Choose the Right Measures, Drive the Right Strategy

By Dennis Campbell

Article Reprint No. B0605D

HARVARD BUSINESS SCHOOL PUBLISHING

INSIGHT, EXPERIENCE & IDEAS FOR STRATEGY-FOCUSED ORGANIZ ATIONS

Balanced Scorecard REPORT

For a complete list of Harvard Business School Publishing newsletters:http://newsletters.harvardbusinessonline.org

For reprint and subscription information for Balanced Scorecard Report:Call 800-988-0866 or 617-783-7500http://bsr.harvardbusinessonline.org

For customized and quantity orders of reprints:Call 617-783-7626 Fax 617-783-7658

For permission to copy or republish:Call 617-783-7587

TO

OL

S

&

TE

CH

NI

QU

ES

3Copyright © 2006 by Harvard Business School Publishing Corporation. All rights reserved.

Could someone who has neverheard of your company pick up your Balanced Scorecard andarticulate your strategy? Howabout someone within your com-pany who is actually responsiblefor implementing your strategy atan operational level? A hallmarkof a good Balanced Scorecard isthat it translates strategy intounderstandable operational terms.Scorecards can fail to achieve thisobjective if they include too manyor potentially irrelevant metrics.Too many metrics can dilute thefocus on your strategic objectives,making it difficult to communicatea coherent implementation planto the employees responsible for achieving those objectives.Moreover, an abundance of metrics that do not have clearlinkages to your overall strategicobjectives may be symptomatic of a larger problem: the lack ofstrategic focus at the top of theorganization.

Measures selected for inclusion in your scorecard should haveclear and demonstrable links to your overall customer andfinancial performance objectives.Understanding the relative importance of different metrics in driving these objectives is anecessary condition for providinggood, actionable information at the operational level, wherestrategy is actually implemented.

Consider the case of Toronto-Dominion Bank (TD Bank). In a banking market where consumersand regulators were typically hostile to mergers and acquisi-tions, TD Bank, one of the largest

banks in Canada, undertook amerger with a relatively smalltrust company, Canada Trust (CT),which was known for exceptionalcustomer service. To assuage theconcerns of regulators, consumergroups, and the customers newlyacquired through the acquisitionof CT, the bank made severalpublic pronouncements promisingto maintain its high customer ser-vice standards and deliver a “com-fortable” and convenient bankingexperience.

Executives at the bank took thispromise seriously, but were nowfaced with the task of translatingtheir service strategy of “comfort-able banking” into operationalterms. To appreciate their chal-lenge, consider the customer

Choose the Right Measures,Drive the Right StrategyBy Dennis Campbell, Assistant Professor, Harvard Business School

Metrics overload is a common problem that can have serious consequences: specifically, it can make it difficult foremployees to see what actions they should take to executestrategic objectives. Having too many metrics dilutes the focusand invariably means many are irrelevant. Here, accountingand performance measurement expert Dennis Campbelltraces a major Canadian bank’s experience in overhauling itscustomer satisfaction metrics to make them meaningful—and actionable—to frontline employees. The results say it all.

Figure 1. The Bank’s Original Customer Satisfaction Tracking Report

With an abundance of data but little information on the relative importance of these 25-plus metrics, this report was not very helpful to branch-level employees.

4 Balanced Scorecard Report May–June 2006

Choosing the Right Measures (continued)

satisfaction tracking report inFigure 1. This periodic report was meant to provide informationto tellers and branch managementabout their performance againstvarious service attributes that areregarded as drivers of the overallbranchwide objectives of highcustomer satisfaction and loyalty.1

But were there too many mea-sures? And were they the rightmeasures to include in the branchscorecard? Without the answer to these questions, how wouldoperational-level employees knowwhich service attributes to focustheir efforts on to achieve highercustomer satisfaction? The score-card in Figure 1 contains a lot of data, but very little informationthat would help branch-levelemployees implement the bank’scustomer service strategy. Thereare more than 25 performancemetrics included in this scorecardand no information on their

relative importance in driving customer satisfaction.

Overhauling CustomerSatisfaction Metrics

The EVP of marketing, ChrisArmstrong, and his team under-took a rigorous and systematicapproach to analyzing whichdimensions of service warrantedfurther investment and whichshould be de-emphasized. Theirapproach began with ensuringhigh-quality data and metrics.Surveys of customers’ experienceswith each dimension of servicewere taken shortly after a cus-tomer’s actual service interactionat a branch. Individual customersurveys were aggregated up to the branch level to form averagebranch satisfaction scores witheach dimension of service onlyafter a sufficient number of customers had been surveyed for each branch.

Statistical analysis was then used to estimate the impact of measures of each service attribute on customer satisfactionand branch profitability.2 Forexample, the analysis revealed,among other things, that:

• Each increase of 1% in mea-sures related to “comfortablebanking” (e.g., the percentageof customers who were satisfiedthat the teller made them feeltheir business was appreciated)led to a 1.7% increase in customer satisfaction, which in turn led to a 0.4% increase in branch profitability.

• Similarly, a 1% increase in measures related to “speed of service” (e.g., the percentageof customers satisfied that their transaction was processedquickly) led to only a 0.8%increase in customer satisfaction,which in turn led to a 0.2%increase in branch profitability.

• Many service attributes tracked in the branch scorecard showedonly a marginal or even norelationship to overall cus-tomer satisfaction and branchprofitability.

Armstrong and his team werenow able to understand preciselyhow much it was worth toincrease performance in eachdimension of service; they knewwhich aspects of service weremost important and which ones could be de-emphasized.Moreover, they found that mostservice measures showed a highly nonlinear “S”-shaped rela-tionship to customer satisfaction.Improvements in service did notyield incremental increases in customer satisfaction until mea-sured service levels were aboveminimum thresholds, after which they yielded dramatic increases in customer satisfaction. However,once service levels climbed aboveupper thresholds, further improve-ments in service yielded smaller

1. CSI score

2. Satisfied (bottom 3 box)

3. Likely to recommend

4. Likely to continue

5. Make you feel like they appreciate your business

6. Process your transaction quickly

7. Have the ability to handle your request

8. Wait time acceptable

9. Greet you pleasantly

10. Address you by your name

11. Give you his or her undivided attention

12. Thank you for your business

13. Process your transactions accurately

14. Treat you in a respectful manner

15. Conduct your banking privately

86.9

1.9

79.8

90.7

91.6

96.3

96.3

91.6

97.2

77.6

96.3

86.9

97.2

98.1

93.5

Teller CSI YTDPerformance

levels

1 2 3

1= Need to improve; 2 = Room to improve; 3 = Maintain

Figure 2. Post-Analysis Customer Satisfaction Tracking Report

By streamlining metrics, presenting them in order of relative importance, and coding performance, this new report helped employees be more effective in their customer satisfaction

efforts. The bank also gained customers and boosted customer satisfaction—a rare achievement following a merger.

5Balanced Scorecard Report May–June 2006

Choosing the Right Measures (continued)

and smaller increases in customersatisfaction.

Armed with these insights, execu-tives at the bank were now in a position to provide high-qualityinformation and feedback to operating units in the form of thecustomer satisfaction trackingreport shown in Figure 2. Severalfeatures distinguish this new reportfrom that shown in Figure 1:

• The number of metrics trackedwas reduced from 26 to 15.Eleven performance metrics that showed either a marginalor no relationship to the objec-tives of customer satisfactionand branch profitability weredropped. The remaining metricswere identified as strong driversof these objectives.

• Measures of service performanceare listed on the scorecard inorder of their relative impor-tance in driving customer satis-faction and branch profitability,starting with the attribute “appre-ciate your business,” which wasfound to be the strongest driver.

• Current performance againsteach of these metrics is nowcoded as red, yellow, or greencorresponding to whether that metric is in immediateneed of improvement, has room to improve, or is at anacceptable level. Targets indicat-ing red, yellow, and green levels of performance (in Figure 2, indicated in numberedcolumns) were established foreach individual metric based onthe thresholds identified in the“S”-curve analysis. Performancebelow the lower threshold,which needed to be crossedbefore service improvementscould yield incremental benefits,was coded red; performanceabove the upper threshold,

where service improvementsbegan yielding diminishingreturns, was coded green; andperformance between thesethresholds was coded yellow.

This new reporting tool and the measurement and analysis it was built on served as thelinchpin for translating the bank’sservice strategy into operations.Branch-level employees, whosecompensation was linked to customer satisfaction, now hadhigh-quality information aboutwhere to focus their efforts toachieve the largest gains in cus-tomer satisfaction and, ultimately,branch profitability. Fewer mea-sures kept the focus on those that mattered most in driving the bank’s strategic objectives of increasing satisfaction and profitability in its branch network.Targets that indicated where service improvements would havethe largest impact on satisfactionand profitability allowed branch-level employees to allocate theireffort more effectively among theremaining metrics included in thescorecard. The net result? Thebank was able to deliver on itspromise of high customer service.Better yet, in its post-merger climate, where many banks experience customer attrition ratesof 5% to 10%, the bank increased its net customer acquisitions. Itscustomer satisfaction index rosefrom 81.5% to 84.5%.

The type of statistical analysisdescribed here requires data onperformance measures acrosscomparable operating units,across time, or both. Banks areparticularly amenable to this type of analysis because they aredecentralized and have multiplecomparable branches. But otheroptions exist. You can wait untilyou have amassed enough perfor-

mance measurement data overtime, although you may need 20 to 30 periods’ worth to ensurevalid results. A second option: be creative about how you defineyour unit of analysis. Many impor-tant performance measures aredefined at the individual customeror employee level. The key tofocusing on measures that matter,and avoiding metric overload, isto take a systematic approach toselecting measures that have clearand demonstrable links to youroverall customer and financialperformance objectives. If thedata is just not available to makethese links “demonstrable,” thenmanagers across organizationalunits and at different organiza-tional levels should have exten-sive discussions and come tosome consensus on which ofthese links are at least most“defensible.” Putting this kind ofeffort into the design of yourscorecard can provide strategicclarity at the top of the organiza-tion and allow you to providegood, actionable information atthe operational level where strategyis actually implemented. �

1. Customer satisfaction was measured as the percentage of customers rating themselves as “satisfied” or “highly satisfied.” Similarly,performance on each service attribute was measured as the percentage of customers ratingthemselves as “satisfied” or “highly satisfied”with that particular aspect of service. For example, the measure corresponding to the service attribute “appreciate your business”captured the percentage of customers ratingthemselves as “satisfied” or “highly satisfied”with the tellers’ making them feel appreciated.

2. A combination of factor analysis and multipleregressions were used to estimate the relation-ships between measures of service attributes,customer satisfaction, and branch profitability.

Dennis Campbell is author of“Putting Strategy Hypotheses to the Test with Cause-and-EffectAnalysis,” BSR September–October2002 (Reprint #B0209E).

T O L E A R N M O R E