1. Improving Organizational Productivity with Building Automation Systems
Trevor Nightingale, National Research Council, Director
Monday, January 22
10:30 AM - 11:30 AM
Room S104A

2. Project Funders
CABA Research Funders:
Other NRC Project Funder:

3. Executive Summary
New approach to quantifying organizational productivity effects
“Better buildings” strategies have positive effects on organizational productivity metrics similar in size to other corporate strategies
RBC’s LEED office buildings associated with better employee job satisfaction, and manager- assessed performance
Further work will identify specific mechanisms and technologies

4. What is Organizational Productivity?
Output $ / Input $
Multiple KPIs on both sides of the equation
In general, organizational productivity is the efficiency with which an organization operates. It is the balance between input costs and output values. Organizational productivity improves when costs are reduced (e.g., recruiting costs lowered, benefits costs, lowered, or energy and maintenance costs drop), or when outputs increase in value (e.g., sales volume increases; newly introduced products increase sales; revenue rises with product quality improvements). Clearly, for a white-collar, office based organization, this is a more complex equation than in an older factory model, where input costs for raw material, and market sales for finished product, are relatively straightforward to quantify. Multiple studies have shown that buildings affect many of these inputs and the value of outputs.

5. Approach
Use the buildings part of the budget to enhance the salaries and benefits part of the budget
The starting point for interest in this area is the commonly recognized observation that for modern, white-collar, service industry workplaces, the costs of staff far outweigh all other costs of doing business.
This is a widely-cited breakdown of the costs associated with office workplace costs over a 10-year period (Brill, Weidemann, & BOSTI Associates, 2001). Another common rule of thumb is that the annual operational costs of an office space are, on average $300/ft2 for staff payroll, $30/ft2 for space rent, and $3/ft2 for utilities (Best, 2014). Thus, one would not want cost savings in buildings to come at the expense of staff’s ability to do their work. Ideally an organization would identify building strategies that support the productivity of the organization, and are cost-effective as a whole.
In other words, a relatively small investment in building design and operation can have a relatively big benefit on organizational productivity through positive effects on staff (and energy use). The results of this project demonstrate that this is possible.
Brill, Weidemann, & BOSTI Associates, 2001

6. Approach
Traditional productivity thinking not applicable to modern office work
Multiple metric approach from CABA and WGBC
Balanced scorecard concept widely accepted in other contexts
No broadly accepted definition of what constitutes appropriate metrics for office work. At one time decision-makers sought very simple cause-and-effect relationships; i.e. “If I replace <BUILDING FEATURE X> with <BUILDING FEATURE Y> then productivity (simply thought of as the amount of work produced by an individual) will increase by Z%”. This is partly a hangover from an industrial production line model of productivity in terms of output of standard units.
There is increasing acceptance that such a model is not applicable to most white-collar workplaces, where output is rarely measurable in such terms. Instead, productivity is better represented by a basket of metrics, sometimes measured in different units, that all influence the overall balance of costs and revenues in an organization.
Building features can affect both input costs and output values. Poor indoor air quality could result in increased illness absence (a cost). Employees who are not ill might have increased workloads to compensate for absent colleagues, and this could in turn result in lowered output quality (reduced output value). This is a more complex and nuanced approach than the simple industrial relationship, but offers a pathway to move forward in this domain that an overly simple metric does not offer.
Oseland and Burton (2012) noted that while most researchers acknowledge that there is an under-appreciated relationship between office design and environmental conditions and (organizational) productivity, the fact that it is difficult to quantify means that it is.
The only metric typically used in relation to the workplace has been size, which has led to strategies that promote density, and thus a saving on the one metric used, rather than the well-being and performance of those inhabiting the space. Indeed, Oseland and Burton (2012) found that “… only one in eight organizations had productivity metrics in place and none monitored the relationship between the environmental conditions and business performance.” Lacking metrics in this domain, organizations tend to focus on the real-estate costs of space, which are easily measured and monetized.
This is exacerbated by the classic split incentive problem, facilities management typically reports to the CFO/COO, with a motivation to cut costs, rather than to an executive with a responsibility for employee well-being or workflow optimization (The Stoddart Review, 2016).
Multi-metric definitions for organizational productivity have emerged from major international organizations in recent years; i.e. CABA and the WGBC. Not all of these metrics are measured in the same units, or have simple translations into dollars. Fortunately, decision-makers are now used to balancing the costs and benefits of multiple metrics when making decisions: “balanced scorecard”, “triple bottom line” are common approaches.
Our recent project with CABA enabled us to develop this multi-metric approach, and then to compare the effects of “better buildings” on these metrics to the effects of other corporate strategies.

7. Approach – analogy to improving energy performance
Does my energy performance need to be improved?  benchmarking
Assess several strategies to improvement
Benefits often better expressed in equivalent terms
Apply this process across multiple metrics related to organizational productivity
(Monetization is done locally)
IMAGE +
It might be helpful to compare elements of our approach with org. prod. to way that energy efficiency upgrades are evaluated.
The energy process usually begins with benchmarking. If a building operator benchmarks their building’s energy use against other similar buildings and observes underperformance, they obtain a signal that energy saving in their building is a reasonable priority, and is likely to be fruitful. Similarly, establishing suitable benchmarks for the proposed organizational productivity metrics will enable decision-makers to understand whether or not their own building’s metrics are lower than desired. This might be expected to motivate them to look at options to improve these metrics.
Several options to improve energy performance would then be considered, and their expected costs and benefits, based on existing knowledge and case studies, compared.
Note that when talking about energy upgrades, the benefits are often expressed in equivalent units, it is like taking so many cars off the road, or planting so many trees. The positioning of such information in different units (cars, not dollars, not kWh) can be more appealing to some decision-makers.
As an example, consider a relatively straightforward metric of clear organizational value, absenteeism. Suppose we demonstrated that a particular building technology reduced absenteeism by one day per person per year. It may be true that a workplace health program can also be shown to reduce absenteeism by one day per person per year. In this (fictitious) example, one could say that installing this building technology was like giving everyone access to a workplace health program (with respect to absenteeism, at least!).
Monetization is always done locally. Energy costs/benefits must be translated from (e.g.) % effects to $ by experts locally, as the local costs of labour, components, and energy will vary from place to place, and local utility incentives will also vary. Similarly, multiple approaches to improving org. prod. may be considered and their costs/benefits evaluated using local multipliers. Rather than providing simple monetary equivalents we propose to provide the information to organizations such that they can go through a monetization exercise themselves, should they wish to do so, using assumptions and multipliers that unique to them.
# cars off the road
Planting # trees

8. Organizational Productivity Metrics
Absenteeism
Employee turnover intent
Self-assessed performance
Job satisfaction
Health and well-being
(Complaints to the FM)
Benchmark developed for each
IMAGE +
This is the set of org. prod. metrics used in this study, each of which have self-evident value to any organization.
The choice of these metrics was not arbitrary. They were derived from a conceptual model of the interplay of workplace environment elements, employee effects and behaviours, and organizational outcomes established by a logical connecting of multiple well-established studies. No single study has ever measured all of these metrics and demonstrated their interaction.
Complaints to the facility manager was a target metric, but we were unable to find any usable studies – this is an important research gap.
Benchmarks were derived from national statistical databases, or sizable peer-reviewed studies of representative buildings.

9. Corporate Strategies Better buildings
Office type (private vs open-plan)
Workplace health programs
Bonuses
Flexible work options
Peer-reviewed literature synthesized for effects on each organizational productivity metric
4000+ abstracts, and 500+ full publications reviewed
Studies from real office workplaces only
IMAGE + NRC owned
Compared the effects of better buildings options to other corporate strategies.
We chose to focus on strategies that we believed would be familiar to a building manager (e.g. Bonuses) or strategies that might be implemented by, or with the participation of, the building manager themselves (e.g. Office Layout, Flexible Work Options).
Office Type: Focus on comparison of private offices vs. open-plan. The justification for the general trend towards open-plan has typically been the expectation that it will bring increased flexibility, transparency, and enhanced communication between team members, although the underlying economic driver has been real estate-cost savings. It is a very familiar strategy, and thus serves as an excellent touchstone against which to compare other strategies.
Workplace Health Programs: WHPs often form part of the benefits package in large organizations, with the general belief that this investment will support employee well-being, which in turn will benefit the organization’s productivity . WHPs tend to be offered as packages of measures designed to promote good health; e.g., health counselling, gym access, nutrition programs, stress management, physical tests (e.g., blood pressure).
Bonuses: It is a common belief that financial incentives offered to individuals will elicit employee behaviours and perceptions that will benefit organizational productivity. We focus on bonuses provided for general job performance, typically evaluated by a manager.
Flexible Work Options: Studies have typically looked at a package of options. This category can include flexibility in scheduling working hours to be conducted in the organization’s own building, the availability of multiple workplace locations within the building, or the ability to telework.
The scope of this work is highly multi-disciplinary, and thus the literature search encompassed many fields, including business, medicine, psychology, engineering, and facilities management.
An important criterion in searching for relevant publications was to find work conducted in real workplaces. There are many lab studies relevant to this topic, particularly related to better buildings strategies (and NRC has done many itself). However, the focus of the research was on org. prod. outcomes, and it is very difficult to create a meaningful org. prod. context in a lab. Studies in real workplaces are much more challenging to do, and are rarer, but were most relevant to this research.

10. Workplace Health Programs
Results
Strategies (IV)
Better Buildings
Office Type
Workplace Health Programs
Bonuses
Flexible Work Options
Benchmark 
Metrics or KPIs (DV)
Unit 
2 – 15
Absenteeism
 0.4 – 1.5
 3.2
 0 – 1.8
 1.0 
day/per/yr
18 – 30
Employee Turnover (int.)
 1.3
 18 
0 – 100
Self-assessed Performance
 2 – 10
 8– 15
 0 – 10  %
60 – 80
Job Satisfaction
 4 – 9
 5 – 10
 0 – 12
30 – 60
Health & Well-being (symptoms)
 5 – 9
55 – 75
Health & Well-being (overall)
 6 – 10
 11 – 12
 6
This table summarizes all of the information we synthesized from those hundreds of published papers. It is very dense, and we can describe it in a systematic way – here’s an example for absenteeism …
1. Begin with the first metric: absenteeism, it’s definition and unit of measurement are below. The benchmark comes from international and industry databases – the range is wide 2-15 days/person/yr depending on country, industry etc. The full report shows these breakdowns, so the user can better define a suitable benchmark for their context. Studies show a consistent trend for better buildings to lower absenteeism by day/person/yr, the range representing different studies looking at different technologies/approaches. Reading the columns to the right shows effects for other corporate strategies. Note that although we are putting these strategies side-by-side and comparing their effects on the various metrics, the mechanisms by which these strategies have their effects may be quite different.
The final Matrix showing the benchmarks for each metric, and the effects of various corporate programs. The benchmark has a purple background if it was derived from national/international statistical surveys, and no background if it was derived from targeted research studies or theory. The arrow in each cell indicates the direction of the effect. The number in cell indicates the size of the effect (in the same units as the benchmark). An arrow without a number indicates that the direction of the effect is established, but a size was not derivable (in our preferred terms) from the published studies. The effects attributed to Better Buildings strategies are highlighted as they are the primary interest of the project.
The number in the cell represents the preponderance of available information, and is based on our judgement (i.e., it is not the result of a quantitative meta-analysis). We present a range if several studies contributed a variety of results. That range might start at zero if several studies found no effect and several found consistent effects. Empty cells denote combinations of corporate strategies and KPIs for which we found no relevant studies on which to base a conclusion.
Study results and benchmarks have been rescaled into common units using straightforward assumptions and linear translation, and not any formal statistical method.
Explanation of units for each metric:
Absenteeism [Unit: Days/person/year] LOWER is better
Focus on short-term sick leave that an employee takes for their own illness, and following their own assessment of their health. This is often self-reported by employees on surveys, but might also be derived from HR databases.
Employee Turnover [Unit: scale (likelihood to look for another job)] LOWER is better
Our focus was on data assessing whether someone leaves their job voluntarily (the inverse of employee retention)
This metric is based on survey data on turnover intent. Questions in different surveys are worded differently, as are the response scales. A typical question might read, “How likely is it that you will make a genuine effort to find a new job with another employer within the next year?” A typical response scale might have seven distinct categories with end labels “Extremely Unlikely” to “Extremely Likely”. To convert this to a common scale these labels would be assigned a value; e.g. 5 and 95, respectively, with other labels in between pro-rated. All responses from a group would then be averaged to find a value for a population, organization, or building. Note that an organizational average of 25 on this scale would mean that, on average, employees say they are “somewhat unlikely” to be looking for another job, not that 25% of employees are actively looking for a new job.
Self-assessed Performance [Unit: % scale] HIGHER is better
Employees have been asked to self-assess their own productivity in their own workplace. A typical question phrasing by researchers might be, “Please estimate how you think your personal productivity at work is increased or decreased by the physical environmental conditions”, with a seven-point response scale from -30% to +30%. In NRC’s interpretation, this is unlikely to be a reliable measure of an employee’s actual material output in percentage terms, and is more a measure of environmental satisfaction in the sense of how the indoor environment supports the employee’s ability to do their job.
Job Satisfaction [Unit: scale] HIGHER is better
Many different question wordings (and single items or averages of multiple items) and scales have been used in many studies. An example single-item scale is, “Taking everything into consideration, what is your degree of satisfaction with your job as a whole?” rated on a seven-point scale from “Very Unsatisfactory” to “Very Satisfactory”. We used judgement to interpret the equivalence of different question formats and to normalized the data from each study to a common scale (see note on turnover intent above).
Health and Well-being (symptoms) [Unit: scale] LOWER is better
In the buildings research domain, the prominence of Sick Building Syndrome (SBS) led to the relatively frequent use of surveys to assess associated symptoms. These symptoms included dry eyes, runny nose, back pain etc. Survey methods included asking about frequency or intensity, or both, and these survey items prevailed in some studies after the SBS phenomenon subsided. Again, the equivalence of these different question formats was considered to re-scale the results from individual studies to a common scale. For example, 10=very mild/infrequent health symptoms; 90 = frequent/intense health symptoms in a population.
Health and Well-being (overall) [Unit: scale] HIGHER is better
Surveys are often used to determine an individual’s general state of health or well-being. These have included both national and international social surveys, and individual research studies. These data were also normalized from the scales reported to a common scale. For example, 10=very poor overall health in a population; 90 = excellent overall health in a population.
Overall, we see that better buildings have a beneficial effect on all of the metrics, and generally the size of these effects is similar to the size of the effects of other corporate programs.
We expect that expressing the data in this way will give more opportunity for better buildings strategies to be part of the investment conversation with corporate decision-makers. Traditionally, better buildings might have only been see as a way to save energy or maintenance costs, now they can be seen as a vector to improve other organizational productivity metrics, and may be considered alongside other strategies often employed for these kinds of improvements.

11. Unlocking Existing Data Sets
Energy
Facilities
Lower energy costs, reduced maintenance, flexibility
Better Buildings
Improved organizational productivity
Healthier, happier, more effective people
Better environmental conditions & greater responsiveness
That was synthesis of results from many studies, in many organizations, each on one or two metrics. How could one conduct a comprehensive study of many metrics in a single organization using original data?
Here are the two main pathways by which a BAS can affect organizational productivity metrics. The upper pathway is via reducing energy costs; the lower pathway is the focus of our recent work, which is via improved workplace conditions that in turn support employee health, well-being, and ability to perform their tasks effectively.
Energy is relatively easy to measure, there’s a meter for it held by the facilities staff, which is one reason why it gets so much attention.
But a key insight from the WGBC report was the recognition that data on many of these important metrics on the lower pathway already exist in an organization and are collected routinely – the organization actually has meters for them too. In other words, one does not necessarily have to engage in an expensive or invasive data collection campaign to explore the relationship between the built environment and organizational productivity in an organization. Rather, it may be a matter of securing permission to use existing data for this purpose, collating them, parsing them by building, and associating them with local building characteristics.
The facility management (FM) company often maintains a database of complaints about the built environment registered by individuals, as well as the response time and cost. The FM might also keep historical records from the BAS, which will provide data on some physical indoor environment conditions, such as space temperature and RH, and zone-level CO2 concentration.
The real “gold” can be found in corporate HR databases that might already hold data pertaining to staff retention/turnover, absenteeism, and other aspects of employee health and well-being. The HR departments in many organizations also conduct regular employee opinion surveys that contain data on job satisfaction and organizational commitment. The marketing departments in large organizations might conduct customer satisfaction surveys, and the finance department will likely have data on business unit performance. Many office building landlords regularly administer tenant satisfaction surveys that contain items related to environmental satisfaction.
BAS data
Complaints
Facilities
Absenteeism
Retention
Performance HR
Surveys

12. Applying this Model to RBC
Data from RBC Facilities (RE team)
Building characteristics (age, size, location, lease etc.)
Green credentials
Work order history (complaints to Facility Manager)
Mapping of employees to buildings
Data from RBC HR (HR team)
Demographics (age, gender, education, dependents, languages …)
Job classification, Salary, Staffing Actions
Manager-assessed performance
Employee Opinion Survey (EOS)
Our project with RBC was an example of analyzing a sub-set of these metrics using existing organizational data.
Data Confidentiality of utmost importance, achieved through:
pre-anonymization
encrypted drives
no network copies
no unsupervised processing

13. RBC Data Summary
Some data from one time point, others over multiple years: Focus on
70 office buildings,1130 branches, 70,958 employees
~120 million data points (only a subset of the total!)
Focus on large office buildings (>100 employees)
13 LEED certified, 33 conventional
Control for other differences between buildings
Each LEED building matched with a conventional building similar in location, size, age and average employee profile
10 Matched pairs, 14,569 employees

14. Outcome Variables Variables used in analysis 4 composite EOS variables
Manager assessed performance
FM complaints per employee
EOS has >100 questions, asked annually. RBC’s consultants create 16 composite variables from the individual items, and we further roll these up into 4 measures, each rated We also converted manager-assessed performance into a 0-1 scale. HVAC-related complaints were also tallied (although we found no effect by building type here, so I won’t report further – nevertheless, I think this is a rich source of data worthy of further investigation.

15. Results summary Two analysis methods:
Compare building averages in matched pairs
Compare individuals between building pairs
Green buildings demonstrated higher EOS scores, and lower variability
Manager-assessed job performance higher in green buildings
Results here are supported by statistical significance tests

16. Building Averages
Average ratings generally high (so, overall, RBC is a pretty good place to work). Green building means tend to be higher, and sds lower. Supported by Wilcoxon Signed Ranks significance tests.
(GRN note – I know this is dense, and I don’t expect people to see the details, just observe some trends. I thought this would be easier than trying to explain the Wilcoxon graph to this audience in the time available)

17. Individuals in Paired Buildings
Stars indicate s.s. effect on that metric for that matched pair. Shaded cells are cases where the green building was rated more highly. Boxes with heavy outlines indicate larger effects.
Multivariate analysis of covariance (MANCOVA) shows differences between most building pairs on many metrics, which support Wilcoxon findings - in most cases the green building is favoured. Specifically on the manager-assessed performance metric, there are s.s. differences for 2 pairs, and in both cases the green building was favoured.
Both of the slides show that not every green building is better than every conventional building (and not on all measures). But across a portfolio there are consistent benefits.

18. Conclusion Multi-metric approach to productivity established
“Better buildings” show benefits on these metrics
Enhances the business case for better buildings
Promising start, but replication and extension required
Complements prior research on the benefits of green buildings on energy use, occupant comfort, and real estate values.
In new work we’d like to cover more metrics, and to identify which specific aspects of green (or better) buildings lead to the positive effects.

19. Next Steps Working with CABA and others to extend this work
Use of archival data (like RBC study)
We need more complete data sets (e.g. Productivity KPIs, Comfort, BAS data)
From more organizations
Focus on effects of specific systems (e.g. Pre-post retrofit data)
Development of internal tools for on-going tracking
Ultimately, we want to know what specific technologies, operational practices, and indoor environment conditions produce better organizational productivity outcomes. If metrics are recorded historically we can examine these before and after the retrofit of a particular technology, or we can look at a population of buildings with various combinations of technologies and statistically isolate the effects associated with each technology.
Results based on archival data can be used by organizations to use data they already have to develop tools to track productivity-related effects of any real-estate modifications they implement. HR departments are used to using such data to study the effects of a new HR practice, so why not do the same thing to study the effects of building attributes?

20. Next Steps Original data collection via field study
On-site physical measurements
Detailed survey data
To drill down to specific mechanisms we’ll need to do some original and customized data collection on the physical conditions in buildings and occupant perceptions and behaviours. This involves Post-Occupancy Evaluation protocols, in which NRC has much experience and global leadership.
Image: The Corporation © Big Picture Media Corporation

21. Physical Measurements
Spot measurements
Temperature, humidity, air speed, formaldehyde, particulates, TVOC, CO2, light level, noise, SII
Longitudinal data
Temperature, humidity, air speed, CO2, light level, noise
Examples of equipment we use for on-site physical measurements.

22. Survey Data On-line questionnaire: Environmental satisfaction
Job satisfaction and Job demands
Demographics
Organizational commitment
Workplace image
Internal communications and relationships
Engagement
Comfort-related modifications and complaints
Satisfaction with local amenities
Health, Mood, Absenteeism
Sleep quality
We ask the same core set of questions in every project, and add other modules depending on local circumstance. This is a list of the concepts addressed by our latest comprehensive survey. These are all outcomes that we expect to be affected by the built environment.
We typically get a response rate of 30-35%, which is very good for a voluntary on-line survey that takes ~20 mins.

23. Growing Datasets
30+ buildings in North America; physical measurements, and survey data from occupants
Survey used by other researchers in 100+ buildings in US, Germany and Switzerland
Pre-post study at OAA HQ net-zero retrofit
Pre-post study at PDPIII
We have a lot of experience doing this, and our methods have been used by others around the world. This allows us to build a benchmarking dataset against which we can compare results from any new building.
We are currently involved in pre-post studies for two important projects. Place du Portage is the biggest single site we have studied. Multiyear Multi-hundred million dollar project to retrofit a building housing 5200 federal office workers. Physical measurements from >200 workstations, completed surveys from >1900 occupants.

24. Measuring Organizational Productivity
KPIs more accessible with new technology?
Engagement
Communication
Collaboration
Cognitive task performance
IoT, sensors, data analytics:
Use of alternative work arrangements
Work & communication patterns
Localized physical conditions
New technology, leveraging the Internet of Things, smartphone apps and wearables offer intriguing possibilities to measure productivity metrics in new ways. These methods still have to be developed and validated, but offer great potential.

25. Join Us If you are interested in working on these issues …
If you have data you’re willing to share …
If you want to advance the cause of better buildings, boost the business case, and increase market uptake …
Let’s talk!

26. CABA study: www.caba.org/White-Papers
Thank you
CABA study:
RBC study:
CONTACT
(613)