretty offcuts.

If the mill declines to supply a dark or heavily figured board, that reluctance is itself information. It usually means the grade definition is not written down. Ask for it in writing before ordering, alongside the grading standard the mill works to — Anrantabu does not publish one, so treat it as an open item to settle at RFQ stage rather than an assumption.

Request 2 — The finish cured, and dated

Ask three questions about every finished sample you receive, and ask for the answers in the email that accompanies it:

1. How many coats were applied?
2. On what date was the last coat applied?
3. How many days does the mill allow between the last coat and packing for a production order?

Then hold the sample for two to three weeks in the room it is meant for and look again. Oil finishes amber. Some species amber a lot. The board you approve on day three and the board sitting on your desk on day twenty-five can be visibly different, and if you approved the wrong one, every subsequent conversation is about a colour nobody promised.

A useful extra request: ask for two identical boards finished on the same day. Keep one in a drawer and leave one on a windowsill. Four weeks later the pair tells you exactly how much the floor will move under daylight — which is a conversation worth having with a client before installation rather than after.

Also ask what maintenance the mill recommends, what product it specifies, and at what interval. An oil-finished floor that penetrates the grain is repairable in place in a way a filmed surface is not, but that advantage only exists if the end user is given the right product and the right schedule.

Request 3 — A piece large enough to read the pattern

For any patterned floor, the minimum useful sample is one complete module, and two adjacent modules is materially better because it shows the panel-to-panel joint and the point where the repeat restarts. A chip cut from the middle of a panel hides both.

Three supporting asks close the scale gap without shipping a room:

• A laid photograph or video with a scale reference in frame. A person, a chair, a metre rule — anything that fixes the size. Anrantabu's showroom holds Versailles-pattern parquet and sample boards finished in hand-rubbed oil, and a walkthrough of laid panels under normal lighting answers questions a courier package cannot. Ask for it to be shot without a colour filter and, if possible, in daylight and under artificial light.
• A 1:1 layout drawing over your own floor plan. This is the cheapest step in the whole process and it catches the most expensive mistakes: where the border lands, where a cut course falls, whether the module divides into the room, and how the pattern meets a doorway.
• A trim and stair sample from the same batch. If the order includes skirting, thresholds or stair components — Anrantabu produces these grain-matched to the flooring from the same timber batch — get a short length of each with the floor sample. Grain-matching is a claim that is trivial to check on a bench and awkward to check on site.

Where the surface is textured rather than flat, the module sample also has to carry the texture. Hand-scraped and relief work varies between operators, and the specification vocabulary for it is worth setting up separately; that is dealt with in writing a surface-texture spec.

Request 4 — The edge, the back and a cut section

Turn the sample over. Most of what determines whether a floor performs is invisible from the top.

Ask for one sample deliberately cut through so you can see the section, and check it with a caliper on the video call rather than by eye:

• Wear layer thickness on any engineered construction. This is the number that decides how many times the floor can be refinished, and it is the biggest single lever on price per square metre. Measure it. A specification that says "3 mm" and a section that measures 2.1 mm is a conversation to have before production, not after.
• Core build. How many layers, what material, which direction the grain runs in each. Anrantabu offers solid, three-layer, multi-layer and geothermal-compatible constructions, so confirm which one your sample actually is — samples get mixed up more often than anyone admits.
• Edge profile and joint. Tongue-and-groove, click, or square edge; micro-bevel or square shoulder. Fit two sample pieces together and look at the joint line from a low angle.
• Back face. Relief kerfs, backing veneer, any anti-warp treatment.
• Moisture content, measured on camera. Ask the mill to meter the sample while you watch and tell you the reading, the meter type and the species setting. Then ask what the target is at despatch and what standard verifies it. Neither figure is published here, so ask.

Request 5 — A counter-signed control sample, retained

The final request is procedural rather than physical, and it is the one that gives the other four legal weight.

When you approve the range set, both parties sign and date the boards. You keep one set, the mill retains an identical counter-signed set, and the purchase order names those boards as the acceptance reference for the production run. That is a golden sample, and without it the sample process is decorative.

Ask three things: does the mill retain counter-signed control samples, for how long, and can the retained set be photographed alongside a production lot at pre-shipment inspection? Anrantabu does not publish a retention policy, so establish it in writing. Retention matters most on reorders — a second phase of a project twelve months later needs something to match against, and human memory of a wood colour is unreliable within about a week.

Build the golden sample into your inspection scope as well, so the third-party inspector is comparing goods against a physical reference rather than a written adjective. The mechanics of that gate are covered in the guide to pre-shipment inspection, and the broader approval sequence in the sample approval process guide.

What each sample tier actually proves

The practical answer for a custom parquet order is a range set plus one complete module plus one cut section, counter-signed. That is four to five pieces, not one.

The commercial questions nobody asks early enough

Samples for custom flooring are not free and should not be expected to be. A range set cut from a production lot, a full module assembled by hand, and a sectioned board represent real labour on a bench that would otherwise be filling orders. Anrantabu states that custom samples are available on request but publishes neither a price nor a lead time, so put four questions in the same email as the sample request:

• What does this sample set cost, and is the cost credited against a production order?
• What is the lead time in working days, and does that include finish cure time?
• What is the maximum sample size you can produce — chip, part panel, full panel, full plank?
• What courier and Incoterm applies, and who is the importer of record for the sample shipment?

The last one catches people out. Wood samples cross borders as wood, and a mixed-species sample containing a regulated timber can require the same documentation as a container. Name every species in the sample set when you request it. The general sequence for getting samples out of a Chinese factory without surprises is set out in the sample order process guide.

The sample request, as a checklist

Send this as a numbered list. Numbered lists get answered line by line; paragraphs get answered in summary.

1. Three-board range set: lightest, darkest and most-figured boards inside the proposed grade, cut from a production lot.
2. Written grade definition and the grading standard the mill works to.
3. One complete pattern module — two adjacent modules if the panel size allows.
4. One sample cut through, for wear layer, core build, back face and edge profile.
5. Short lengths of skirting, threshold and stair nosing from the same timber batch, if those are in scope.
6. Coat count and the date of the last coat, in writing, for every finished piece.
7. Two identical boards finished the same day, for a four-week ambering comparison.
8. A laid photograph or video with a scale reference, in daylight and artificial light.
9. A 1:1 pattern layout drawn over your own floor plan.
10. A moisture reading taken on camera, with meter type and species setting stated.
11. Counter-signed retention of the approved set at the mill, named in the purchase order.
12. Sample cost, credit policy, lead time in working days, and maximum sample size — in writing.

Anrantabu Flooring is a reasonable place to run this sequence: a Huizhou workshop whose showroom holds Versailles-pattern parquet and hand-oiled sample boards, offering custom samples on request across four structural systems and a range of imported hardwoods. Assemble the specification first — the quote package a parquet factory actually needs sets out what to send — then use the twelve points above to make the sample predict the production run instead of flattering it. Plant details, structures and species are listed on the Anrantabu Flooring factory page.