Where early figures come from
At feasibility there is nothing to measure, so the cost has to come from somewhere else. That somewhere is benchmarking: taking the outcome of comparable completed schemes and adjusting it to the scheme in front of you.
Done properly this is a legitimate and necessary technique. Done as a straight copy it is the single most common source of the gap between an appraisal and a tender return, because the previous scheme carried its own conditions and they do not travel.
The failure is rarely the choice of benchmark. It is the absence of adjustment between the benchmark and the scheme being priced.
The word covers two different activities that are worth separating. Benchmarking to test whether a figure is plausible is a sense check, and it is quick and useful at any stage. Benchmarking to produce the figure in the first place is a pricing method, and it carries every difference between the source scheme and yours unless somebody removes them.
What has to be adjusted
Six things move a rate materially. Building form, because a tall narrow block and a wide shallow one with the same floor area do not cost the same. Storey height and structural grid. Specification level, which is where most of the difference sits on residential work. Site constraints, including access, working hours and neighbouring buildings. Procurement route and the amount of risk transferred. And time, meaning inflation from the benchmark date to the anticipated tender date.
Each of those is capable of moving a rate by a meaningful percentage on its own. Applied together and unadjusted, they explain most of the surprises that arrive with tenders.
The adjustment does not have to be elaborate. It has to be written down, so that the person reading the figure later can see what was taken into account.
The date on the rate
A rate carries a date whether or not anybody records it. A figure derived from a scheme that tendered two years ago describes a market that no longer exists, and carrying it forward without an inflation basis is not conservatism, it is silence.
The correct treatment is to state the benchmark date, state the assumed tender date, and state the basis on which the movement between them has been applied. We set that out in construction inflation allowance.
Where benchmarks come from matters
Internal benchmarks from your own completed schemes are the most useful, because you know what they included. Published indices are useful for movement over time and weak for absolute level. Rates quoted informally by another developer are the least reliable of all, because you do not know their specification, their procurement route or whether the figure was out-turn or tender.
Whatever the source, the same test applies: can you say what the figure includes. If not, it is a number without a boundary, and the boundary is where the money is.
Benchmarking is not a substitute for measurement
A benchmark answers whether a scheme is roughly viable. It does not answer what it will cost. As soon as there is enough design to measure, measurement replaces benchmarking, and the benchmark becomes a sense check rather than the basis.
Developers occasionally keep benchmarking long after there is something to measure, because it is faster and because the number is already in the appraisal. That is the point at which a benchmark stops being a technique and becomes a reason not to look.
What this means for you
Ask where the rate in your appraisal came from, what it was adjusted for, and what date it carries. If those three answers are not available, the figure is a guess wearing a suit.
Where the figure is deciding what you can bid for a site, it is worth testing properly before the bid rather than after the acquisition. That work sits inside cost planning and estimating and is where a client side quantity surveyor earns most of its value.
Where you have built more than a handful of schemes, the most valuable thing you can do is keep your own out-turn costs in a form that can be reused: final account values, floor areas, specification level, procurement route and dates. Most developers hold this information across a dozen unconnected files, which is why they end up benchmarking against somebody else’s building instead of their own.