The TR measurement is construction's most important safety metric. Learn to do it systematically and document the results correctly.
TR measurement is often thought of as a safety inspection where faults are hunted down. It is however primarily a measure, and the value of a measure lies in producing a comparable figure.
The idea of the method is simple. The site is walked through and every observation is marked either correct or incorrect. No scale, no interpretation, no severity grading. The ratio of observations produces an index that is comparable from week to week and from site to site.
That forced binary is the strength of the method. With two options, the measurement does not depend on the measurer's day. With a five point scale, two people give the same scaffold different values and the figures cannot be compared.
Observations are usually divided into the same main categories: working practice, scaffolds and walkways, machines and equipment, fall protection, electricity and lighting, order and waste management, and dust.
The categories are not an end in themselves, but they serve two purposes. They make sure the round does not focus on whatever happens to be conspicuous, and they show afterwards where problems recur.
The second is the one most often left unused. A single index tells you the level. The distribution across categories tells you the cause. If fall protection produces an entry week after week, the issue is not carelessness but that something in the working method or in the availability of equipment produces the same gap again.
A common mistake is to turn the index into a target against which success is measured. That leads to a predictable outcome: the measurement starts producing better figures without the site changing.
There are many ways to get there and none requires dishonesty. The round is timed after the clean-up. A difficult area is skipped because access was poor just then. A borderline case is read as correct, because it will be fixed tomorrow anyway.
The value of the index is in the direction, not the level. The question is not whether the figure is 92 or 88 but whether it is the same figure as a month ago and whether the same categories keep producing entries.
For this reason the index should not be tied to rewards. A measure that is rewarded stops measuring.
A measurement that leads to no correction is wasted time. Yet this is where the chain most often breaks.
The observation is recorded, the round ends, the result is written down, and the correction depends on somebody remembering. On the next round the same gap is there.
The chain is closed by three things. They can be done on paper, but on paper they also break most easily, because nobody can see what is still open.
The observation has an owner. Not "the site" but a name.
The observation has a deadline. Without one, correction happens before the next round, meaning a week from now regardless of severity.
The correction is signed off. Otherwise you do not know whether it was fixed or whether the work stage simply ended.
If one of the three has to be chosen, it is the first. An observation without a name belongs to everyone and therefore to no one.
In FiSAS TR measurement, marking an item wrong opens an observation form with a photo and a location on the floor plan, and the person responsible gets an email notification about the observation to fix. The TR index is calculated during the round, and the report is available as a PDF. TR measurement is included in the Master plan.
A weekly maintenance inspection is required on a construction site, and TR measurement is the most common way of carrying it out. In practice the round is skipped in exactly the weeks when it would matter most.
The reason is familiar. In a busy week the round is the thing that can be moved, because nothing immediate follows from not doing it. A busy week is however also the one where scaffolds are moved quickly, waste piles up and working methods are cut short.
Two things help in practice.
The round has to be short. If it takes an hour, it gets postponed. If it takes fifteen minutes, it gets done.
It has to happen while walking. The entry is made where the observation is, not afterwards in the hut. A round recorded afterwards loses the precision that makes the method work in the first place.
In a safety observation the wording is always an interpretation. "Railing inadequate" means different things to different people.
A photograph shows what was wrong, and whoever fixes it does not have to guess what was meant. It also shortens the repair time, because the person fixing it knows what they are looking for before they get there.
The photograph has a second function. If something happens on the site, a documented observation history shows what was known, when, and what was done about it. In an accident investigation that is an entirely different thing from a recollection that the matter had been noticed.
On a shared site most of the work is often done by people other than the principal contractor's crew. The observations therefore concern people who are not under your direction.
Three practical approaches that work better than reminders.
Observations are gone through in the contractors' meeting rather than emailed. An observation raised face to face gets fixed more often.
The same measure applies to everyone. If your own crew's gaps are recorded more leniently, the measure loses credibility, and the willingness to fix things goes with it.
A recurring gap is raised to contract level. When the same contractor produces the same observation for the fourth time, it is no longer an isolated fault but a working method.
A single measurement says little. A three month series says a lot, and three things are worth reading from it.
Do the same categories recur. If they do, the problem is structural: equipment, method or induction.
Does the level change with the work stage. Demolition and frame work produce different observations from interior finishing. If that difference is absent, the measurement is not capturing the difference between stages.
Are observations fixed before the next round. This is the health indicator of the whole process. If the repair time is growing, the process is fading even if the index looks unchanged.
The common assumption is that safety costs time. In practice the observation categories often say otherwise.
Order and waste management are among the most common categories, and they are directly productivity. A site where materials are in their places and routes are clear is not only safer but faster. The time spent searching and moving things does not show in any measure, but it is real.
The same goes for scaffolds. A proper scaffold is safe and working on it is faster than on a poor one.
This is a useful argument when safety has to be explained to the crew. Appealing to rules works badly. Appealing to the hour a day lost to searching works better.
These are worth separating, because the obligation and the method are constantly confused.
The obligation: a weekly maintenance inspection has to be carried out on a construction site. It covers among other things the condition of the site and the scaffolds, access routes, machinery and fall protection. The obligation applies to sites where construction work is done, and the responsibility sits with the principal contractor.
The method: TR measurement is a way of carrying out that inspection so that a comparable figure remains. The method is not compulsory, but it is established, and its advantage is precisely that it measures.
The difference shows in what is left behind. A bare statutory inspection can be done by walking the site and fixing what is seen. That satisfies the obligation but produces nothing that can be compared with last week.
TR measurement produces a figure, and the figure produces two things a plain round does not: a trend over time, and the ability to compare sites with each other. The second is what turns the method into a management tool rather than mere compliance.
Outside building construction, an equivalent method is used in civil engineering. The principle is the same, the observation categories differ.
Descriptions of the method tend to stay abstract. This is how a round goes when it works.
Before the round nothing is prepared. If the site is cleaned for the round, the measurement measures cleaning rather than working practice.
On the round the whole site is walked, not only the easy part. Observations are recorded as they are seen, each either correct or incorrect. A borderline case is marked incorrect: that is the rule of the method, and it keeps the measure strict.
An observation is not one object but one matter at one place. The same scaffold can produce a correct mark for access and an incorrect one for railings.
For a deviating observation a photograph is taken and an owner assigned. This takes ten seconds and is the most productive part of the whole round.
After the round the observations are gone through with the people they concern. With subcontractors on site that means face to face at the next site meeting, not an email attachment.
On a small site the whole thing takes fifteen minutes. If it takes an hour, something is wrong: either recording is slow or the round is being used to solve observations rather than record them.
The index is the share of correct observations out of all observations. If a round produces 100 observations and 85 are correct, the index is 85.
This is worth understanding, because it explains two things that confuse people.
The number of observations varies with the site. A large site produces more observations than a small one, yet the index remains comparable, because it is a ratio. This is the most important property of the method.
One serious gap and one minor one weigh the same. The method has no severity grading. A missing railing and a waste bin in the wrong place are each one incorrect mark.
That sounds like a weakness and is a deliberate choice. Severity grading would bring interpretation back, and removing interpretation is exactly what makes the figures comparable.
The practical consequence: the index does not say whether a site is dangerous. It says how large a share of things is in order. A serious gap is addressed immediately regardless of what the figure says, and that decision is not made on the basis of a measure.
On Finnish building sites a level around 90 or above is regarded as good. Your own baseline is better taken from what the first rounds show than from what is measured elsewhere.
The method was developed for large sites, and small construction companies often ask whether it is any use on a three person job.
The answer is that the benefit differs from a large site but is not smaller.
On a large site the measurement produces information nobody would otherwise see. The site is too large for one person to observe.
On a small site everyone sees everything. The value of the measurement is not in noticing but in the observation being recorded and fixed rather than passed over. On a three person site a missing railing is certainly noticed, and that is exactly why it is easy to ignore: everyone knows about it, so nobody does anything about it.
A recorded observation with an owner breaks that chain.
The second benefit is external. When a client or a supervisor asks about safety, a documented series of measurements is a different thing from an assurance that matters are in order. In a small company this can be the differentiator in a tender where everything else is close.
A recurring observation is the most valuable output of the method, and it is usually left unused.
A single gap is an event. The same gap four times running is a system producing it. The second is not fixed by reminding, because reminding has already been tried three times.
Three common causes for a recurring observation, each with a different remedy.
There is not enough equipment. If there are railings for ten metres and a need for twenty, the gap moves from place to place from round to round. The remedy is a purchase, not a policy.
The working method produces it. If waste always accumulates in the same corner, the issue is where the work is done and where the bin was placed. The remedy is an arrangement.
Induction does not cover it. If a new subcontractor produces the same observation in their first week every time, the induction is missing a point.
So it is worth reading the distribution across categories rather than only the index. The index tells you the level, the distribution tells you where to intervene.
TR measurement works because it forces a simple question and produces a comparable figure. Its value is not in the figure but in what happens to the observation behind it.
Three things decide: the round happens even in a busy week, the observation is recorded where it is rather than afterwards, and every observation has a name and a deadline.
If you want to know whether your own practice is alive or nominal, look at last month's observations and count how many have been signed off as fixed. That number says more than the index.
