Documente Academic
Documente Profesional
Documente Cultură
B4. Accounts Receivable. Since this business unit did Preliminary Findings, Lessons Learned,
not have a good customer satisfaction metric, Accounts and Recommendations
Receivable (AR) data, which were available for 39 sub-
divisions, was used as an additional business metric for 1. “Importance” scores provided marginal value. The
this study. Since we were looking for an indication of first part of our analysis looked at the importance that
business performance, the metric we used was the pro- employees placed on 41 attributes of the EVP. Results
portion of AR that was unpaid after more than 180 days. showed that most employees rated each of the attributes
Using the proportion rather than actual dollars served to as “very important” or “extremely important,” providing
control for the different organization sizes. little variance in employee responses. Although we ex-
pected more variation in EVP importance scores, these
An example of the linkage model illustrating the results were not totally surprising when considering that
hypothesized relationships between all of the employee our attributes are anchored in the most current research
measures and business outcomes is shown in Figure 2. on employee engagement and retention. That is, many of
the attributes that we included on our survey have been
Data Analyses shown to be important to the general workforce. With
almost no variability, the importance scores did not pro-
Relationships between employee survey responses and vide any additional value for the analysis. Likewise, there
business outcomes were tested at the appropriate level for was no need to utilize the discrepancy score for each at-
which data were tracked (e.g. survey results for the 39 sub- tribute, since the discrepancy scores did little more than
divisions that had productivity data compared to produc- reflect the fulfillment scores. Instead, we utilized the ful-
tivity outcomes for those same sub-divisions). Correlations fillment scores as our independent EVP measures since
were calculated to estimate the relationship between each this is where we saw the greatest variability.
EVP attribute measure (importance, fulfillment, and dis-
crepancy) with our four employee engagement measures Moving forward, one suggestion for obtaining a more
and four business outcome measures. We also conducted balanced importance score distribution is to have em-
correlation analysis between each engagement measure ployees force rank a fixed number of attributes into
and our four business outcome measures. We used a vari- three categories of importance—critically important, im-
ety of multivariate analyses (e.g., multiple regression, path portant, and less important. This approach, sometimes
analysis) and other structural equation modeling methods referred to as Q-Sort, would expose the “relative impor-
to examine the relationships between all of the employee tance” that employees place on attributes that would oth-
measures and business outcomes. erwise receive a rating of “very important” or “extremely
2. Discretionary effort scores yielded less variance 3. Various relationships between EVP fulfillment
than expected. Although employees showed greater scores and intention to stay. Again, the primary goal of
variability in their responses on the intention to stay and this research was to understand the relationship between
organizational advocacy components of employee engage- EVP fulfillment and intention to stay. As shown in Figure
ment, most respondents rated themselves highly on the 3, we plotted the correlations of the 41 fulfillment scores
discretionary effort measure—reporting that they often and the intention to stay results at the business unit level
go above and beyond what is expected of them. These (614 respondents). We segmented the correlation matrix
results may be due to a few reasons. into four regions to illustrate the strength of the relation-
ships and the potential for improving each fulfillment
First, it is possible that employees within this business unit score. A few of these findings are noteworthy.
do indeed exhibit higher levels of discretionary effort by
willingly exceeding the performance requirements of their First, Region I of the matrix shows attributes that are
jobs. Second, given the resource constraints that high high-impact (higher correlation with intention to stay,
employee turnover has placed on this business unit, it is p = < .0001) and offer the most opportunity for improve-
also likely that many employees may have felt “obligated” ment (fulfillment scores fall below the mean). Although
or“required” to assume the responsibilities of those em- the 13 attributes in Region I provide a logical starting
ployees who have left the organization. As such, our dis- point for developing retention strategies, we recommend
cretionary effort measure may have been tapping into prioritizing them against the criteria of costs, benefits,
employee behaviors that are not as “discretionary” as they and scalability before determining a retention strategy.
may appear. Finally, high discretionary effort scores may For example, of the 13 attributes in this region, 7 of them
be due to social desirability bias—where individuals have are related to the EVP categories of Learning and Develop-
a tendency to over-report socially desirable personal char- ment and Compensation—both of which require the use of
acteristics and under-report socially undesirable character- finite resources (e.g. financial, mentors). As such, these
istics.24 By asking employees to self-report discretionary strategies are restricted in the number of employees that
effort, we were essentially asking them to rate an aspect they can impact. Likewise, providing all employees with
of their performance. As such, employees may have been job fulfillment (r = .48) and challenging work (r = .39) has
inclined to inflate their discretionary effort ratings. limitations since not all jobs lend themselves to variation,
and managers are restricted in the number of employees
To the extent that social desirability bias may have re- they can impact due to span of control. Rather than apply
duced the validity of our discretionary effort results, one these limited strategies broadly, we will further analyze
recommendation for minimizing its effects in the future the survey results of critical employee segments, such as
is to have employees’ self-report discretionary effort high-potentials and mission-critical groups, to determine
using Behaviorally Anchored Rating Scales (BARS).25 Unlike whether these attributes are areas of concern within
the Likert scale used for our study—where employees these populations. Although we plan to identify strate-
answered how often they go above and beyond—BARS gies that will impact the broader population, critical
can be used to capture specific and varying levels of employee segments will be a priority when it comes to
discretionary effort (e.g. volunteers for additional work allocating limited resources—given that these employees
assignments). This approach makes it more difficult for bring disproportionate value to the organization.
One attribute in Region I that we plan to widely address the potential to impact multiple employees simultane-
is resources (r = .37): making sure that all employees have ously, such as leaders providing open and honest communi-
the resources and information they need to do their jobs. cation (r = .43), communication of goals (r = .38), and
This attribute is the cornerstone of rational engagement showing concern for employees (r = .38). Further, this re-
and provides a foundation upon which emotional engage- gion has four attributes related to Performance Manage-
ment can be built. We suspect that resource issues may be ment that can be leveraged to optimize the organization’s
tied to the process efficiency (r = .28) concerns identified in performance management platform. Not only are all of
Region III; deficiencies in these two areas (resources and the attributes mentioned scalable, but they are low-cost
process efficiency) may be what are leading to elevated and likely to sustain returns beyond the initial invest-
workloads (r = .37), see mid-point of Figure 3. Our plan ment since they are aspects of organizational culture.
is to conduct a segment analysis of the survey results to Lastly, there are a few attributes in Region IV that have
pinpoint whether these concerns are systemic or specific higher fulfillment scores, but weak relationships with in-
to a particular employee segment. We will also analyze tention to stay, (p = < .0001). They include: profit sharing
the written responses to the open-ended questions to (r = .18), retirement savings plan (r = .21), competitive health
determine the best corrective actions for improving benefits (r = .21), value of health benefits (r = .22), and paid
these areas. time off (r = .23). We suspect that weaker relationships in
For additional broad-impact retention strategies, we can these areas may be due to a large portion (over 50%) of
draw from attributes in Region II (higher impact and this business unit’s members having low tenure and
higher fulfillment). Although these attributes offer less being younger in age relative to the general workforce.
room for improvement, they have stronger relationships That is, for many of the respondents, the attainment of
with intention to stay than the remaining two regions these attributes may appear to be out of reach and less
(III and IV). We recommend that the leadership team tangible than the shorter-term rewards, such as base
continue to build on attributes in this region that have pay and bonus, which had stronger relationships with
intention to stay. Further, these results are in line with returns on intention to stay. For example, all 11 catego-
external research that suggests the current generation of ries explained 35% of the variance in intention to stay
young employees does not place nearly as much impor- scores, only 3 points above the three-category model.
tance on paternalistic attributes of an employer. 26 As This analysis confirmed that the leadership team should
such, there is no action needed relative to these attri- build its retention strategy on no more than three core
butes regarding this business unit’s retention strategy. areas in order to get a higher return on its investment.
4. A few areas of focus are all it takes to impact in- We conducted additional multiple regressions to identify
tention to stay. In addition to understanding the correla- the optimal number of attributes within the best three-
tion between each EVP fulfillment score and intention to category models that were needed to gain the most
stay, we assessed the impact of multiple EVP attributes potential impact on intention to stay. In general, only two
on intention to stay. Although correlation analysis can attributes in each category were needed to achieve the
help to explain the relationship between an EVP attribute maximum potential impact. The good news is that there
and an employee’s desire to remain with the company, its were several two-attribute options in each category that
limitation is that it does not allow us to understand the can produce a similar increase in intention to stay.
simultaneous impact of multiple attributes on intention These findings are important since they reduce the likeli-
to stay. Since the impact (correlation) of each attribute is hood of over-investing in retention strategies that would
not additive, we utilized multiple regressions to cancel yield a marginal return. If, for example, we relied exclu-
out the overlap or shared variance of multiple attributes. sively on the correlation matrix (Figure 3), we would have
As shown in Figure 4, regression analysis of the 11 been led to believe that improving all four Learning and
overall EVP categories found that most of the variance Development attributes in Region I would have yielded
in intention to stay was explained by 3 categories. For the most potential increase in intention to stay through
example, the most variance that the best single EVP learning and development activities. However, multiple
category (Day to Day Work) accounted for in intention to regressions revealed that improvement in only two of
stay was 23%. That is, improving attributes associated these attributes was needed to achieve the maximum po-
with Day to Day Work can influence an employee’s desire tential increase in intention to stay. Further, it informed
to remain with the company by up to 23%. The most us that any combination of these four attributes would
variance that the best two-category combination (Day to produce a similar result.
Day Work & Learning and Development) accounted for in 5. No relationships between intention to stay and
intention to stay was 29%, an increase of (+6) over the voluntary turnover (at this time). One goal of this
one-category model. The best three-category models research was to test the extent to which our intention to
accounted for 32% (+3) of the variance in intention to stay index correlates with Voluntary Turnover. However,
stay. Many of the best three-category models included: because this study was a pilot, the turnover data available
Day to Day Work, Learning and Development, and Compen- for our research preceded the collection of intention to
sation. Beyond three EVP categories, there were marginal stay data. As such, we were forced to test the relationship