Rise of Analytical Performance

8/3/2019 Rise of Analytical Performance

1/21

The Rise of Analytical Performance Management

Tomas H. DavenportPresidents Distinguished Proessor o Inormation echnology and Management,

Babson College

Sponsored byIn collaboration with


2/21


Table of Contents

Introduction ..............................................................................................1

A Brief Overview of Performance Management Practice ............................2

Te Holy Grail of Analytical Performance Management ............................4

Current Practices in Analytical Performance Management .........................5

wo-Variable Analysis ........................................................................5

wo-Variable Analysis with Controls .................................................6

Tree-Variable Relationships ..............................................................7

Key Variables Used ............................................................................8

Unit of Analysis .................................................................................9

ypes of Data ..................................................................................10

echnology Employed .....................................................................11

A Stage Model of Analytical Performance Management ...........................11

Key Barriers to Analytical Performance Management ..............................13

Summary and Conclusions ......................................................................15

Activity-Based Costing: An Alternative Route to

Analytical Performance Management ....................................................16

About the Author ....................................................................................17Endnotes .................................................................................................18


3/21

1

Introduction

Were seeing a severe downturn in economic ortune or many companies. Banks, automobile companies,

airlines and other rms across multiple industries are announcing record losses. CEOs have no shortage oexcuses rom oil prices to a credit panic to lack o consumer demand. But couldnt these actors have been

anticipated, at least to some degree? Couldnt the actors that truly drive a companys business success be

measured, closely monitored and predicted in such a way that massive downturns could be avoided or at

least moderated?

Imagine the benets i your organization could accurately predict its nancial perormance into the uture.

What i you had solid evidence that some actors aected your business much more than others? What i

your organization was certain that the perormance actors and variables it monitored were actually the right

ones the ones that make a dierence to the organizations nancial success? And what i you had real data

and analysis showing that your companys strategy is actually working (or not working)? Tese benets areoered byanalytical perormance management, a quantitative approach to understanding and predicting

perormance that is a real possibility today or many rms. Although not yet widely adopted, the analytical

approach to perormance management oers so many benets that it seems inevitable, and the organizations

that have adopted it thus ar have achieved consistently positive results.

Perormance management is the use o metrics and indicators, both nancial and nonnancial, to measure,

monitor and improve organizational perormance. Perormance management is hardly a new phenomenon

in business, although it has been propelled in recent years by business intelligence systems that contain and

display perormance metrics that have been extracted rom transactional inormation systems. Perormance

management (sometimes called enterprise perormance management or corporate perormance manage-ment) is a thriving industry, with an academic discipline to support it (management accounting), a class o

inormation systems designed to assemble and display it, and many ollowers o developments in the eld.

Analytical perormance management is the systematic exploration o quantitative relationships among per-

ormance management actors. Te approach allows rms to know not just to speculate which nonnan-

cial perormance variables are associated with nancial perormance. While not very dicult, only a ew

rms have yet applied analytical approaches to perormance management, which means an opportunity or

competitive advantage or those rms that adopt the approach.

In this research report, the concept o analytical perormance management is described both in theory andin practice. Te research is drawn rom interviews at 16 companies that have adopted analytical perormance

management to some degree,1 as well as a global online survey (with a nonrandom sample) o almost 2,500

managers.2 Te company interviews provide insight into leading practices in analytical perormance manage-

ment, and the survey provides an overall perspective on attitudes toward the concept and barriers to achieving it.



4/21

2


A Brief Overview of Performance

Management Practice

Perormance management and business intelligence tools can lead to a broad set o possible activities.Te most typical activity arising out o these systems is reporting o standard inormation at standard re-

quencies. A typical perormance management system might, or example, report sales results by geography

each quarter. More sophisticated approaches to reporting might involve the ability by managers to speciy

and receive custom reports, or even to drill down into high-level data to understand trends, patterns or

anomalies. Perhaps the most sophisticated approach to standard reporting is to present alerts that reported

inormation is outside o previously dened parameters. Even with alerts, however, the ocus is backward-

looking at perormance that has already happened.

According to the global online survey o managers (Figure 1), most organizations practice perormance man-

agement to some degree. Only 7 percent said they did not employ perormance management approaches.However, the survey also suggests that integration o perormance management approaches across a diverse

rm can be dicult. While 37 percent o respondents said they had integrated perormance management

across the entire organization, an even larger number, 56 percent, said their organizations practiced peror-

mance management, but only in some parts o their organizations or not in an integrated ashion. Firms may

nd it dicult, or example, to achieve agreement on which actors to measure and how to measure them.

Figure 1

Question: What is the current status o perormance management in your organization?

In corporate perormance management, the greatest advances in recent years involve reporting on nonnancial

perormance metrics as well as the more traditional nancial indicators. 3 Nonnancial metrics, which

address such domains as customer satisaction and loyalty, employee engagement and satisaction, the

perormance o key business processes, and such intangible attributes as brand equity and the capability

to innovate, have prolierated in recent years as a means o understanding and predicting nancial peror-

mance. Because reporting on past nancial perormance is at best a orm o damage control, companies are

striving to develop measures that assess the perormance o strategic initiatives and potentially provide early

warning o uture nancial problems, or early indication o success. As yet, however, organizations are still

37%

7%24%

32%

IntegratedacrosstheentireorganizationNotcurrentlyusedinmyorganization

PracticedinsomeareasoftheorganizationPracticedthroughouttheorganization,butnotintegrated

(37%)

(7%)

(24%)

(32%)


5/21

3


in the early days with respect to the use o nonnancial metrics. No external authorities require that non-

nancial metrics be reported, although researchers have long argued their importance, 4 and some standards

bodies have debated whether their reporting should be made mandatory.5

A major step in the relationship between nonnancial and nancial metrics, however, came with the devel-

opment and relatively broad adoption o the balanced scorecard. Tis idea, developed by managers at semi-

conductor rm Analog Devices and ormalized and elaborated upon by academic Robert Kaplan and consul-

tant David Norton, is simple in concept but has been more dicult to implement. Te idea o a balanced

scorecard suggests that nonnancial and nancial metrics be given equal weight, and that they be displayed

side-by-side or even on the same page. Tis balanced presentation ostensibly allows managers to easily see

trends in both nonnancial and nancial perormance, and to begin to observe (at least in rough orm) some

drivers o nancial perormance.

Balanced scorecards, however, are not sucient to understand perormance analytically. As Kaplan andNorton have argued, simple presentation o multi-dimensional data on a scorecard display is not sucient

to understand relationships among nonnancial and nancial variables. o begin to address this problem,

Kaplan and Norton proposed that organizations employ a strategy map,6 which demonstrates cause-and-

eect relationships in a logical ashion. Te strategy map is both a visual representation o an organizations

strategy and a representation o how items in a balanced scorecard are linked to each other. It is thus desir-

able as a means to link strategy to perormance management, as well as a hypothesis or what actors are

driving nancial perormance.

In the global online survey o managers (Figure 2), one question addressed to what degree organizations are

currently employing these approaches to perormance management. Among business executives (other thannonprot managers and management consultants, who were excluded rom all analyses reported in this

paper) who answered the question, the responses are below:

Figure 2

Question: Which o the ollowing does your organization use in managing perormance?

(27%)

(12%)

(34%)

(13%)

(14%)

34%


6/21

4


Tese results suggest that despite many years o advocacy by Kaplan and Norton, strategy maps, and even

single organizational scorecards, are still somewhat rare. However, strategy maps are not the ultimate tool or

understanding the relationships between nonnancial and nancial perormance actors, and they were not

intended to be. Kaplan and Norton argue in a review o organizations management systems that the nalstage o the system should be to test and adapt strategy. Tey argue that Companies, especially those with

large numbers o similar operating units, can use statistical analysis to estimate correlations among strategy

perormance numbers. Such analysis will usually validate and quantiy links between investments in, or

example, employee skills or I support systems, and customer loyalty and nancial perormance.7

Tis type o analysis is what I reer to as analytical perormance management.

Despite this encouragement and a variety o potential benets, the actual use o analytical perormance

management approaches is inexplicably somewhat rare. Relatively ew companies have employed statistical

analysis to correlate nonnancial and nancial perormance variables. In the survey data (Figure 3), just over

a third (36 percent) o managers said that they analyzed the quantitative relationships among nancial andnonnancial indicators o perormance, and our interviews with companies suggest that this is a liberal inter-

pretation by survey respondents. Te same percentage said they dont analyze relationships among indicators

at all, and a slightly smaller percentage (28 percent) said they analyze logical relationships among indicators,

as in a strategy map.

Figure 3

Question: How does your organization assess the relationships among fnancial and nonfnancial

perormance indicators?

The Holy Grail of Analytical Performance Management

Te idea behind analytical perormance management is that companies would create statistical models

o cause-and-eect perormance relationships. Tat is, they would create correlation or regression models

that link a variety o nonnancial variables to one or more nancial variables. In an ideal world, they would

consider or control or all possible variables that might have a substantial eect on nancial perormance,

including customer relationships, employee attitudes and behaviors, the level o innovation, the value o

brand equity, and other domains. Tere would be one overall equation that described the relative contributions

(36%)

(36%)

(28%)


7/21

5


o each perormance driver at an aggregated enterprise level and allowed statements such as, With other peror-

mance drivers held constant, a one-point rise in customer satisaction yields a $200 million increase in prot.

Imagine the benets o such a model. Because company executives would know which nonnancial actorsactually drove perormance, those actors would be the ones measured and present on scorecards and reports.

No longer would organizations gather and report metrics simply because they are amiliar, or because a stan-

dard balanced scorecard ormat suggests them. Business strategies, at least those explicitly addressed through

strategy maps and measureable variables, would be testable. Firms would also be able to compare key units

on their achievements on key metrics. Te ability to ocus on the measures that really matter to perormance

would make managerial lie considerably easier.

Given an explanatory quantitative model o perormance, rms would also be able to predict the impact o

increases or decreases in nonnancial variables on nancial perormance. Eventually, external organizations,

such as Wall Street analysts, would be able to make similar predictions (though this could be either positiveor negative or company valuations). Finally, as Kaplan and Norton note, companies could determine

whether their key strategies are actually producing the desired results.

Regardless o these maniold benets, most eorts to manage perormance analytically in the real world have

allen considerably short o the holy grail. However, enough rms have undertaken analytical perormance

management to show that it is indeed possible. In the next section I describe the actual state o the art.

Current Practices in Analytical Performance Management

Over the last decade or so, rms have begun to explore the quantitative relationships between nonnancial

and nancial perormance variables. We interviewed 16 rms that had undertaken some orm o analytical

perormance management, so we can point out emerging best practices.

wo-Variable Analysis

In the early orms o this analysis, the relationships analyzed involved only two variables, relating one non-

nancial perormance actor to a single measure o nancial perormance. Examples o this sort o analysis

include the ollowing:

HiltonHotelsconcludedusingveyearsofdataat42hotelsthata5percentimprovementincustomer

retention (customers who say they are likely to return to Hilton) would result in a 1.1 percent increase

in annual revenues the ollowing year at a typical property (and Marriott has ound similar relationships

in its own data).8

HarrahsEntertainmentfoundthatforeach1percentitgrewitsshareofcustomergamingbudgets,

its share price increased by $1.10.

BestBuydiscoveredthatforeverytenth-of-a-point(onave-pointscale)increaseinemployee

engagement at a particular store, it increased operating income at the store by $100,000. 9


8/21

6


VictoriasSecretfoundthatraisingitsaverageconversionrate(thepercentageofcustomerswhoenterthe

store and actually buy something) by 1 percent brought more than $35 million in sales and more than

$15 million in operating prot.

Since these are only two-variable relationships and other possible causal actors are neither included in the

model nor controlled or, it is possible that these are spurious relationships. However, the power o such

statements is obvious. Companies can concentrate on increasing the independent variable with some con-

dence that nancial perormance will improve.

wo-Variable Analysis with Controls

he most sophisticated companies in analytical perormance management have begun to employ multiple

variables to explain and predict nancial perormance. While the models may include several types o vari-

ables as controls, most o these rms are still attempting to explore a particular two-variable hypothesis. For

example, oronto Dominion Bank has examined the relationship between customer service and the nancialperormance o its retail branches. Dennis Campbell, a proessor at Harvard Business School, has document-

ed the banks eorts in a series o case studies.10 While the banks analyses control or other variables (such as

the average income o the customers in the geography served by the branch), the primary ocus is on the ser-

vice-protability relationship. Te bank ound that the analysis provided management insights that were

useul to inorm potential actions. For example, the analysis revealed that customer service dierentials

among branches explained 19 percent o the variation in branch protability. As a result, the bank began to

reward branch personnel or meeting service targets. However, the bank also ound that the service improve-

ments only aected protability in the middle o the distribution; hence branches that were particularly low

or high in service levels were not eligible or service-incentive payments. oday, however, oronto Dominion

still measures service levels extensively, but no longer statistically relates them to branch protability.

Another example o a primarily two-variable relationship with controls was explored at Store 24, a conve-

nience store chain in New England. Te companys managers had adopted both a balanced scorecard and a

strategy map, and had gathered metrics on each key variable in the scorecard. Te executives had also adopt-

ed a strategy that clearly stated relationships between key perormance variables. It was based on the belie

that an entertaining service experience led customers to buy more in stores. Called the Ban Boredom strat-

egy, it involved placing displays in stores on entertaining topics (e.g., popular movies), about which store

clerks were supposed to engage customers in dialogue. Te chain had gathered measures o how well each

store had implemented the entertainment strategy, and how customers reacted to the stores service.

Measures o the skill o the store-level teams were also compiled. While the companys managers themselvesdid not explore the relationships among customer service variables and store perormance, a group o aca-

demic researchers did analyze the data.11

Te results o the analysis revealed that Store 24s strategy was not working. Te greater the likelihood that a

store had implemented the entertainment strategy, the lower its protability. Te analysis controlled or key

variables that might have infuenced the outcome, e.g., the demographics and income levels o the stores

neighborhoods. Te analysis did reveal that when store employee skill levels were high, the strategy did

work. However, the overall level o skills was suciently low to make the strategy unsuccessul. Store 24


9/21

7


executives abandoned the strategy ater two years, but the researchers analysis ound that the strategy could

have been ound to be ineective ater analyzing only one years worth o data.

Tree-Variable RelationshipsWhile Store 24s and oronto Dominions models were primarily on two-variable relationships with controls,

some rms have ocused on three-variable relationships. In particular, the service-prot chain hypothesis,

adopted by several retail and service rms and explored in greatest quantitative depth by Sears in the 1990s,

species that employee satisaction and engagement drive customer satisaction and loyalty, which in turn

leads to improved nancial perormance. Tis hypothesis has been widely adopted as a management model 12

but has not oten been tested in statistical models.

Sears, however, did test what it called the employee-customer-prot chain model extensively between 1993

and 1997, a period that coincided with a dramatic nancial turnaround at the retailer.13 Te companys

eorts are among the most extensive in the brie history o analytical perormance management. Developingthe statistical relationships among nonnancial and nancial perormance variables proved to be among the

easier tasks at Sears. More dicult, according to the leaders o the project, were building management align-

ment around the model and the measures, and educating ront-line service personnel about the model and

its implications. While the analytical approach to perormance management oered signicant benets to

Sears, it is not in use at this time (though Eddie Lampert, the current primary owner o Sears Holdings,

is reputed to be a believer in and heavy user o quantitative testing approaches).

Several other companies have experimented with analytical perormance management, although there seem

to be ew approaches that are as broadly implemented as those described above. One large coee retailer, or

example, has just begun to explore what it calls linkage among employee, customer, and nancial peror-mance indicators. Inosys, the India-based proessional services rm, makes extensive use o the balanced

scorecard approach and uses an analytic hierarchy method to subjectively weight the importance o dier-

ent perormance metrics. Inosys also did a one-time statistical analysis using 16 quarters o data to under-

stand the primary drivers o growth and protability outcome metrics. Te companys strategy and planning

analysts ound the analysis useul in identiying important metrics, but it has not yet become an ongoing process.

Te act that the companies that produced the best examples o multivariate analytical perormance manage-

ment do not continuously practice it suggests that usage o the approach may be episodic even when it

leads to substantial perormance gains. Sears adopted analytical perormance management approaches

because o its need or a service-led turnaround. It accomplished the turnaround, moving rom desperatestraits (losing $3 billion) in 1992 to being named the most innovative general merchandise retailer by

Fortune in 1999. Store 24 was acquired by another convenience store chain, which may provide a reason

or its lack o continued analysis. oronto Dominion embarked upon its initiative ater a merger (with

Canada rust) brought a dierent set o perspectives among the newly-merged organizations about the

nancial value o customer service. It achieved many o its goals in both customer service (metrics up every

year since the 2000 merger) and nancial perormance (D Canada rust has achieved 20 quarters o double-

digit earnings growth year over year). For each o these organizations, however, new business contexts and

new managers led to dierent priorities. While they maintain a strong interest in metrics and perormance,

they do not currently eel the need or statistical analysis o the drivers o nancial perormance.


10/21

8


Key Variables Used

Customer relationship-oriented variables tend to be the most commonly employed by rms exploring

analytical perormance management and were the highest-rated among a variety o nonnancial measures

in terms o importance in driving nancial perormance in the survey (Figure 4). Customer attitudes andbehaviors may include customer satisaction (highest-rated in the survey by respondents, and most common-

ly used, as shown in Figure 5), customer loyalty (as expressed by share o wallet analysis, lietime customer

value or other metrics) or the net promoter metric14 that has gained recent popularity. One rm had

attempted to correlate net promoter score with nancial perormance at the business-unit level, but had

ound that initial correlations broke down under detailed study. Such customer-oriented variables are most

commonly assessed either through loyalty programs or customer intercepts and surveys.

Figure 4

Question: On a scale rom 1 to 5, please rate how important you believe each o the ollowing key indicators is in

driving fnancial perormance and indicate which ones, i any, your organization measures.

Customer satisaction 4.55

Customer loyalty 4.42

Employee satisaction/engagement 4.13

Product/service innovation capability 4.02

Brand reputation 4.29

Sustainability 4.06

Figure 5

Question: Please indicate i these items below are measured in your organization.

Customer satisaction 81%

Employee satisaction/engagement 71%

Brand reputation 62%

Customer loyalty 59%

Sustainability 48%

Product/service innovation capability 46%

Te other common set o nonnancial variables involve some orm o employee attitudes, including employ-

ee satisaction and engagement, and (less requently) human capital management.15 Employee attitudes are

typically assessed through periodic surveys. Both types o data are requently sourced rom external rms,

such as Gallup or employee engagement and the Service Management Institute or customer intercepts.

Human capital management involves not only employee attitudes, but a complex set o variables that may

include learning-oriented variables and even knowledge management. American Standard, or example, has

done considerable work relating various measures o human capital to growth in prots at sales oces and

to lower accident rates at manuacturing plants.16


11/21

9


Other nonnancial variables are less commonly correlated to nancial perormance. One rm we inter-

viewed had experimented with a measure o innovation (as expressed by the percentage o revenues rom

new products and services that exceed a certain threshold level) as a predictor o overall revenues and prots.

A technical services organization ound the average time it takes to close a case to be a strong predictor ogross margins. An oil rener ound renery uptime closely correlated to prots. A proessional services rm

ound that macroeconomic and client nancial indicators were strong predictors o its own nancial

perormance.

Academic researchers have addressed a variety o other nonnancial predictors o nancial perormance,

though we know o no companies that have employed these variables in a systematic approach to analytical

perormance management. Tey include marketing and marketing productivity,17 brand equity,18 inorma-

tion technology capabilities19 and innovation.20

Perhaps obviously, the choice o variables and relationships to explore should be driven by an organizationsstrategy. Just as the Store 24 example above illustrates, analytical perormance management models should

be viewed as tests o key business relationships that strategies have posited or hypothesized. Ater strategic

relationships have been tested, organizations should move on to test the importance o variables that arent

deemed to be strategic, but might have some relevance to perormance anyway.

Unit of Analysis

Te goal o analytical perormance management is to yield analyses and results that are interpretable and can

be acted upon. Te idea o a single aggregated corporate model incorporating all relevant nonnancial driv-

ers o perormance is an appealing prospect in analytical perormance management. However, it is dicult

to apply in large, complex organizations. Interviews among early adopters o analytical perormance manage-ment suggested several problems with corporate-level analysis. First o all, there would probably be too many

actors to easily interpret. Secondly, relationships between variables would probably become weaker at higher

levels o aggregation. Any relationships that exist, or example, between employee satisaction and nancial

perormance would be weakened, i not washed out altogether, by the inclusion o employees who do not

work directly with customers and whose engagement (or lack thereo) might not aect nancial results.

Understanding and acting on analytical perormance models is easiest when there are clear, relatively local

units o nancial perormance to which nonnancial measures o perormance can be tied. In a retail chain,

the relevant unit would be the store or perhaps a regional grouping o stores (as at Sears and Store 24). In

banking, branches orm a natural local unit o analysis (as at oronto Dominion). In a manuacturing com-pany, the relevant units might be sales oces (as at American Standard).

I the company doesnt have such natural units, it probably only makes sense to do analytical perormance

management or business units with similar employee, customer and product types. Otherwise too many

dierent actors will be mixed in or the results to be interpretable. At Cisco Systems, or example, managers

dealing with customer service metrics realized that it did not make sense to combine corporate customer metrics

rom its enterprise telecommunications equipment business with those or its Linksys consumer division.


12/21

10


Businessunit/department/divisionlevelCorporate/organizationlevel

Localfacility,e.g.,storeorbranchlevelOther

Surprisingly, however, in the survey the local acility was the least common unit o analysis supplied by

respondents when they were asked at what level they analyzed relationships among nancial and nonnan-

cial indicators (Figure 6). Only 7 percent analyzed relationships at that level, whereas 28 percent attempt-

ed to analyze them at the corporate level, and 63 percent at the business unit, department or division level.Many o these organizations, however, may be analyzing relationships logically rather than statistically,

so the problems cited above may not come into play.

Figure 6

Question: At what level do you analyze the relationships among fnancial and nonfnancial perormance

indicators?

ypes of DataTere are two dierent types o data that can support analytical perormance management. Te ideal typeis time series data. Tis would involve capturing data on all relevant perormance measures over a substantial

period o time perhaps several years o quarter-by-quarter measurements. ime series data would allow

organizations to determine whether there are lag eects or particular variables in their infuence on nancial

perormance. Sears, or example, benetted greatly rom the analysis o lag eects:

We were able to demonstrate that the lag indicators (a compelling place to invest) are driven

by the lead indicators (a compelling place to shop and work). We were able to explore how

these actors were related to each other within specic time periods and across dierent time

periods. wo-quarter lag analysis showed that it takes a while or the customer to see that we

are doing something dierent beore we are able to achieve measurable nancial results.

21

Te alternative to time series data is to employ cross-sectional data. As the previous section describes, this

would involve data on perormance across a range o units stores or branches, or example. Tis would

allow statistical models to identiy the variables that explain variation across perormance in the local units.

As noted above, most organizations that have adopted analytical perormance management have used cross-

sectional data. Tis requires that the local units measure nancial and nonnancial perormance variables

in the same way.

63%28%

7% 2%

(7%)

(63%)

(28%)

(2%)


13/21

11


echnology Employed

Most o the early adopters o analytical perormance management have done the analysis in a somewhat

ad hoc ashion. Tat is, they did not yet have an institutionalized, recurring approach or analyzing the data

rom perormance reports or scorecards. Tereore, it was not particularly important that analytical capabili-ties were combined with typical reporting or scorecard display capabilities in their analytical perormance

management systems.

However, or organizations that wish to practice analytical perormance management approaches on

an ongoing basis, it would be important to employ technologies that allow both reporting and statistical

analysis in an integrated ashion. It would also be useul to have these capabilities integrated with other

approaches to perormance management, such as activity-based costing (see sidebar on that topic at the end

o this paper). Tis would be in addition to the usual prerequisites or business intelligence, which include

clean, integrated data that is accessible in some sort o data warehouse or data mart.

A Stage Model of Analytical Performance Management

Based on the activities o companies, a clear progression o activity emerges with regard to analytical peror-

mance management. Figure 7 depicts a ve-stage model o activity related to analytical perormance. It ranges

rom highly prosaic perormance management situations (inability to le required nancial perormance

reports) to the most sophisticated ones (incorporation o analytical perormance management into decisions

and actions).

Figure 7 A Sta Modl Of Analytcal Pfomanc Manamnt

Figure 8 includes a list o key activities to move rom one stage to another. o master Stage 1, a company

needs only to master having accurate and timely nancial reports. Not all companies are able to do this; wit-

ness the number o restatements and late lings (or example, a General Accounting Oce study ound 919

restatements among US rms between 1997 and 2002).22 Most public rms, however, can master Stage 1.

Stage 5

Incentives and Actions

Stage 4

Analytical Model

Stage 3

Strategy Map

Stage 2

Balanced Scorecard

Stage 1

Financial Reports


14/21

12


Figure 8

rqmnts to mov fom sta to sta

Stage 1 to Stage 2 Identifyandagreeuponnonnancialmeasures.

Collectandreportnancialandnonnancial

measures in scorecard ormat at regular intervals.

Stage 2 to Stage 3 Discussandidentifymission,visionand

customer value proposition.

Debateandagreeupontheperformancedrivers

o the business.

Stage 3 to Stage 4 Collectconsistentscorecarddataoveranextended

period (at least a year).

Turnstrategymapintoaquantitativehypothesisand test it with data.

Stage 4 to Stage 5 Embedtheresultsofanalyticalperformance

management into perormance assessment and

compensation decisions.

Educateemployeesandmanagersaboutthe

actors that truly matter to perormance, and

the relationships with nancial perormance.

Stage 2requires the development o nonnancial measures, and their integration with nancial measures

on a common scorecard. While many companies have developed balanced scorecards, doing so requires

an organization to agree on a set o nonnancial perormance measures, to decide which measures are most

important to report and manage, and then to gather the metrics on a regular basis. Once a company has

reported measures on a scorecard or an extended period (the Store 24 example suggests that some results can

be achieved with one year o data), it has the data in place or a more analytical and quantitative approach.

Many companies are in this position, so they have the building blocks in place or analyzing the data.

Stage 3 requires the organization to hypothesize the logical relationships among scorecard elements in a

strategy map. Strategy maps are the articulation o the organizations business strategy in the orm o itsmission, vision, customer value proposition and key nonnancial perormance actors. Since the relation-

ships are only hypothesized, this stage is not extremely dicult, but it does require senior management teams

to discuss and debate their strategy and the drivers o their business. Te survey data in Figure 2 suggest that

14 percent o respondents have achieved this level o analysis.


15/21

13


Stage 4builds on the perormance measures, accumulated data, and hypothesized relationships rom the

earlier stages to create a statistical model o the linkages between key nonnancial and nancial perormance

variables. Tis is treated as one stage, but it will undoubtedly require experimentation and iteration over

time. Te ultimate objective is an analytical perormance management model that is statistically robust, thatclearly identies the key actors that drive nancial perormance and the relative weights o those actors, and

that can be interpreted and acted upon by management. Although the survey suggests that 36 percent o

organizations that analyze relationships between perormance measures do so in a quantitative ashion, our

interviews and experience would suggest that the number o rms at this level is much smaller. Several o the

rms mentioned above, including Store 24, Hilton and Marriott, Best Buy, Inosys and Harrahs, could all

be considered to have reached Stage 4 or some period o time.

An analytical model is o little value unless acted upon. Firms at Stage 5educate managers and employees

about the relationships in the model and apply the models results and implications to decisions and actions.

In short, they use analytical perormance management to actually improve perormance. I, or example, themodel suggests that employee engagement drives perormance, managers will take steps to ensure that

engagement stays at a high level throughout the organization. I customer satisaction is a primary driver,

managers should ensure that employees are aware o the importance o customer satisaction and are incent-

ed to maintain it. Te results rom perormance management analyses allow managers to optimize their

resources and ocus attention on the aspects o the business that truly matter. While ew rms have achieved

this level, those that have achieved very positive results. Sears clearly achieved Stage 5 during the mid-1990s,

when it widely educated employees about the actors driving company perormance, and it assessed and

compensated based on the results o its model. oronto Dominion Bank also achieved Stage 5 when it began

to compensate employees based on the results o its model linking customer satisaction to nancial peror-

mance in branches.

However, once a company reaches Stage 5, it may not stay there over time with respect to a particular set o

measures and models. Te actors that drive business perormance will change as the business environment

changes, so rms will periodically need to go back to Stage 2 and perhaps even to Stage 1 to ensure that

they have the right perormance measures in place, and then the right scorecards, strategy maps and analyti-

cal perormance models based on those measures. Analytical perormance management is not a one-time

journey, but rather a continuous process o renement and change.

Key Barriers to Analytical Performance ManagementTere are several actors that prevent more organizations rom undertaking analytical perormance manage-

ment. For many organizations, simply not recognizing that analytical perormance management is possible

is a barrier. For others, not having sucient business context to overcome the diculty o these approaches

may stand in the way. Te survey data (Figure 9) suggest that there are many actors o relatively equal

weight that stand in the way o analytical perormance management.


16/21

14


Figure 9

Question: What are the primary barriers preventing your organization rom adopting analytical,

quantitative perormance management (click on all that apply)?

Questionable data quality/accuracy/consistency 39%

Disagreement on perormance indicators/metrics 37%

Inability to aggregate data rom various sources 31%

Insucient data 30%

Lack o executive support or interest 28%

Lack the skills required 28%

Metrics and data are a problem or many organizations, and many o the problems cited in the survey dealwith that set o issues. Not having agreement on key metrics across the organization is one barrier; an enter-

tainment company we interviewed, or example, suggested that nonnancial perormance measures varied

widely across the organizations business units, and no one was responsible or bringing about more com-

monality. Even i there is agreement on what metrics really matter, gathering them can be dicult and

expensive. Data quality was the most requent issue cited in the survey as a barrier. Other organizations in

our interviews mentioned data requency as an issue. A retailer noted, or example, that it is dicult to get

requent enough measures o employee engagement and satisaction; ew organizations gather such data

quarterly. One electronics retailer had increased the requency o employee engagement surveys rom annu-

ally to semiannually.

Lack o analytical skills is an issue or some organizations. Several organizations we interviewed elt that they

had no in-house employees who understood both the metrics and the relevant statistical techniques.

However, Sears and oronto Dominion employed third-party analytical consultants, and the Store 24 work

was perormed by academic researchers. It is relatively easy or any rm to nd such external help.

Not quite so easy is nding employees who can do analytical work and whose organizational aliation cuts

across relevant unctions. One retailer that was beginning analytical perormance management work had a

group o behavioral/organizational psychologists in human resources. While the analysis o perormance

management variables also included customer satisaction, there was no one with the relevant analytical skills

whose responsibilities included customer relationships and employee attitudes.

We ound no group in any o the companies we interviewed whose responsibilities encompassed nonnan-

cial perormance measures and their relationship to nancial perormance. One would expect that such

responsibilities would be ound within an accounting or nance organization, but this seems not to be the

case, at least in the great majority o rms. Perhaps in the uture organizations will create new roles in nan-

cial organizations that can address all these issues.


17/21

15


Summary and Conclusions

Analytical perormance management is clearly possible, and it oers substantial potential benets to those

ew organizations that have undertaken it. It isnt particularly dicult rom a quantitative analysis stand-point, and it oers the potential o conrming or disconrming key strategy propositions.

While it is not yet common practice, many companies more than 40 percent in our survey say they are

denitely moving in a more analytical direction on perormance management (Figure 10). Another 28

percent say they would like to become more analytical. While respondents to this survey (who are probably

unusually interested in perormance management issues) are likely to be more aggressive in implementing

new perormance management approaches than the average manager, the survey results suggest that given

better tools, more widespread awareness o benets, and greater understanding o methods and approaches,

substantially more organizations will practice analytical perormance management in the uture.

Figure 10

Question: Please indicate which o the ollowing best describes your organizations current position on using

analytical approaches to perormance management. By analytical, we mean the use o quantitative

inormation such as statistics to draw conclusions.

Greater adoption may require new strategies or gathering data over time and across local units. It may also

require senior executives to address perormance management topics in ways they do not currently employ.

In any case, because gathering the data is the most time-consuming aspect o analytical perormance man-

agement, it behooves organizations to start now in gathering data and beginning to analyze it. Similarly,

rms should begin creating the skills to analyze and interpret perormance data, and create either temporary

or permanent capabilities. Ultimately, to practice analytical perormance management only requires a strong

desire to better understand the drivers o business perormance, and the straightorward application o

resources toward that goal.

Te motivation or adopting analytical perormance management is perhaps obvious, but it bears repeating.

Given the possibility o knowing what measures really matter, and the ability to understand and predict

nancial perormance, rms cannot aord to continue with major revenue and earnings surprises particu-

larly those involving losses. Until these tools are employed, corporate perormance management will remain

an art rather than a science. Only when it becomes a science will rms get control over their nancial

perormance.

(14%)

(28%)

(17%)

(40%)

(1%)


18/21

16


aking a scientic approach to perormance management is also not a one-o exercise. Companies that have

tried analytical perormance management did so because they needed to improve dramatically. Ironically,

once the improvement was made, ew embedded it as a regular activity perhaps thinking all lessons were

learned. But as markets and the economy change, nothing can be taken or granted. Rules and traditionsare being replaced every day. Yesterdays lessons may have no meaning today. Te only way to know whats

driving your business is to analyze and prove it not just once, but continuously. Otherwise, like Store 24,

by the time you realize the strategy isnt working, it could be too late, or have cost you dearly.

With much o the global economy currently acing an economic downturn, perhaps we will see a resurgence

o organizations looking to analytical perormance management or help. Te techniques are proven and

possible so whats stopping you?

Te approach outlined in this report takes a

high-level perspective on analytical perormance

management, relating nonnancial metrics at a

corporate or unit level to the same units nancial

perormance. Its also possible and desirable to

examine perormance analytically at the level o

business transactions and individual customer

relationships. Tis more granular approach can

complement the top-down approach describedin this report. For example, a top-down analysis

may indicate that the strength o customer rela-

tionships is a key driver o nancial perormance.

However, its also necessary to ensure that cus-

tomer relationships and transactions are prot-

able i the organization is to prosper nancially.

Activity-based costing (ABC) is the recommended

tool or the more granular analysis o transac-

tions and relationships. ABC analysis assigns

an organizations costs (both direct and indirect)

to the activities it employs to make products

and services, and serve customers. While ABC

work is not statistical in nature, it is clearly

analytical, with careul methods or assigning

costs to activities.

Trough diligent ABC analysis, a company

can determine whether individual customer sales

transactions, and overall relationships with custom-

ers, are protable. It can also assess how to make

customer relationships more protable, e.g., by

steering customers to certain marketing channels

or touch points or by discouraging relationships

with overly demanding customers.

For example, the Canadian telecom provider

Fido Solutions (now a subsidiary o Rogerselecommunications) analyzed its product oer-

ings using ABC and then mined its customer

data to determine which customers were most

and least protable. Fido stopped trying to retain

its least protable customers and saw the number

o unprotable ones drop by hal. Te analysis

allowed Fido to ocus its resources on the

customers most likely to contribute to its

nancial perormance.

As at Fido, ABC work complements top-down

analytical perormance management by pinpoint-

ing how costs build up in creating products and

serving customers, and oering managers ways

to reduce costs and become more protable.

Most organizations, thereore, should employ

both approaches.

Activity-Based Costing:An Alternative Route to Analytical Performance Management


19/21

17


About the Author

Voted the third leading business-strategy analyst (just behind Peter Drucker and om Friedman) inOptimize Magazine, Tomas Davenport is a world-renowned thought-leader who has helped hundredso companies revitalize their management practices. He combines his interests in business, research, and

academia as the Presidents Distinguished Proessor in Management and Inormation echnology at BabsonCollege. om earned a Ph.D. rom Harvard University in social science and has taught at the HarvardBusiness School, the University o Chicago, Dartmouths uck School o Business, and the University oexas at Austin. He has also directed research centers at Accenture, McKinsey & Company, Ernst & Young,and CSC. Sample audience reviews include:

Te smartest business speaker out there! Had the audience on the edge o their seats! Energized.A rock star! I learned more rom this speaker in one hour than rom any Ive ever heard.Appropriately provocative.

An agile and prolic thinker, om has written or co-authored twelve best-selling business books and hasbeen a creator and early author or several key business ideas including: knowledge management (our books,

one the best-selling Working Knowledge); human approaches to inormation management (two books);business process reengineering (on which he wrote or co-authored the rst article, the rst book, and the rstcasebook); and realizing the value o enterprise systems. oms book, Te Attention Economy, was namedone o the ten best books o 2001 by Amazon.com and by Borders.com and was the winner o the Library

Journalaward or one o the best business books o 2002. Whats the Big Idea: Creating and Capitalizing onthe Best Management Tinking, was published in May 2003 and has become a top 10 best-seller in severalcountries. Published by Harvard Business School Press in September 2005, Dr. Davenports Tinking or aLiving: How to Get Better Perormance and Results rom Knowledge Workers, is an Amazon bestseller.

Whats next? om has conducted a major research study on Competing on Analytics: How Fact-BasedDecisions and Business Intelligence Drive Perormance. Te related article, which was published in the

January 2006 Harvard Business Reviewdecision-making issue, was the organizations bestselling 2006 reprint.

Te much anticipated book on the topic, Competing on Analytics: Te New Science o Winning, publishedMarch 6, 2007, is already in its ourth printing. A related Harvard Business Review case study appeared inthe May 2007 HBR with another book on the topic to ollow in 2009.

om was one o the rst management thinkers recruited to blog or Harvard Business Online and his TeNext Big Ting blog is a reader avorite. om has written over 100 articles or such publications as HarvardBusiness Review, Sloan Management Review, Caliornia Management Review, and the Financial imes, and isquoted requently in Te Wall Street Journal, New York imes, Business Week, Fortune, Business 2.0, the BostonGlobe, and Fast Company. He is also interviewed requently by the broadcast media.

In the Spring o 2008, om received two impressive honors. CIO Insight named his Competing onAnalytics book one o the all-time op 15 Most Groundbreaking Management Books and Zi Davis once

again included him as one o only our I management thought leaders on their 100 Most InfuentialPeople in I list. om has been named one o 10 Masters o the New Economy byCIO Magazine, one o25 E-Business Gurus byDarwin, one o the most trusted consultants byOptimize Magazine, and one othe top 25 consultants in the world byConsulting Magazine.

With his vast storehouse o industry stories, research & data, and cutting edge ideas, om Davenport bal-ances research-based business acumen with practical application. His areas o expertise include improving theproductivity o knowledge workers, inormation and knowledge management, attention management, ideageneration, innovation, competing on analytics, managing enterprise applications or business value, andbusiness process reengineering. More inormation on om Davenport is available at his website,

www.tomdavenport.com.
http://www.tomdavenport.com/http://www.tomdavenport.com/


20/21

18


Endnotes

1 Suzanne Gratton assisted in scheduling and carrying out the interviews. SAS provided nancial support or the research and commissioned an online

survey. Monet Meek and Jonathan Hornby of SAS advised the study and were particularly helpful. Dennis Campbell o Harvard Business

School, Nick Bontis o McMaster Business School, Jeanne Harris o Accenture, and Len Schlesinger o Babson College also provided useul adviceon the paper.

2 Te survey was conducted by Harvard Business School Publishing in June and July o 2008. Respondents were recruited through Web site advertisements

and oered a ree article i the viewer clicked on the ad and participated in the survey. A total o 1,060,497 Impressions displayed the message in dierent

sizes and placements and 4,248 clicks to the survey were recorded. O the 4,248 clicks to participate in the survey, 2,769 surveys were completed. A lter

was applied to the data and excluded responses that answered Sta as the respondents position, or less than $100,000,000 in company revenues.

Further ltering excluded nonprot organizations and consultants, yielding 1,841 respondents. Te number o valid responses to particular questions

varied considerably.

3 Robert G. Eccles, Te Perormance Measurement Maniesto, Harvard Business Review, January-February 1991.

4 Baruch Lev, or example, has argued convincingly or the reporting o intangible metrics, and that they are important to an organizations market value.

See, or example, Baruch Lev, Sharpening the Intangibles Edge, Harvard Business Review, June 2004.

5 Te International Accounting Standards Board, or example, has published accounting standards or intangible assets (IAS 38) and has a research project

to create a more robust approach to valuing such assets. See http://www.iasplus.com/agenda/intang.htm.

6 Robert S. Kaplan and David P. Norton, Strategy Maps: Converting Intangible Assets into angible Outcomes. (Harvard Business School Press, 2004.)

7 Robert S. Kaplan and David P. Norton, Mastering the Management System, Harvard Business Review, January 2008.

8 Hilton Hotels inormation rom Dennis Campbell, Putting Strategy Hypotheses to the est with Cause-and-Eect Analysis, Balanced Scorecard Report,

Harvard Business School Publishing, September/October 2002, p. 15; Marriott inormation rom personal correspondence with Stephan Chase, Vice

President o Customer Knowledge, Marriott International, June 12, 2008.

9 Scott Flander, erms o Engagement, Human Resource Executive, Jan. 1, 2008, online at http://www.hrexecutive.com/HRE/story.jsp?storyId=59866490.

10 Dennis Campbell and Brent Kazan, D Canada rust A, B, and C cases, Harvard Business School, April 2008.

11 Dennis Campbell et al., esting Strategy with Multiple Perormance Measures Evidence rom a Balanced Scorecard at Store 24, Harvard Business

School Working Paper, February 2008.

12 James L. Heskett et al., Putting the Service-Prot Chain to Work, Harvard Business Review, March/April 1994 (reprinted July 2008).

13 Anthony J. Rucci, Steven P. Kirn, and Richard . Quinn, Te Employee-Customer-Prot Chain at Sears, Harvard Business Review, January-February

1998.

14 Fred Reichheld, Te Ultimate Question: Driving Good Profts and rue Growth. Harvard Business School Press, 2006.

15 Laurie Bassi and Daniel McMurrer, Maximizing Your Return on People, Harvard Business Review, March 2007.

16 Laurie Bassi and Daniel McMurrer, Maximizing Your Return on People, Harvard Business Review, March 2007.

17 Roland . Rust et al., Measuring Marketing Productivity: Current Knowledge and Future Directions,Journal o Marketing68: October 2004, 76-89.

18 Mary E. Barth et al., Brand Values and Capital Market Valuation, in John Hand and Baruch Lev, eds., Intangible Assets: Values, Measures, and Risks.

Oxord University Press, 2003.

19 Anandhi S. Bhardawaj et al., Inormation echnology Eects on Firm Perormance as Measured by obins Q, Management Science45:7, July 1999.

20 Alan Hughes, Innovation and business perormance: small entrepreneurial rms in the UK and the EU. In HM reasury (eds.): Economic growth and

government policy: papers presented at a HM reasury seminar held at 11 Downing Street on 12th October 2000. London: HM reasury, pp.65-69.

21 Steve Kirn, Te Balanced Scorecard at Sears: A Compelling Place or Feedback and Learning, Balanced Scorecard Report, July-August 2000, p. 9.

22 General Accounting Oce, Financial Statement Restatements: rends, Market Impacts, Regulatory Reponses, and Remaining Challenges, 2003. Washington,

DC: GAO-03-138.


21/21

SAS insttt inc. Wold Hadqats +1 919 677 8000 To contact your local SAS ofce, please visit:www.sas.com/ofces

SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. indicates USA registration.

Other brand and product names are trademarks of their respective companies. Copyright 2008, SAS Institute Inc. All rights reserved. 103610_508832.0908

Rise of Analytical Performance

Documents

Transcript of Rise of Analytical Performance