Granular Metrics, Feedback and Experimentation Arnoldo Hax Alfred P. Sloan Professor of Management Contributions of the Delta Model The Triangle Adaptive Processes Opening the mindset to new strategic position The Best Three Strategic Options: Product does z Best Product not always win Linking Strategy with Execution Execution is not the problem, linking to strategy is Total Customer Solutions z System Lock-in z Execution is captured through Three Adaptive Processes: Operational Effectiveness z Customer Targeting z Innovation z whose roles need to change to achieve different Strategic Positions Aggregate Metrics Granular Metrics and Feedback Measuring success Discovering performance drivers Good financials do not always lead to good results Aggregate performance metrics need to reflect each of the Adaptive Processes and their role based upon the strategic position Managing by averages leads to below average performance Business is non-linear. Performance, particularly bonding, is concentrated. Granular metrics allow us to focus on underlying performance drivers, to detect variability, to explain, to learn, and to act. Product performance z Customer performance z Complementor performance Business is non-linear. Performance, z The Delta Model and Granular Metrics Metrics Select Performance Indicators Act from Variability • Define program • Define structured tests • Rollout Learn from Variability • Hypothesize cause and effect • Evaluation Detect Variability • Drill down to detailed segmentation • Isolate variability Explain Variability • Identify performance drivers • Correlate drivers with variability Drivers of variability for selected performance indicators Performance indicators Drivers of variability Product cost and quality z Customer profit z Complementor contribution z Business segment economic value z Scale z DensityDensity-e.g. concentrat concentration of serv service ice z Location cation z Labor productivityproductivity- practices and training training z Equipmen Equipment product productivity ivity-- design design z Process design z And And so on Customer size z Customer revenue z Tenure z Acquisition cost z Channel Channel mix z Customer care supp support z Customer investments in relation relationshi ships z And And so on Size of releva relevant complementor products z Complementor investment in bu business z Relative size in customer value chain z Complementor contributio o customer economics contribution tto z Exclusivity of relationship relationship z And And so on ROI (return on on investment) – profitability z RiskRisk- volatility and covariance covariance z Option value z Investment base z Cashflows Cashflows Experimentation as the basis for effective change change Large Size of Change Small The middle road: lower returns, or unacceptable when first mover advantage is high Unacceptable risk, risk, as a starting point. point. Highly desirable as as endpoint endpoint Ineffective: lower returns, or unacceptable when first mover advantage is high Testing: the relevant area for experiments leading to large change Slow Speed of Change Fast Cost De-Averaging: Using Granular Metrics To Drive Performance 1. The Value Chain of the Local Business Data Circuit Switching Transmission Order fulfillment Etc. Marketing Billing Average Cost = $395/order 3. The 80-20 Cost Effect 100 80 % of costs $1,000 Cost/order 2. Individual Order Cost 60 40 20 $100 Most Expensive Orders Least Expensive 0 20 40 60 % of orders THE BEHAVIOR OF COST - THE CASE OF BUSINESS DATA SERVICES Figure by MIT OCW. 80 100 The drivers of the cost of order-fulfillment fulfillment Cost of order fulfillment Location of work centers Activities of order fulfillment Manpower productivity Equipment Order Fulfillment Headcount required to process 1,000 orders per month 70 Competitor 1 - Houston Company Competitors Competitor 1 - Chicago Chicago } 39% gap 1 1,000 Client performance curve New York Tulsa Charlotte Competitor 1 ­ Denver Kansas City Competitor 1 ­ San Francisco Competitor 2 ­ San Jose Competitor 2 Indianapolis 2,000 Best-in-class performance 5,000 12,000 27,000 Scale = 72% Log/log graph Competitor 3 ­ Jacksonville 60,000 Orders/Month T HE BE H AV I O R O F CO S T BY L O CAT I O N Figure by MIT OCW. Customer ready 'Defect-free Path' Automatic switch map Fast facilities test $100 (per access leg) Valid design Facilities available Order Facilities not available Access available Pass test Access not available 10x cost difference Access available Access not available Skin test Invalid design 'Defect-prone Path' Manual switch map Fail facilities test Fail access test THE BEHAVIOR O F C O ST B Y FIV E A C TI V I TIES IN TH E VAL UE CHAI N Figure by MIT OCW. $2,000 (per access leg) Customer not ready 1 MILLION POSSIBLE ORDER PATHS Pass access test The behavior of cost by type of defects defects Client Defect-free Business Data Service Orders % of orders Cost/ order Averag e cycle time 71% $100 5 Days 18% 24.0% $900 25 Days 60% Total cost Focus of program developmen t Business Data Service Orders Business Data Service Orders 5.0% $1,800 45 Days Average 1 $395 12.3 Days 22% Craftsmen Efficiency Total hours/cleared fault 10 8 6 S:1 4 2 0 A B C D E F G H I Craftsmen J K L M Fault not fixed correctly & repeats Pair throw: reroute around fault Request assistance Refer fault to someone else (specialist) Fix fault N Cumulative cost (%) 100 80 The Need for Granular Metrics 0 0 25 100 Cumulative craftsmen (%) THE BEHAVIOR OF COST BY INDIVIDUAL WORKERS Figure by MIT OCW. Records Errors Concentration 100 100 80 80 % of Records Errors % of Line Faults Line Fault Concentration 100% of demand pair reclamations are concentrated in 17% of junctions 60 40 20 0 0 20 40 60 80 100% of records errors are concentrated in 19% of equipment 60 40 20 100 % of Junctions 0 0 20 40 60 80 % of Equipment THE BEHAVIOR OF COST BY TYPE OF EQUIPMENT Figure by MIT OCW. 100 Cost/Unit Benchmarks Cost Drivers Best Practice Sessions Select Associates Director/Process Reviews Standard Supervisor a b Individual Performance Individual Reviews Standard a b c Supervisor Supervisor Review Ind. Actual Target Total District/Center Reviews Supervisor Total Actual Target 'Supervisor Coaching' OPERATIONAL EFFECTIVENESS: ADAPTIVE FEEDBACK USING GRANULAR METRICS Figure by MIT OCW. Profit De-Averaging: Using Granular Metrics To Drive Performance THE NEED FOR GRANULAR METRICS 3x 2x 35 Profit Margin 1x 0 -x Cumulative percentage of customers 10 20 30 40 50 60 70 80 90 100 -2x 85% of Profits 46% of Profits 40% of Profits -3x 21% of Profits -4x -5x -6x -7x P RO FIT MA RG IN CO N T RIBU T ION Figure by MIT OCW. The drivers of customer profitability in the credit card business s Profit by customer Revenue per customer Number of customers Balance level per month Months of tenure Monthly Profit / Customer 75th Percentile + 1.5* Box Size ('Upper limit of connected data') $70 $60 $50 $40 $30 $20 $10 $0 $(10) $(20) 75th Percentile Median 25th Percentile $10-$100 $100-$200 $200-$400 $400+ Revenue/Customer 25th Percentile -1.5* Box Size ('Lower limit of connected data') VA R I A B I L I T Y O F P R O F I T S B Y U S A G E L E V E L S Figure by MIT OCW. Profit Margin (ROS) Company Scale 18% Customer Share 22% 15% 5MM 10MM 20MM 30MM Number of Customers 30 20 10 0 -10 -20 -30 -40 -50 $1000 $2000 $3000 $4000 Balance Level $/Month Customer Tenure 30 20 10 0 -10 -20 -30 -40 -50 10 Months of Tenure FURTHER E X A MI N AT I ON O F P R O F I T DR IVER S Figure by MIT OCW. 20 30 Cumulative Profit 5x } } Tailor offers to individual customers (i.e., increase response & profit with differential offers tailored to customer's preferences) 3.5x x 0 Offer one product to customers, based on both response & profitability (i.e., acquire customers likely to have good volume/ margin/lifetime characteristics) } Offer one product to those who will respond (i.e., maximize response for a given offer, assuming average economics) 150,000 300,000 Cumulative Number of Customers Solicited T H E E F F E C T S O F C U S TO M E R TA R G E T I N G Figure by MIT OCW. } Offer one product to everyone Untargeted solicitations yield a negative profit Customers’ targeting behaviors behaviors Behaviors Typical competitors Capital One Offer z Test/year z Service tier z Key metric z Targets z Solicitations/year z Segmentation frequency z Risk orientation z Data sources z Staff analysts 75 products 30 4 ROE, Delinquency High response likelihood 50 million Biannual Risk averse 80 20 5,000 products 15,000 13 NPV High NPV potential 400 million Ongoing Price for risk 50,000 150 z Capital One- achieving a Total Customer Solutions through Customer Targeting Variation 2 Variation 1 Select Offers Further improvement Offer Creation Core offer • Interest Interest rates rates • Fees • Promotions Mass marketing Screen for profitability • Full life cycle customer profitability Variation 2 Variation 1 Test cells Collect customer response Success metrics metrics Customer NPV z Acquisition cost z Activation z Percent with revolving balances z Fee income per account z Charge-off per account z Tenure z Typical competitor Capital One $62 $77 58% 25% $55 $71 4 years $427 $67 79% 40% $73 $54 6 years Value Created by Business Unit % of Total Company Value Created 60 100% of value created 20 42% more value created -21% of value created -21% of value created i) 20% of new capital ii) 32% of R&D i) 36% of new capital ii) 25% of R&D i) 20% of new capital ii) 22% of R&D 15 10 5 0 i) 24% of new capital ii) 21% of R&D -5 -20 -40 0 25 50 % of Equity Investment 75 100 1) 6 BUs contribute 100% of value created 2) 16 more BUs contribute another 42% of value created 3) Those 22 BUs are worth $ 15.5 billion, on an equity investment of only $4.2 billion 4) 44% of new capital investment and 53% of R&D spending over the next three years is going to BUs earning economic profits 5) 56% of new capital investment and 47% of R&D spending are going to BUs that will be generating economic losses SOURCES OF VALUE CREATION Figure by MIT OCW.