Planning guide

Where to start setting out a tiled floor (and why the corner is usually wrong)

Set out floor tile from measured reference lines, avoid fragile edge slivers, and verify the start point with a worked 3200 × 4100 mm example.

Published 2 September 2026

The first tile does not have to sit where the calculation starts. That distinction is the easiest way to make sense of floor set-out.

On site, you establish two square reference lines, dry-lay enough tiles to verify the pattern, and then begin fixing tiles in an area you can physically work from. The mathematical start is the origin of the repeating tile-and-joint grid. It determines where every grout line and perimeter cut will land, even if the first tile you bond is in the middle of the room.

Starting with a full tile in a corner feels efficient. It also gives one wall complete control over the layout. If the opposite wall lands 12 mm into the next tile, you have designed a fragile 12 mm strip that must be cut and installed for the full run. The room may be large and the tile order may be correct; the visible result is still poor.

The set-out decision in one sentence

Choose the grid position that gives the widest possible worst edge cut across the complete pattern, then transfer that position to square reference lines.

“Complete pattern” matters. In straight lay, each row repeats at the same horizontal position. In half bond, alternate rows move half a tile pitch. Third bond has three row positions. A centred first row can look excellent while the next row leaves a sliver at the opposite wall.

The pitch is the tile plus its grout joint. A 300 mm tile with a 3 mm joint repeats every 303 mm. Do not lay out ten tiles as 3000 mm: nine internal joints add 27 mm before you even consider the perimeter.

A worked 3200 × 4100 mm room

Take a rectangular room measuring 3200 × 4100 mm. The tile is an actual 300 × 600 mm, the joint is 3 mm, and the pattern is half bond with the 300 mm side running across the room.

The pitches are:

• Across: 300 + 3 = 303 mm
• Down the room: 600 + 3 = 603 mm

At a corner origin, the even rows begin with a full tile. The odd rows shift by half the 303 mm pitch, or 151.5 mm. At the far side, one of those shifted rows ends at 3181.5 mm. The remaining piece is:

3200 − 3181.5 = 18.5 mm

That is the number area calculators miss. Dividing 13.12 m² by 0.18 m² per nominal tile can estimate an order, but it cannot say that one row finishes with an 18.5 mm strip.

A two-axis layout search finds a better lattice origin at 238 mm across and 544 mm down. The alternate horizontal edge pieces become 235 mm and 83.5 mm. The top and bottom pieces are 541 mm deep. The worst piece anywhere on the perimeter is therefore 83.5 mm.

The optimised layout draws 81 pieces: 47 full and 34 cut. The larger number of cut pieces is not a defect. The goal is a buildable, balanced perimeter, not the largest possible count of untouched tiles. A layout that “saves cuts” by creating ten identical 20 mm strips has optimised the wrong thing.

You can reproduce this case in the free tile layout planner — that link carries these exact measurements, so the drawing and the before-and-after figure appear with no setup. Switch the start position under the drawing between At the wall and Balanced to see the 18.5 mm and the 83.5 mm for yourself.

How to transfer the calculation to the floor

1. Measure the finished boundaries

Measure where the tile will actually finish, not from framing that will later receive lining or trim. Check width and length in at least two places. If opposite walls are not parallel, note both measurements. A rectangular planner cannot model the taper, so use the smaller reliable dimension for the first risk check and expect the cut to vary along the wall.

Also check squareness with diagonals or a 3-4-5 triangle. A correct offset measured from a wall that is not square will reproduce that error across every grout line.

2. Measure an actual tile

“300 × 600” is often a nominal format. A real tile might be slightly smaller, and manufacturing calibre can vary between batches. Ten rows magnify a 1 mm assumption into 10 mm before joint variation. Measure several tiles from the delivered batch and use a representative dimension.

For pressed or intentionally irregular edges, the specified joint is part of the system. Check the manufacturer’s minimum joint and any restriction on running-bond offset. Long rectangular tiles may restrict half bond because curvature can put a high edge beside the low centre of the neighbouring tile, increasing lippage.

3. Mark two independent reference lines

Measure the calculated X offset from a dependable baseline and snap the first line. Set a second line square to it for the Y offset. Do not use a wall as both the measurement reference and the squareness reference unless you have verified it.

The crosshair on the printable Cutfall plan marks the lattice origin. Depending on the chosen pattern, that point may be inside a tile, on an extension of a grout line, or offset from the first physical tile you intend to fix. That is fine: the reference grid is what matters.

4. Dry-lay the risky runs

Dry-lay one complete row with spacers toward the wall that receives the smallest calculated cut. For half bond, verify both row parities. For third bond, verify all three. You do not need to cover the floor; you need enough real tile and real joints to confirm the repeating pitch.

Compare the measured last piece with the plan. A few millimetres of difference can come from tile calibre, spacer compression, an out-of-square wall, or a measurement taken above floor level. Resolve the cause before mixing adhesive.

What counts as too narrow?

There is no universal code dimension for a “sliver.” Material, cutting method, edge exposure and installer skill all matter. Cutfall uses a planning threshold equal to the lesser of one-third of the tile’s short side or 50 mm. That is a warning, not a claim that a 51 mm piece is always good.

Below about 15 mm, the warning becomes stronger. A strip that narrow is hard to support during cutting and gives very little surface for bedding and adjustment. If the geometry cannot avoid it, revisit the pattern direction, use a border, adjust the joint only within the manufacturer’s permitted range, or discuss how the room’s focal and hidden edges should trade off.

Doorways deserve judgment. A narrow cut under a future cabinet is different from the same cut running through the main threshold. The numeric optimum is an excellent default because it prevents hidden surprises, but a real plan can deliberately favour a visible edge once the consequence is known.

Common set-out mistakes

• Centred grout line versus centred tile. These are different layouts. Compare both rather than using the word “centre” without specifying which one.
• Ignoring the joint in the arithmetic. The error accumulates once per repeat.
• Checking only the first row of a bond pattern. Alternate rows can own the true minimum cut.
• Using nominal tile size. The box description is not a site measurement.
• Optimising full-tile count. Fewer cuts can still mean one unusable cut repeated many times.
• Trusting a rectangular drawing in a tapered room. Verify both ends and dry-lay the narrow side.
• Ordering exactly the pieces drawn. Breakage, selection, future repair stock and box rounding still need an allowance.

The final check before adhesive

You should be able to answer four questions from the floor markings alone:

1. Where are the two square reference lines?
2. What is the smallest expected perimeter piece for every row position in the pattern?
3. Which wall or threshold is deliberately being favoured, if any?
4. Does a dry-laid run agree with the calculated pitch?

If any answer is uncertain, the layout is not ready. Moving a chalk line is free. Discovering the same error after five rows are bonded is not.