Measure for Measure

Benchmarking is the process of using a formal analysis to compare certain characteristics of a building with those of another.

To get the most of out the process, you must determine what you want to accomplish. This is the first issue discussed below; the remaining ones will help you evaluate whether benchmarking is helping you meet your goals as effectively as possible.

1. What’s your goal?

   The various easons to benchmark include to:

   1. Reduce your operating expenses or space utilisation
   2. Identify how to make your building greener
   3. Justify to management a desired improvement
   4. Know whether or not you are managing effectively
   5. Monitor how your buildings are performing over time — are their rankings improving?
   6. Find out how your clients are performing compared to the rest of the population.

   Note that (a) and (b) relate to the intention to improve a service; the remainder are used only for comparison. Most who benchmark are interested in comparisons only — to see how a building is doing compared to others. However, this can motivate improvement in the building’s performance.

2. Know your area

   There are different ways to measure a building, such as using gross or net figures, and several standards (IPD or North American for gross; BOMA or IFMA for net). Most systems use gross area, but even within that, there are different standards.

   Different standards can vary by as much as 15 per cent when measuring the same building, so it follows that benchmarking data could be out by 15 per cent.

   The best systems give FMs a choice and then allow them to compare their results to only those that are measured the same way. Systems that force someone to measure through a specific standard aren’t as reliable, as they are dependent on the FMs being willing to re-measure their site.

3. Average or median?

   Most benchmarking tools use medians, while some use averages. Medians are what the middle-performing building is doing, regardless of how high or low the buildings at the extremes are. Averages (or means) use data from all the buildings and calculate the average. Medians are considered more accurate for benchmarking as they downplay any outlying data at the extreme ends of the range.

4. Which report?

   There are many types of reports one can generate, and you will have to know which ones you may want to generate so that you can select the best benchmarking tool for you:

   • Building performance metrics (for example, utility consumption per unit area, area cleaned per custodial worker, maintenance cost per unit area, and so on)
   • Best practice reports (to generate improvements, as opposed to comparisons only)
   • Show multiple buildings for the same client on one report
   • Show multiple years of benchmarking for the same client on one report
   • Print quality reports to give to management.
5. Identify your filters

   There are lots of types of filters that one can generate to compare buildings. Many of these are fairly common and include climate, building size, building age, hours of operation, primary use.

   Then there are some that may be more important, but are not tracked in many systems:

   • Maintenance craft
   • Preventive versus corrective maintenance
   • In-house versus contract labour
   • Union verses non-union labour
   • Security clearance requirements

   Keep in mind that the more filters selected, the more buildings needed in the benchmarking database to generate meaningful comparisons.

6. Using valid data
   1. How up to date is the database? Knowing when the data was input for the portion of the database you will be using is very important. Certain parts of some benchmarking databases are updated only once every three years and most of this data was captured at least 3-6 months previously and may represent data from the previous fiscal year. Such data is usually too dated to be meaningful.
   2. What are the safeguards to prevent anomalies or data with errors from being input? There are many opportunities for human error, ranging from holding a key down too long (resulting in an extra digit or two), or just a building so unique that it probably shouldn’t be benchmarked. Some systems use tools to flag these situations so that the FM must approve them, while others ignore them.
7. The filter set

   With any benchmarking system, it works well only if there are enough buildings in the comparison set (filter set) to yield meaningful results. This usually means a minimum of 25-30 buildings with your selected filters (when applying three to four filters simultaneously, this can result in a database of over two or three thousand buildings). The more filters one selects, the more buildings one will need in the database. In this sense, the database can never have toomany buildings.

   Geographic filters (for example by city) can be important. Unfortunately, most systems do not contain enough buildings in enough cities to make benchmarking viable, so workarounds become relevant and can be done through use of specialised filters.

   Here are a few examples that your benchmarking system should have:

   • For utility consumption, look at cities with similar temperatures and humidity
   • For labour costs, look at cities with similar labour rates for certain common tasks.
8. Refine your filters

   The object here is to determine which filters have the most impact and then limit the report to just the three or four that will be most meaningful. For example, let’s assume that we have identified six potential filters.We start with the one most likely to have an impact and generate a report for that filter, and note in which quartile our building falls (see chart).

   We then try another filter. If our building appears to move significantly to the left or right, we should keep that filter. Otherwise, we turn it off and look at a third one. We keep doing this until we have tried all six filters, and then keep the three most meaningful.

9. User friendly

   The following aspects of user friendliness are important:

   • Secure web that allows you to access results from any location
   • Ability to batch input data from multiple buildings via an Excel (or equivalent) data entry form
   • Not requiring that all data fields be input
   • Not requiring that all data fields be input at the same time (not all data are always available).

   Ease of use characteristics include:

   • It should be obvious how to use the system without much training
   • Even if it is obvious what to do, keystrokes for common tasks should be kept to a minimum
   • The system should be fast, reducing the user’s waiting time to go from screen to screen.
10. Draw conclusions

    Without accurate conclusions, all the above work is meaningless.

    Let’s assume that the FM already has correctly selected the most appropriate benchmarking tool for the desired output. The FM will still need to exercise judgement in the following areas:

    • Knowing which reports to generate
    • Knowing which filters to apply
    • Areas of required experience and expertise
    • Knowing which best practices we should implement

Benchmarking is not an exact science — it is both an art and a science. The first part of benchmarking is to understand your benchmarking goals. Then, one needs to select a system that will facilitate the achievement of those goals. Finally, one must apply a mixture of good judgment, trial and error, expertise and experie