Using high-technology to assess workplace collaboration

Jay L. Brand, PhDGabor Nagy, PhDIdeation, HAWORTH

INTRODUCTION

The office design community has begun to recognize that corporate real estate and facility management decisions cannot be made independently of their influence on occupancy quality. The quality of the workplace should not be compromised simply to save money; in fact, many leading organizations now consider workplace design an important strategic investment to drive organizational effectiveness and business success. One consequence of this shift from cost-centered to human-centered design involves defining success for office renovations and new office workspaces in terms of the quality of human experience rather than traditional knowledge worker productivity. Measuring productivity assumes a broad similarity between knowledge work and production work (e.g., in factories).

However, it may be impossible to subdivide the components of “knowledge work” into their constituent elements, measure the time required and accuracy for each, and integrate across them to estimate the quantity/quality of work per unit time--the fundamental “process” approach for measuring production in factories. Instead, the quality of ideas (e.g., creativity; innovation) constitutes the fundamental “value” in knowledge work, and this relates more to work attitudes and other psychosocial issues than to analysis of the speed or efficiency of behavior. In other words, “knowledge work” differs fundamentally from “production” work. For the latter, speed (time) and accuracy (quality) jointly determine “productivity,” whereas in the former, quality ultimately replaces quantity as the essential value proposition; one billion-dollar idea can “make or break” a knowledge-based organization. Additionally, for knowledge work, quality fluctuates as an unpredictable function of many factors, including even macro-economic ones. Supporting the quality of knowledge workersʼ experience(s) related to work represents one remaining useful investment strategy.

In this regard, Teresa Amabile and her colleagues (Amabile & Kramer, 2011) have provided an excellent framework for describing effective organizational contexts for high-quality (creative; innovative) knowledge worker performance. Although beyond the scope of this article, we believe that social network analysis (SNA), the focus of this paper, can be combined with “The Progress Principle” to provide useful, integrated metrics for understanding the value of organizational investment in knowledge work. SNA describes interaction within teams, quantifying collaboration, and “The Progress Principle” describes supportive, collaborative organizational contexts to encourage creativity and innovation. In contrast, summing or averaging across individual performance to predict organizational effectiveness fails to address many of the critical issues for business success in the new, international economy.

June 3, 2013 1

Brand & Nagy 1

One exception to this somewhat gloomy prospect for determining/predicting the value of knowledge worker productivity involves objective assessment of interaction patterns within office environments (Olguín-Olguín & Pentland, 2010). Investment in human capital includes subsidizing continuing education and investing in training programs. However, “social capital” assumes that unique value exists within the interactions between and among people and groups of people, independent of employeesʼ individual contributions. If these interactions can be measured and made visible, implications for collaboration can ultimately be derived and related to aspects of workplace design. Furthermore, even in our increasingly mobile work world, unique advantages for face-to-face interaction remain, primarily due to the rich, nonverbal communication it provides. One of these advantages involves quick, efficient resolution of ambiguities.

Face-to-face interaction always occurs “somewhere,” and thus an interesting “chicken or egg” conundrum arises: Do social networks reflect the influence of “place” on collaboration, or do they represent communication patterns to which place design should respond? If social networks ultimately link organizational performance to collaboration and innovation (cf. Pentland, 2012), workplace strategy may be subordinate to organizational design, at least in terms of its role in supporting collaboration. However, strategic alignment (perhaps even integration) between workplace and organizational design can provide unique advantages for driving business success compared to alternatives.

OVERVIEW & APPROACH

Haworthʼs Ideation team has been using technology that involves wearable badges and space-anchored antennas to provide this type of objective measurement of interactions, including where they occurred, who was involved, their frequency & duration, and certain of their attributes (e.g., speaker/listener; active/passive - measured with accelerometers). We do not record the actual verbal content of such communications, but we capture directionality or orientation of the people involved in order to eliminate proximities that do not involve communication, a disadvantage of approaches using RFID or ZigBee technology. This new technology can be implemented in many possible ways using various methods, but measuring interaction patterns over a two-week period provides a somewhat representative “snap shot” of the patterns of communication for the individuals and groups participating.

This technology requires participants to wear a badge similar to an employee ID card that transmits & receives infrared signals approximately every second. Antennas can be placed in any space or area to “anchor” the interactions between/among badges, as well as to anchor the location of individual badges. However, even in the absence of an antenna, the badges continue to capture interactions (who, with whom, frequency, duration; just not, where?). Every evening, participants return badges to a base station for recharging and uploading data, and they pick them up each morning. Although strictly speaking, this approach gathers objective measures of badge/antenna proximities, 3D accelerometers (active/passive meetings), and voice power detectors

June 3, 2013 2

Brand & Nagy 2

(speaker/listener roles) to characterize interaction, understanding communication and collaboration is the ultimate goal for such assessment.

After data have been collected for two weeks (at least 10 business days), software such as UCINET may be used to visualize & evaluate the interactions among the participants. The broader and more complete the participation (groups of 150-250* are ideal to capture “Big Data”), the better the estimates of the social networks based on interaction patterns will be. Interaction duration thresholds (e.g., 1-minute, 3-minutes, 10-minutes) can be used to focus on meaningful communication. Empirical reasons have been established for accepting a three-minute threshold as the minimum for eliminating “chit chat” although there may be important cultural differences in this regard. For example, in New York City, people tend to focus “on topic” much faster than in West Michigan; thus, thresholds for filtering “socializing” could be shorter there, perhaps only one or two minutes, whereas in West Michigan they may need to be longer, even five or more minutes.

RESULTS

At this point, these interaction data can be evaluated similarly to any other social network analysis (SNA). Individual actors within such networks are represented as dots, called nodes, in network diagrams; the connections between nodes, called ties, are represented by lines (see Figure 1). Individual (ego level) analyses evaluate connections among individuals within and between teams, identifying speakers, listeners, hubs (individuals central to their group/team), outliers (individuals on the periphery of interaction within their group), bridges or liaisons (individuals linking two groups; these can serve as “connectors” or “bottlenecks”) and other functional roles.

Group level analyses can identify cliques (a group of individuals where everyone interacts with everyone else; this situation also represents 100% density), fragmentation (higher interaction within than between groups), isolates (groups with no ties to other groups), propinquity (the tendency for spatially proximal nodes to be connected), cohesion & centrality (convey the nature of intra-group and inter-group connections), and innovation potential (relatively large, highly interactive groups characterized by high density, low centrality (no bottlenecks), and high cohesion (suggests egalitarianism and trust among group members). The authors have developed an exploratory algorithm to predict this “innovation potential;” investigation to verify is ongoing . . .

SPACE UTILIZATION

Precise, objective space (or time) utilization estimates can also be derived from these data, but this approach provides additional valuable information, such as how many people were using the space, which departments/groups were involved, whether the meeting was “active” or “passive,” (based on movement patterns among participants) and the speakers and listeners, along with the nature of the ties among group members and their ties to other groups.

June 3, 2013 3

Brand & Nagy 3

ADJACENCY ANALYSIS

Although nodes in a network can be labeled with peopleʼs names, we usually maintain confidentiality for employees by using aliases consisting of letters for departments paired with numbers to designate individuals (e.g., G1, G2 . . . GN). The research literature primarily provides analyses at the ego (individual) level or across entire networks; however, for workplace strategy, we believe that group-level evaluations provide the most meaningful insights. For example, by evaluating the ratio between intra-group and inter-group ties, we can provide “bottom-up,” data-driven reasons for adjacency recommendations using adjacency diagrams (also called “bubble diagrams.”) Three categories defined by the nature of these empirical inter-group connections (weak, moderate, strong), correspond with “no necessary adjacency,” “optional adjacency,” and “critical adjacency,” respectively, between any pair of groups throughout the floor-plate. Briefly, members of highly interactive groups (e.g., cliques) should share dedicated space (such as project or war rooms), and any two groups with strong inter-group ties should ideally be co-located [or adjacent].

With respect to these empirical adjacency analyses, itʼs interesting to speculate about the design dilemma mentioned earlier: Which comes first - the “chicken” (providing dense spaces with clear boundaries helps groups forge strong bonds) or the “egg” (groups with strong bonds require their own space with clear boundaries)? In this regard, our data suggest that propinquity (spatial proximity influences interaction) is an important outcome, meaning that for most groups, stronger connections characterize members of groups sharing space on the same floor compared to members of groups occupying space on different floors; research also shows that proximity to a stairway reduces this spatial segregation tendency.

CONCLUSIONS & IMPLICATIONS

Finally, several important design implications can be derived from these social network analyses and diagrams:

1. Low space utilization rates plus high connectivity suggests a higher ratio of group/team collaborative workspace(s) to individual workspace(s)

2. Longer interactions (60 minutes or more) tend to involve only two to three people, suggesting that most spaces for interaction can be relatively small; although triads may be important for collaboration, they can make space planning difficult!

3. Ties between floors even within highly connected groups are weaker than those among group members sharing space on the same floor (propinquity)

4. Empirically derived group adjacencies can be an excellent starting point for blocking & stacking

June 3, 2013 4

Brand & Nagy 4

5. Ideally, space planning should support the dynamic nature of interactions within and between groups

6. Finally, a few unassigned “touch-down” workspaces for private, concentrative work should be embedded within each groupʼs team community or neighborhood (cf. Cain, 2012)

*This is simply a practical limitation, as monitoring larger groups of people becomes unwieldy.

REFERENCES

Amabile, T. & Kramer, S. (2011). The progress principle: Using small wins to ignite joy, engagement, and creativity at work. Cambridge, MA: Harvard Business Review.

Cain, S. (2012). Quiet: The power of introverts in a world that canʼt stop talking. New York, NY: Crown Publishing Group.

Olguín-Olguín, D. & Pentland, S. (Alex) (2010). Sensor-based organisational design and engineering. Intʼl J of Organisational Design and Engineering, 1(1/2), 69-97.

Pentland, S. (Alex) (April, 2012). The new science of building great teams. Harvard Business Review, reprint R1204C (http://www.hbr.org/search/R1204C).

FIGURE 1 - A social network diagram, illustrating nodes, lines, “speakers” (red) and “listeners” (green); circle size corresponds to extent of speaking or listening

!"!#

!$

!%

!&

!'

!(

!)

!*!"+

!""

!"#

!"$

June 3, 2013 5

Brand & Nagy 5