Who does what
Almost nothing goes wrong in the middle of a trade's scope. It goes wrong at the edges, where one trade stops and the next is supposed to start.
A speaker that nobody cut out for. A cable pulled to the wrong side of a wall. A lighting panel with no space left in the switchboard. A gate intercom that was everyone's job and therefore nobody's.
None of these are hard problems. Each one costs a week.
They have the same cause, and it is not incompetence. It is that two trades each assumed the other had it, and no document ever said which.
The edges are where the money goes
A builder running a significant home is coordinating a dozen trades against a programme. Each one knows its own scope well. What none of them owns is the boundary.
Take a single in-ceiling speaker. Somebody specifies it. Somebody supplies it. Somebody runs the cable to it at rough-in. Somebody cuts the hole once the ceiling is on. Somebody terminates it. Somebody tests it, and somebody signs off that the room sounds right.
That is six actions. On most sites they are assumed to belong to two or three parties, and the assumption is rarely written down. The gap usually shows up on the day the ceiling is being painted, which is the most expensive possible day to discover it.
Multiply by every keypad, every camera, every blind motor, every data point, and the aggregate is not a snag list. It is a fortnight.
What the matrix actually is
The responsibilities matrix is a table. One row per element of the technology scope, and five columns: who specifies, who supplies, who installs, who terminates, who proves it works.
It is issued with the design and documentation set, before tender, so the builder can price it and every subcontractor can see their own column. It is dull to read, which is the point. A document that is interesting usually means somebody is arguing.
Two examples of how the lines actually fall, because they fall differently depending on the discipline.
Lighting control. The panel is designed and commissioned by Baillie. The electrical contractor prewires, cuts out and fits off. The switchboard allowance, the panel location and its ventilation are settled at design and documentation. Not at second fix, with a builder standing in the room asking where it goes.
Ceiling speakers. Positions and cable type are documented. The cable is run at prewire to that schedule by the electrical contractor. Baillie cuts out, fits off and commissions, so the levels are set in the finished room rather than assumed from the plan.
Those two run in opposite directions, and both are correct. What matters is not which party does which action but that the answer exists in writing before anyone is on site.
Why it belongs to the technology architect
The matrix is not the builder's document to write, though the builder is its main beneficiary.
Writing it requires knowing what each element needs: that a lighting panel generates heat and needs a ventilation path, that a speaker set out against the structure rather than the reflected ceiling plan will be visibly out of line for the life of the house, that a camera at a gate needs a cable route nobody has drawn yet because the landscape design is not finished.
That knowledge sits with whoever designed the technology. So the matrix comes from the same desk as the drawings, the rack elevations and the cable schedules, and it goes out as one set.
What it does to a tender
This is the part that is easy to underestimate.
A builder tendering a vague technology scope receives numbers that cannot be read against each other, because each integrator has guessed at a different job. One has priced the cutting out. Another has assumed the electrician does it. A third has excluded the gate intercom entirely because the drawings were silent on it.
Those three numbers are not comparable, and the lowest one is usually the one that has quietly excluded the most. Which the builder discovers later, as variations.
A documented scope with a responsibilities matrix produces prices for the same work. The builder can compare them honestly, and the number that wins is the number that reflects the job. That is a better outcome for the builder, and in the long run a better outcome for the integrator who priced it properly.
A line with no name against it is the line that gets missed.
The clause that does the most work
Near the end of the matrix sits a line that is worth more than the rest of it: who proves it works.
Supply and installation are visible. If a speaker is not in the ceiling, everybody can see that. But commissioning is invisible until it is wrong. A camera that records but never alerts. A blind that closes but not with the lighting scene. A keypad that works except for the third button.
Naming a party against proof means somebody is contractually required to demonstrate the finished behaviour, in the room, to the person who will live with it. That is the difference between a handover and a list.
What it looks like when it is missing
A home brought to Baillie at second fix gets a system designed around decisions already made. Some of those decisions were taken by people who did not know they were making them: an electrician who pulled a cable to a convenient stud, a cabinetmaker who filled the only space the rack could have gone, a plasterer who closed a wall that needed one more conduit.
Nobody did anything wrong. There was simply no document that said otherwise.
The matrix is cheap. It is a table, written once, as part of a design phase that was going to happen anyway. Against the cost of one trade returning to site, it has paid for itself before the first page is read.