the Helm and the Lens – Volume 4b

The companion volume described what a property does. This one asks whether any of it is true.

In every other standard of this series, that question has a familiar shape. An auditor establishes whether the system conforms to its requirements and whether it is effectively implemented and maintained, and a system can pass both tests while the thing it manages stays exactly where it was. This standard adds a third test and it changes the profession. The organisation must demonstrate continual improvement of its energy performance, which means that the auditor — internal or external — must establish not only that the system works but that a number moved, and that the movement was real.

The second half of that sentence is where the difficulty lives. A hotel’s energy figure moves for reasons that have nothing to do with management: a milder winter, a fuller house, a service that left the building, a baseline that was quietly rebuilt. Every one of them produces a genuine improvement in the reported figure and none of them is an improvement in energy performance. The volume opens on exactly that — a nine per cent gain in Helsinki that turned out to be four things happening in the same year, and less than one point of management inside it — and almost everything that follows is a method for telling the two apart.

There is a second difficulty and it belongs to the building rather than to the standard. The evidence an auditor needs was never built to be audited. Meters were installed to bill, control systems to keep guests comfortable, maintenance records to show that a visit took place. None of them was designed to support a conclusion about performance, and a great deal of auditing in this discipline consists of establishing whether a number can bear the weight that is about to be placed on it.

The third chapter takes the position of the internal auditor: the programme built around the hours at which a hotel actually consumes rather than around its departments, the checklist that goes to the plant room, the near-miss nobody recorded, the interview, the supplier, the gap between the baseline and the building, root cause, calibration, access to the set points, the report that ownership will read, the follow-up, and what the audit owes the management review. At its centre is the energy performance test — how an improvement is demonstrated rather than asserted.

The fourth chapter takes the position of the certifier: the chain that makes a certificate mean something, the choice of a body, the two stages of initial certification, the competence of the team, the findings and what closes them, what a certificate actually says, the surveillance visit in the year performance fell, the mark and the claim, the combined audit, the lender who reads the certificate, the group certified across four climates, change inside the cycle, and recertification. At its centre is objective evidence of improvement — what a person who has never met anybody in the building will accept.

Those two central subchapters are the spine of the volume, and the other twenty-eight lean on them.

Who this book is for

The quality manager who has audited two other standards competently for years and is given two days for this one. The coordinator who holds the energy management system alongside another job and has to prepare for a stage two visit. The general manager who signs the management review and reads the first page of every audit report. The owner or asset manager deciding whether a certificate is worth what it costs and what may honestly be said about it. The certification auditor who learned this standard in industry, where a production line has a measurable output and a stable load, and now finds a hotel in the audit programme. And the person — there is usually one — who is asked, the week before a visit, whether the figure in the pack is actually true.

Three facts about auditing energy in a hotel govern everything in this book. The first is that the number can be entirely accurate and mean nothing. Weather, occupancy, boundary and baseline all move it, and a figure that is arithmetically correct can be the sum of four things none of which was done by anybody. Accuracy is the beginning of an auditor’s work in this discipline and not its conclusion.

The second is that the evidence was built for another purpose. The meter answers the supplier’s question, the control system answers the guest’s and the maintenance record answers the contractor’s. An auditor who takes any of them at face value is auditing the purpose they were built for, and the first question in almost every subchapter of this book is what a given record can support and what it cannot.

The third is that the auditor is never in the building when the consumption happens. Audits are conducted on weekdays in daylight. A hotel’s energy is decided at two in the morning, on changeover Saturdays and in the shoulder weeks of a season — by schedules that nobody reviews, set points changed for a guest who asked, and plant left in the position a technician found convenient. The audit that does not go looking for those hours will certify the building the day shift sees.

To these the discipline adds a fourth, and it is the rule on which the whole of the fourth chapter rests. The property demonstrates; the auditor does not disprove. An audit team that finds no evidence of an improvement is not required to establish that it did not occur, and should say precisely that — not that the measure failed, but that nothing exists by which anybody could establish that it worked. Unsupported is not untrue, and the difference is not a courtesy. It is the finding.

What this book is, and what it is not

It is a practitioner’s guide to auditing an energy management system in a hotel, from inside and from outside. It is not a certification scheme, it is not an auditor training course, and it does not replace the texts that actually bind anybody. It will not tell you whether a given property would pass. Nobody who has not seen that property’s data can tell you that, and a book that pretends to is selling something.

Where the book states what is required, it states it as a requirement, and it says whose requirement it is. That distinction matters more here than in the companion volume, because in the fourth chapter the requirements come in two families: those the energy standard places on the property, and those the framework governing certification places on the body that certifies it. Several of the most consequential obligations in this volume — the use of the mark, the duty to notify change, what a surveillance visit must cover — sit in the second family, or in the certification agreement a property signs, and appear nowhere in the standard itself. A reader who looks for them there will not find them.

Bodies, schemes and the documents of the certification framework are described throughout by what they do and never by name. This is deliberate. The names change, merge and are reissued on their own timetables, and a book that depends on them ages by the year; the functions they perform have been stable for a long time. Where a reader needs the binding text, the Sources identify which document carries which obligation.

Every case is composite. Thirty properties in thirty cities, none of them repeated from the companion volume, none of them a report of any real hotel’s history, each assembled from patterns that recur across four decades of work in international hospitality. The distribution of climates is deliberate — five cold, five hot or arid, the rest temperate — because in this standard the location changes the problem rather than the setting. Every financial figure is a modelled estimate and is declared as one wherever it appears; none of them is a sector benchmark.

One subchapter stands apart and says so. What lenders, valuers and rating analysts ask of a certified property is commercial and regulatory practice rather than a normative text, and it changes faster than any standard. The subchapter that deals with it describes the shape of the questions, which has been stable, and declines to describe any particular institution’s rules, which have not.

The fixed elements of a subchapter

Every subchapter carries the same elements. A Guiding Question opens it and is the question the subchapter exists to answer. Each one can be put to your own property or your own audit programme today, and each carries a time or a count, because a question that cannot be attempted in an afternoon will not be attempted at all.

What the Standard Says sets out the applicable requirements in the book’s own words, never in the standard’s, and in the fourth chapter it says which family each requirement belongs to. Where a subchapter’s subject has little or no normative basis, the section says that too, rather than inventing one.

The Verification Pact runs a belief commonly held in the market against what actually holds. It is the discipline the whole series is built on, and it is deliberately the shortest thing on the page: a handful of lines, no argument, no qualification.

Two tables then close the practical section. The first separates what is actually required from what is method or custom — in the third chapter, what is required against what demonstrates it and what is merely one way of doing it; in the fourth, what binds against what a property can settle for itself and what is only habit. The third column is the one to read twice, because the documentation produced in this industry against obligations that do not exist is the largest avoidable waste an auditor sees. The second table is the instrument of the subchapter: the thing a reader can take into a plant room, an audit or a meeting and use on Monday.

Every subchapter also closes its practical section with the case that runs the other way — the obvious correction that produces the opposite failure. The auditor so wary of normalisation that nothing can ever be demonstrated. The property so frightened of over-claiming that it says nothing true about a certificate it earned. The group that responds to a sampling problem by abandoning the one advantage of certifying as a group. The building locked down so tightly that it loses everything at once through a workaround. These are not rhetorical balance. Over-correction built a good part of what this book describes, and a reader who takes the argument without its inversion will build some more.

A note on words

Auditing borrows a vocabulary that the industry uses loosely, and five of its distinctions decide whole subchapters of this book.

Conformity is the fulfilment of a requirement. Performance is a measurable result. In every other standard of this series an auditor reaches a conclusion about the first and may say nothing about the second. In this one both conclusions are required, and the second cannot be inferred from the first: a system can conform in every clause and still fail to demonstrate that anything improved.

Objective evidence is data supporting the existence or truth of something, which can be verified. The last three words carry the whole of it. An improvement is demonstrated when a person who has never met anybody in the building can reach the same conclusion by the same route from what is written down, and not otherwise — whatever the seniority or the confidence of whoever is explaining it.

Opening

The surveillance visit was booked for the second week of March, and by the end of January Nuria Sempere knew what the auditor would find, because she had found it first.

The property is two hundred and forty rooms on the Costa Blanca at Alicante: four guest floors in two wings, two pools of which one is heated, a spa, and a laundry serving the hotel and a sister property nine kilometres up the coast. It had always run a season, from the last week of April to the last week of October. Nuria had been maintenance manager for eleven years and had carried the energy coordination since the system was established.

The certificate was twenty-six months old and the first surveillance visit had been clean. The figure taken to it was a normalised improvement of seven point four per cent against the baseline, won on a chiller sequencing change and a new domestic hot water plant, and it had gone into the owning company’s annual report, onto the website, and into a tender the sales office won that February. The figure for the year just closed was four point two per cent better than the baseline and three and a half per cent worse than the year before it. It was the first fall since anybody in the building had been measuring.

Nobody had panicked, and there was a respectable reason not to. During the year ownership had extended the season, opening six weeks earlier and closing three weeks later: forty-seven additional operating days at an average occupancy of thirty-eight per cent, against a main-season average of eighty-six. The plant that runs because a building is open rather than because it is full barely notices how many rooms are sold: spread across thirty-eight per cent occupancy it produces an intensity per occupied room-night roughly two and a half times the main-season figure. Forty-seven days of that, inside a two-hundred-day year, will turn a modest gain into a modest loss on its own.

The normalisation should have handled part of it and Nuria could not say how much. The model had been fitted on three seasons of main-season data, at occupancies between seventy and ninety-five per cent, in the weather of May to October. It was now being asked about days in March at a third full, at temperatures the fitting data did not contain. The consultant who built it said it would still work; he did not say on what basis, and when she plotted the fitted points and the new points on one chart, the new ones sat in a region where there was nothing at all.

He had a second proposal, and that was the one that worried her. Revise the baseline. The argument was not frivolous: the operating calendar of a property is a static factor, an extension of nine weeks is a material change to it, and the standard provides for revision where static factors change materially. He had a draft ready, and restated against the new calendar the year showed an improvement of one point one per cent rather than a fall. She did not sign it. The method had not existed before the result; it had been built to meet it. It restated a whole year rather than the part that had changed. And it would replace one series with another, leaving nothing by which the auditor, her successor or ownership could see what the figure had been before the arithmetic improved it.

So she spent three weeks separating the fall instead. Degree days for both periods. Occupancy by day rather than by month, which is where shoulder weeks stop hiding. The boundary record and the service contracts, to establish that nothing had moved in or out. The revision register, which was empty, as it should have been.

Three things came out of it that belonged to the property rather than to the weather. The heat recovery exchanger on the laundry failed in the second week of July; the laundry’s own indicator rose nineteen per cent and was flagged in week three; the work order went out the same day and the part took eleven weeks to arrive. The chillers ran the summer schedule for six weeks into the shoulder, because the schedule had been written for a season that ended in October and nobody had been asked to write another; that was found in the fourth month and corrected. And the corridor air handling on the guest floors had been running one and a half degrees colder since the second week of August, after a complaint from a tour group, and was still running colder in February when Nuria found the entry in the change log of the building management system. Nobody had opened that log since the previous audit. Seven months of it took forty minutes to read.

Her arithmetic put about one and three-tenths of the three and a half points on the property, the corridor set point the largest of the three; the rest divided between the extended shoulder and a summer the normalisation had absorbed to an extent she could not state with confidence, and she wrote that sentence in exactly those words. What she took to the visit was eleven pages: the fall on the first of them with both figures, the separation with its four documents, the three defects with dates and costs and what had been done, the change log now read monthly, and a proposal to revise the baseline forward from the coming season, by a method written out in a paragraph and dated before any result it would touch, with the unrevised series retained beside the new one for the rest of the cycle.

The visit took two days and this took two hours of them. One minor nonconformity: a deviation in guest-floor consumption that the property’s own monitoring had recorded and nobody had investigated, over seven months. One opportunity for improvement, on the validity range of the normalisation. The forward revision accepted on the terms proposed, and written down as accepted. The certificate maintained without qualification.

The auditor said one thing on the way out that Nuria has repeated to three other coordinators since. A performance standard does not require you to have a good year; it requires you to know what kind of year you had. The fall was never the finding. The seven months were.

The Problem

The first failure is believing that a fall in performance is itself a nonconformity. It is not, and the belief is expensive because of what it causes. A property convinced that a bad year costs the certificate will manage the number rather than the building: the presentation gets adjusted, the awkward months get aggregated away, the baseline acquires a revision it was not owed. All of those are findings. The fall was not.

The second failure is the mirror of the first and is commoner in well-run properties. The fall is accepted as weather and trade, described in one sentence at the management review, and nothing is separated and nothing is done. This is also a finding, and a more serious one, because the system has stopped doing the only thing that distinguishes it from a filing cabinet.

The third failure is the baseline revised after the result. Revision is provided for and is sometimes obligatory, but it has stated triggers and a method that must exist before the figure it is applied to. A revision drafted in the month before a surveillance visit, on a year that disappointed, restating a whole period rather than the part that changed, is not a technical correction. It is the audit’s principal finding wearing a spreadsheet.

The fourth failure is the significant deviation that was recorded and not investigated. This is the requirement written for precisely this situation, and it is the one most often unmet, because investigating a deviation takes a person and half a day and the number will probably come back on its own. In Alicante the number did not come back on its own for seven months.

The fifth failure is the normalisation applied outside the conditions it was built on. A regression fitted on a full-occupancy summer will produce a number for a third-full March, the number will look exactly like all the others, and nobody will know the model has never seen anything resembling that day. The failure is invisible in every direction but one: plot the fitted points and the applied points on the same chart, and it is obvious in four seconds.

The sixth failure is holding one indicator and no others. A property with a single headline figure has a verdict and no diagnosis. It knows that performance fell and cannot say whether the standing load grew, whether a circuit failed, whether a schedule drifted or whether the denominator changed shape, because it has not kept the three or four cheap supporting series by which those are told apart.