Door Hardware Refurbishment Survey 2026: The Fourteen Fields, What Needs a Screwdriver, and How to Sample a Building You Cannot Fully Open

A refurbishment order is only as good as the survey behind it, and the most expensive assumption in this trade is that the drawings describe the building. They describe what somebody intended twenty years ago. Since then doors have been rehung, locks have been replaced with whatever was in the van, and a fire door has been swapped for a plain one by somebody who did not know the difference.

So survey the doors. Not the drawings, not the old schedule, and not the previous supplier's order, the doors.

The unit is the leaf

Not the room, not the opening. A pair of doors is two leaves with two sets of hardware and frequently two different hands, and a schedule organized by room will merge them into one line and order half the hardware.

Give every leaf a permanent ID and write that ID on the leaf, in pencil on the hinge edge or on a removable label. Every later conversation (the order, the delivery, the fitting, the snag) refers to that ID. A survey whose rows cannot be matched back to a physical door has recorded information that nobody can act on.

The fourteen fields

Fields 1 to 7 and 12 to 14 come from the corridor with a caliper and a phone. Fields 8, 9 and 11 need the lock or cylinder out. That split is the whole logistics problem of a survey, and it is why sampling exists.

The sampling rule: by type, not by floor

You cannot open four hundred locks. You do not need to. But the common shortcut (sample one floor thoroughly and assume the rest) is the wrong axis, because buildings are not consistent by floor. They are consistent by procurement batch, and a batch usually maps to a door type: all the flat entrance doors, all the riser cupboards, all the office doors on the refurbished wing.

So group by type first, using the fields you can see: leaf material, lock type, forend shape, lever family, closer present. Then open a sample within each type, on more than one floor, including at least one door that looks slightly different from its group.

That last one is where the surprises are. A group of forty identical-looking doors usually contains three that were replaced separately, and the three are the ones that will not accept the hardware you ordered for the forty.

Record which doors were opened and which were inferred. A survey that does not distinguish measured rows from assumed rows will be read as if every row were measured, and the assumed rows are exactly the ones that generate the reorder.

Photographs: three per leaf, and what each must show

The leaf, whole, from the pull side, with the ID visible. This is the row's anchor and the thing that settles arguments about which door anybody meant.

The lock edge, close, showing the forend and the frame keep together. This one image carries the forend shape, the keep type and the gap, and it is the image most often missing.

The face around the furniture, showing escutcheon, rose, lever and any wear pattern, with a scale in frame, a steel rule is fine. A photograph without a scale records appearance and not dimension.

Photograph rather than describe wherever you can. "Radiused forend" is a reading of a photograph; the photograph can be re-read next month by somebody who disagrees.

Two hands on one floor, and other things that look like errors

Corridors alternate. A floor that looks symmetrical in plan usually has both hands in it, and a survey that records one hand per floor has recorded a plan rather than a building. Count both, and record the split. Our handing page covers the conventions; our universal-versus-handed page covers which products care.

Doors get rehung. A leaf turned end for end in a previous refit carries hardware in a hand that matches nothing else on that floor, and it is not a survey error, it is the correct record of an odd door. Flag it rather than correcting it to match its neighbors.

A fire door with no evidence of a rating is not a fire door for procurement purposes. Record what you can see (a label, a plug, a certificate on file) and record its absence just as explicitly. Our certificate page explains how to check a rating claim; the survey's job is only to record whether evidence exists.

The doors you cannot get into

On an occupied building there will always be a set you cannot survey: tenanted flats, secure rooms, a floor mid-lease. Do not quietly average over them.

Record them as a named, counted set with a stated assumption ("38 flat entrance doors, assumed to match the sampled type B, not surveyed") and price them as a separate line. That way the risk sits somewhere visible, and if it turns out badly it is a variation with a paper trail rather than a surprise on the day of fitting.

This is the same discipline as the dash instead of an invented dimension. An honest "not measured" costs a line of text; a plausible assumption presented as a measurement costs a reorder.

The frame is half the job, and it is the half that gets left out

The fourteen fields above describe a leaf. A lock closes into a frame, and a survey that records only leaves will discover the frame on the day of fitting.

Frame material: timber, steel or aluminum, and it decides almost everything about the keep. A keep in a timber frame can be re-cut on site with a chisel; a keep in a pressed steel frame cannot, and a keep in an aluminum section is usually a proprietary part that has to be ordered with the lock rather than after it.

The existing keep or strike: its type, its size, and whether it is being reused. A new lock frequently needs a new keep even when the backset matches, because the latch geometry moved a millimeter or two. If the keep is being reused, its opening has to be measured, not assumed.

Rebate depth and the closing gap: on a rebated pair this also tells you whether a rebate kit is needed, which is a part most schedules omit until the doors will not close past each other.

Whether the frame is square: old frames settle. A leaf that closes badly today will close worse with a stiffer new latch, and that is a carpentry item rather than a hardware item, but it belongs on the survey, because somebody has to price it and it is cheaper to price now than to discover during handover.

Turning the survey into an order

Group the finished survey by the fields that actually determine a part number, in this order: lock type, then backset, then center distance, then hand, then finish, then cylinder length by half.

The grouping usually collapses a few hundred rows into eight or twelve real line items plus a handful of one-offs, and the one-offs are worth their own conversation rather than being rounded into the nearest group. Order the groups, and order the one-offs as one-offs.

Then check the sum. The number of leaves on the survey must equal the number of hardware sets on the order. This sounds too obvious to write down, and it is the single most common arithmetic failure on a refurbishment, because pairs get counted as one and cupboard doors get forgotten entirely.

What our catalog can and cannot answer for you

A survey produces questions the supplier has to answer back. Across the 588 HYDE catalog records, here is what a published record can currently settle without anybody opening a drawing:

The last row is the useful one to know in advance. Forend dimensions are published on 6 products. If your survey turns on whether a replacement lock's forend covers the existing mortise, the catalog will not answer it for most models. The answer has to come from a confirmed drawing or a physical sample. Ask early rather than at the point of fitting.

That is a limitation stated as a count rather than an adjective, which is how every gap on this site is written. It is also, for a refurbishment, the single most useful thing we can tell you before you start measuring.