Variance Analysis Author: Alton Shader, James Baker March 1998
Agenda What is variance analysis? Linear variance Ice Cream Co. Two component variance analysis Jesse’s Brewery Variance analysis with more than two components Boston Video Key takeaways
What is Variance Analysis? Variance analysis is used to understand and assess the drivers of change in measured variables. Variance analysis helps explain and understand what drives the difference between two measures of the same variable (e.g., 1998 profit vs. 1995 profit) Variance analysis explains differences between measures by breaking those measures into their base components (e.g., 1998 revenue and 1998 OPEX as components of 1998 profit) and quantifying the impact of each component Bain frequently uses variance analysis to quantify and identify true profit drivers help drive future analysis on the most leveraged issues
The Value of Variance Analysis Bain uses variance analysis to gain business insight and to identify the most effective and valuable action steps. Why do we need to perform variance analysis? Gives business insight as to what drives revenue/cost/ profit Leads to 'actions' Why does profit change? What driver has the most impact? What explains differences in relative cost position? Used to determine product line profitability Identifies areas of focus for cost reduction Indicates impact of lowering price Drives customer segmentation strategies
Definitions in Variance Analysis Most situations fall into one of three categories of variance analysis. Linear Two Component More than Two Component Description: Simple comparison of one component against another Understanding individual impact of two variable on a single measure Understanding individual impact of more than two variables on a single measure Example: 1990 total cost per unit versus 1995 total cost per unit 1998 revenue increases driven by 1998 price increase and unit sales Differences in customer revenue driven by number of transactions, product mix and other fees
Agenda What is variance analysis? Linear variance Ice Cream Co. Two component variance analysis Jesse’s Brewery Variance analysis with more than two components Boston Video Key takeaways
Linear Variance Example (Ice Cream Co.) Situation: Orit and Tom’s Ice Cream Co. produce ice cream which they sell in gallon-sized containers at supermarkets around the country They have tracked their cost to produce ice cream per gallon over time and have seen it decline over five years by 25% Question: What has driven costs downward? What might Orit and Tom focus on to achieve better cost savings going into the future?
Linear Variance Example 1990 1995 Variance Production costs: (includes sales, labor, admin) $210,000 $220,500 $10,500 Raw materials costs: $60,000 $63,000 $3,000 Advertising expenses: $30,000 $31,500 $1,500 Total gallons sold: 50,000 70,000 20,000
Linear Variance Example (Variance per Gallon) 1990 1995 Variance Unit production costs: $4.20 $3.15 ($1.05) Unit raw materials costs: $1.20 $0.90 ($0.30) Unit advertising costs: $0.60 $0.45 ($0.15) Total per unit costs: $6.00 $4.50 ($1.50)
Linear Variances Example (Ice Cream Co.) Thorough Bain analysis of the data allocated the reduced costs into three components.
Linear Variances Example Management of Ice Cream Co. appears to have enjoyed increasing economies of scale, particularly relating to production costs, which have driven costs downward Orit and Tom might want to focus in the future on gaining even greater leverage in managing production costs as production volume increases
Agenda What is variance analysis? Linear variance Ice Cream Co. Two component variance analysis Jesse’s Brewery Variance analysis with more than two components Boston Video Key takeaways
Two Component Variance Example - Jesse’s Brewery (1 of 6) Situation: Jesse’s Brewery produces high-quality beers which it sells in cases In the past year, Jesse’s Brewery has experienced 15% revenue growth. During the same time, sales volumes and prices have both increased Question: Jesse’s Brewery would like to understand what percent of the revenue increase comes from the increase in sales volumes, and what percent from the increase in price
Two Component Variance Example - Jesse’s Brewery (2 of 6) What percent of the $27MM increase in revenue is due to the price increasing? the volume sold increasing? both the price and volume sold increasing? this is covariance, where part of the variance is not easily attributable to a single variable Case Price: $18.20 $20.00 Cases Sold: 9.7MM 10.2MM
Two Component Variance Example - Jesse’s Brewery (3 of 6) The best way to approach variance analysis problems is to use the rectangle diagrams to intuitively understand how the changes in the two variables account for the change in the overall quantity. 1998 Revenue $203.5MM 10.2MM Covariance (1998) Volume Variance (part of revenue (Part of revenue change change attributable attributable to volume) to change in both volume and price) 9.7MM (1997) Volume Price Variance 1997 Revenue (Part of revenue change attributable to price) $18.20 $20.00 (1997) (1998) Price
Two Component Variance Example - Jesse’s Brewery (4 of 6) A four-step process leads to the solution. 1998 Revenue (P2 x V2) Step 1: Volume Variance V2 = (V2 -V1) x P1 (1998) Volume Variance Step 2: Co- Price Variance = (P2 -P1) x V1 (Part of revenue change variance V1 attributable to volume) Step 3: : Covariance (1997) = (V2 - V1) x (P2 - P1) Volume Price Step 4: Allocate covariance 1997 Revenue Variance to volume and price variance (Part of based on the proportion of the (P1 x V1) revenue total for each variance change -e.g., if: attributable to volume variance = $2M price) price variance = $1M covariance = $.5M then 66.7% of $.5 is added to the volume and P1 P2 33.3% is added to the price (1997) (1998) Price
Two Component Variance Example - Jesse’s Brewery (5 of 6) Here is an example of an annotated spreadsheet to determine variance. Jesse's Brewery Data 1997 1998 Variance Variance % Price per case $18.20 (P1) $20.00 (P2) $1.8 (P2 - P1) Number of cases sold (in MM) 9.7 (V1) 10.2 (V2) .5 (V2 - V1) Revenues (in $MM) $176.5 $203.5 $27.0 Step 1: Price Variance $176.5 (P1 x V1) $193.5 (P2 x V1) $17.0 (193.5 - 176.5) 63% (17.0/27.0) Step 2: $9.1 (185.6 - 176.5) 34% (9.1/27.0) Volume Variance $176.5 (P1 x V1) $185.6 (V2 x P1) Step 3: $0.9 (P2-P1)x(V2-V1) 3% (0.9/27.0) Covariance Revenue variance excluding covariance $26.1 (17.0+9.1) Proportions for assigning covariance Price 65% (17.0/26.1) Volume 35% (9.1/26.1) Assignment of covariance to Price Variance $0.6 (0.9x65%) to Volume Variance $0.3 (0.9x35%) Adjusted Variances $17.6 (17.0+0.6) Price Variance $9.4 (9.1+0.3) Volume Variance
Two Component Variance Example - Jesse’s Brewery (6 of 6) If price and volume variances are calculated correctly, then we can take their proportions to allocate covariance. Price variance: $17.0MM 65% Proportions are Volume variance: $9.1MM 35% Covariance: $0.9MM 65% 35% to price variance to volume variance $0.6MM $0.3MM + + $17.0MM $9.1MM Total variances: $17.6MM (65%) $9.4MM (35%) Price Volume
Negative Covariance In some situations one variable may fall as the other rises. The formula for covariance does not change. 14.7 new (1998) Volume 9.7 (1997) $17.00 $18.20 Price new (1997) (1998) Even if one variable decreases, the formula applies the same way. The only difference is that price variance and covariance turn negative. Volume = (14.7-9.7) x $18.20 = $91.0 MM Price = ($17.00-$18.20) x 9.7 = - $11.6 MM Covariance = (14.7-9.7) x ($17.00-$18.20) = - $6.0 MM Total Variance $73.4 MM
Negative Covariance To summarize, the methodology does not change in calculating negative covariance. In two variable problems, when one variable increases and one decreases (+ - ), we arrive at a negative covariance if the formulas are applied exactly as before (change only the variable whose variance we are calculating) This negative covariance (P2 - P1) x (V2 - V1) then needs assigning in proportion to the absolute values of the two variables This absolute value approach can be applied to multiple variables using the Excel spreadsheet The example given is also intuitively correct (price falls, volume rises)
Agenda What is variance analysis? Linear variance Ice Cream Co. Two component variance analysis Jesse’s Brewery Variance analysis with more than two components Boston Video Key takeaways
More Than Two Component Variance Example Situation: Boston Video is a chain of video rental stores in the Northeast U.S. In addition to renting video movies and games, the stores sell a variety of movie-related items (popcorn, candy, film magazines). They also get money for late fees Bain has been hired to help Boston Video segment its customer base and understand what drives the differences in spending patterns between segments Question: What variables impact revenue per customer? What is the overall variance by customer type? What are the breakdowns of that variance by customer type? What should Bain focus on improving?
More Than Two Component Variance Example A, B, & C Customers Data Customers were segmented into 3 groups. “A” customers represent only 20% of Boston’s customer base but generate 50% of store transactions and 60% of revenue. 11,411 customers 74K transactions $441K How might you frame an analysis to understand a lot impacts revenues per customer?
Revenue Drivers Framework More Than Two Component Variance Example Revenue Drivers Framework The differences in customer spending levels between segments can be identified and isolated into two broad categories. Revenue per Customer Number of Transactions Revenue per Transaction Rental Revenue Sell Thru Revenue Late Fees and Other Product Mix Quantity
More Than Two Component Variance Example Customer Segmentation The following data was made available about customer habits. Number of Transactions Revenue per Transaction
Revenue Per Customer Per Month More Than Two Component Variance Example Revenue Per Customer Per Month “A” customers spent on average 2.6x the average “B” customer and nearly 10x the average “C” customer. 2.76 x $6.89 $11.82 $17.02 1.29 x $5.38 0.45 x $4.45 Customer segment:
More Than Two Component Variance Example Drivers of Variance How would you complete this slide showing the split between the two variance drivers?
More Than Two Component Variance Example Sample Spreadsheet This spreadsheet details the calculations necessary to compute the variances and covariance for transactions and revenue per customer. Facts: Total Trans/ Rev/ Customer Rev/Mnth Mnth Trans Variance A 19.01 $ 2.76 6.88 B 7.20 1.29 5.58 (11.82) C 2.00 0.45 4.45 (17.02) Variance from A to B %total: calculation: trans variance (10.14) (1.47) 74% (B trans - A trans) times A price rev variance (3.60) (1.30) 26% + (B price - A price) times A trans total variance before covar (13.73) =Total Variance before Covariance covariance 1.92 + (B trans - A trans) times (B price - A price) answer =Total Variance Allocate covariance: 100% trans 1.42 74% to transaction rev 0.50 26% to rev Total Delta Trans (8.72) Regular trans variance plus allocation from covariance Total Delta Price (3.09) Regular rev variance plus allocation from covariance =total variance
More Than Two Component Variance Example Sample Spreadsheet
Variance Analysis (Revenue per Transaction) More Than Two Component Variance Example Variance Analysis (Revenue per Transaction) Now suppose the client wants to further understand what drives revenue per transaction. How would you complete this slide showing the split between the three revenue per transaction variables? Revenue Per Transaction Sources of Variance Customer segment:
More Than Two Component Variance Example Sample Spreadsheet This spreadsheet further allocates revenue per transaction variance into types of product purchased.
More Than Two Component Variance Example Drivers of Variance Variance analysis showed that the primary difference between customer segments was how often they came to the store. Over 70% of the variance between A customers and the others was explained by transaction frequency. 26% 30% 74% 70%
Variance Analysis (Revenue per Transaction) Revenue per transaction analysis was not very telling. Revenue per Transaction Sources of Variance Customer segment:
Summary What drives A customer behavior is the number of transactions they complete so, Bain needs to help improve the frequency of transactions of B and C customers (assuming that A customers continue to remain loyal and that a retention strategy is therefore not the most leveraged work)
Agenda What is variance analysis? Linear variance Ice Cream Co. Two component variance analysis Jesse’s Brewery Variance analysis with more than two components Boston Video Key takeaways
Key Takeaways Variance analysis can be used in many instances to understand changes in a business and to drive strategic and tactical decision-making. Key points to remember are: Variance analysis is a simple way of breaking down changes in a variable into their component parts, to yield business insight Know how to set up a spreadsheet and variance analysis becomes straightforward Often we only look at two components at a time The basic approach is to change only one variable at a time and observe the effect on the total Covariance explains the amount of variance that is attributable to the change of both variances (as opposed to one) and is normally split between variables in proportion to their variances
Sample Spreadsheet Variance from A to C Trans variance Rev variance Total variance before covar covariance Answer Alliance covariance: Trans Rev $ (17.02) $ (15.98) $ (6.73) $ (22.66) $ 5.64 $ 3.96 $ 1.64 100% 70% 30% Total Delta Trans Total Delta Price $ (11.96) $ (5.6) (2.31) 2.76 $ (2.44) $ 6.88 %total: calculation: (C trans-A trans) times A price +(C price - A price) times A trans =Total Variance before Covariance +(C trans - A trans) times (C price - A price) =Total Variance 70% to transaction 30% to rev Regular trans variance plus allocation from covariance Regular rev variance plus allocation from covariance =total variance