Lottery Advertising as a Market Lever: Florida’s Project: Lottery Advertising as a Market Lever: Florida’s Project What can statistical analysis tell us about advertising as a successful business investment strategy?


Introduction: Questions We Face: Introduction: Questions We Face “Why does a lottery need advertising money when it’s a monopoly? People will play anyway, won’t they?” “Why don’t we cut the lottery advertising budget by a third and transfer that money directly to education?” “How do we know that advertising lottery products is worth doing?” “Within our lottery, what would generate the better return? Spending on advertising? More sales reps? More retailers?


Presentation Overview: Presentation Overview Dennis Suggestions for quick ways to show advertising’s ROI Why Florida invested in an advertising impact study And, at the end, brief comments on how the study is being used Mike The study itself—methods and results


Importance of Advertising Research: Importance of Advertising Research Understanding your market Determining advertising strategies and tactics Judging the effectiveness of your advertising


Importance of Advertising Research: Importance of Advertising Research Demonstrating the cost-effectiveness of advertising spending to commission, governor, legislature, public and press Today’s Topic


Slide6: Quick Methods


Demonstrating Advertising’s Benefits: Demonstrating Advertising’s Benefits Compare two similar scratch-off games—one advertised and the other not advertised Ideal: Same price Same or similar prize structure and payout Same or similar theme and playstyle Launched at same time of the year Compare sales and “noble cause” contribution over same time period Method One: Case Study


Demonstrating Advertising’s Benefits: Demonstrating Advertising’s Benefits


Demonstrating Advertising’s Benefits: Demonstrating Advertising’s Benefits In first 13 weeks of sales, advertised game had— $9.4 million more in sales $2.1 million more in profits for education $1.1 million in profits when cost of advertising considered ROI Ratio = $2.11 for every advertising $ spent Net ROI = $1.11 Method One: Benefits of Advertising


Demonstrating Advertising’s Benefits: Demonstrating Advertising’s Benefits Compare spending of tracking survey respondents—those aware of advertising vs those not aware Florida’s monthly tracking survey 500 adults, by phone Questions on playership, advertising, demographics, attitudes, special subjects Weighting--Estimate number of people playing and number having seen lottery advertising Method Two: Survey Research


Demonstrating Advertising’s Benefits: Demonstrating Advertising’s Benefits Number of players % recalling lottery advertising (and % NOT recalling) Average spending among those seeing the ads (vs NOT seeing them) Annualized spending differences Lottery’s advertising expenditures “Noble cause” contribution rate


Demonstrating Advertising’s Benefits: Demonstrating Advertising’s Benefits Among respondents, 73% recalled seeing a lottery add; 27%, not. Average monthly spending— Those recalling the advertising = $33.10 Those not recalling the advertising = $28.50 Averaged results of two calculation methods ROI Ratio = $1.97 for every $ of advertising spending Net ROI = $0.97 Method Two: Benefits of Advertising


Slide13: Why invest in an advertising impact study?


Why Did Florida Undertake an Advertising Impact Study?: Pressures in the Lottery Industry To increase sales and net income in a mature industry Growing competition for consumers’ discretionary spending Challenges from governors, legislatures and boards to justify spending especially on advertising Must be able to defend allocations of business resources Must maximize returns on business investments Why Did Florida Undertake an Advertising Impact Study?


Why Did Florida Undertake this Study?: Opportunity to use statistical and mathematical modeling tools = fresh perspective on what drives our business Enhance–not replace–business experience, intuition and judgment Why Did Florida Undertake this Study?


Slide16: The Study


Introduction to the Study – Basic Questions: Introduction to the Study – Basic Questions Our study was guided by two basic questions: How does the Florida Lottery’s spending on advertising compare to that of other U.S. lotteries and other producers of comparable goods and services? What is the quantitative relationship between advertising expenditures and lottery ticket sales?


Introduction – Analyses Conducted: Introduction – Analyses Conducted The following analyses were conducted in response to these questions: Comparison of Florida Lottery advertising expenditures with advertising expenditures on alternative but comparable products, e.g., soft drinks, movies, candy bars, and snack foods Examination of Florida Lottery advertising expenditures and sales over time Regression analysis to predict the impact of advertising expenditure changes on sales and transfers (net) Regression analysis on market levers beyond advertising


Introduction - Study Methods: Introduction - Study Methods Software tools and data SAS (Statistical Analysis System) La Fleur Almanac data and Florida Lottery data Multiple regression modeling was employed A research method which shows how variables of interest are related by constructing a mathematical model (equation) representing a “best fit” line for a set of data points Regression model validation Models were validated against previous actual sales Models depict “associations” not necessarily “cause and effect” Models are customized to Florida


Competitive Advertising: Competitive Advertising Vendors of Comparable Products


Vendors of Comparable Products: Vendors of Comparable Products The companies selected satisfied the following criteria: Sell consumer products that are more or less comparable to the cost of lottery tickets Sell consumer products that are not life necessities Sell consumer products that could be purchased in the neighborhood Sell products with advertising focused on the same buyer community Provide usable data specifically labeled as advertising, marketing, and/or promotional expenses, not commingled with other reporting categories


Vendors of Comparable Products: Vendors of Comparable Products


Slide23: Advertising to Sales Ratios for Companies Providing Comparable Products The horizontal line marks Florida’s 2003 advertising-to-sales ratio


Competitive Advertising: Competitive Advertising Sector-Aggregated Data


Sector-Aggregated Data: Sector-Aggregated Data We found a compilation of 81 sectors emphasizing retail stores and services. The data were originally collected in 2001 and 2002 by trade associations and were made available on the web by the advertising department of the Virginian-Pilot News The U.S. Census Bureau is another source of sector data


Sector-Aggregated Data: Sector-Aggregated Data While these data are of interest for comparison, it is useful to keep the following in mind Products and services in these sectors are not always comparable to lottery products. Their competitive environments may be quite different Businesses included in a composite are of various sizes, types, and focuses, perhaps diluting the meaningfulness of comparisons


Slide27: About 1/3 About 2/3 The vertical line marks Florida’s 2002 advertising to sales. Distribution of Advertising to Sales for Selected Products, Retail Stores and Services The median of the distribution is 2.50%, the mean is 3.25%. The horizontal line marks the height of intersection.


Trends in Florida Lottery Advertising Spending and Sales : Trends in Florida Lottery Advertising Spending and Sales


Trends: Sales: Trends: Sales The first panel suggests some falloff in sales with time. The adjusted numbers of the second panel display the falloff more clearly.


Trends: Advertising: Trends: Advertising The trend in the adjusted numbers for advertising expenditures also matches the trend in sales.


Trends: Disposable Income : Trends: Disposable Income Is the falloff in per capita sales over time due to a falloff in per capita disposable income over time? The figure below dispels that possibility


The Relationship Between Sales and Advertising: The Relationship Between Sales and Advertising


The Relationship Between Sales and Advertising: The Relationship Between Sales and Advertising Regression analysis was used initially to model the relationship between sales and advertising without consideration of other market levers The regression was based on historical data Two models were fit The first was limited to data from Florida The second added data from 19 other state lotteries


The Relationship Between Sales and Advertising: The Relationship Between Sales and Advertising Florida Data


Inflation-Adjusted Regression of Sales on Advertising - Florida Data: Inflation-Adjusted Regression of Sales on Advertising - Florida Data The Regression of Sales Against Advertising for Florida


Discussion of Regression Model: Discussion of Regression Model As indicated in the previous slide, the model suggests that a dollar in additional advertising is accompanied by $56 in additional sales If we assume a transfer rate of 33.5%, an increase of $1 in advertising is accompanied by an increase of $19 in net income The visual fit of the model is encouraging however the model is built on only 13 observations


The Relationship Between Sales and Advertising: The Relationship Between Sales and Advertising Florida and States Like Florida


Operational Variables to Cluster States: Operational Variables to Cluster States Eight features were used to find lotteries similar to the Florida Lottery: Presence or absence of video lottery Online sales as percent of total sales Percentage of non-food agents Instant game payout rate Agents per million adults Sales representatives per million adults Commissions per agent Demographic variables


Florida and States Like Florida: Florida and States Like Florida The cluster analysis grouped the following 20 states as similar: Excluded states: Video lottery or population density


Two Regression Models: Two Regression Models The Regression of Sales Against Advertising for Florida Multi-State Model Florida-Only Model


Discussion of Regression Model: Discussion of Regression Model Based on the multi-state model, an increase of $1 in advertising is accompanied by an additional $15 in sales If we assume a transfer rate of 33.5%, an increase of $1 in advertising is associated with an increase of $5 in net income


Further Analysis: Further Analysis Regression Modeling of Advertising and Other Market Levers


Market Levers Considered: Market Levers Considered Retailer infrastructure Percent of retailers that are not convenience stores or grocers Retailer density (retailers per million adults) Percent of online retailers LSR density (sales representatives per million adults) Retailer compensation Commissions as a percent of sales Player prizes and incentives Prizes as a percent of sales Advertising expenditures


Model Approach: Model Approach Model fitting began with all 7 market levers and iteratively removed levers that did not sufficiently contribute to explaining the relationship between levers and sales Florida and 19 other states were used


Florida’s Regression Model: Florida’s Regression Model The resulting model for Florida: Sales = 791,190,000 + 11.00*Advertising Expenditures + 73,741*Number of Retailers + 2,997,296*Number of LSRs


Model Validation: Model Validation Except for 1995, model fit is good, differing on average less than 2% of sales


Other Modeling Results : Other Modeling Results Still further regression analyses revisited retailer density and trade styles using Florida-only data Both market levers turned out to be pivotal in predicting Florida sales


Summary of Study Findings: Summary of Study Findings


Summary of Study Findings: Summary of Study Findings When compared to vendors of comparable products, the Lottery appears to be spending less on advertising as a percent of sales than the typical vendor When compared to sector-aggregated spending, the Lottery appears to be spending less on advertising as a percent of sales than is typical in most industry sectors The fall of Florida Lottery sales and advertising (per adult) from the 1990 levels appeared to be associated


Summary of Findings: Summary of Findings Based on modeling, an increase of $1 in Florida advertising is accompanied by an estimated increase of $15 in sales accompanied by an estimated increase of $5 in net income Other market lever variables also are key for Florida, especially retailer density and trade style


How the Study Results are Helping the Florida Lottery: How the Study Results are Helping the Florida Lottery


How is this Study Helping the Florida Lottery?:  Business Investment Decisions Based not just on business experience and judgment Not just on cases or surveys But also on mathematical models that make specific projections Models that help us compare the costs and returns of different business investment options How is this Study Helping the Florida Lottery?


How is this Study Helping the Florida Lottery?:  Business Investment Decisions Example: Retailer Density Decision to seek spending authority to increase our number of retailers Example: Proportion of On-line Retailers Stuck to conversion plan to convert all retailers to full-service retailers Example: Retailer Trade Styles New corporate recruitment effort to increase number of retailers outside the convenience store trade style How is this Study Helping the Florida Lottery?


How is this Study Helping the Florida Lottery?:  Advocacy Credible method for showing the sales and profit returns of its business proposals to the Legislature How is this Study Helping the Florida Lottery?


How is this Study Helping the Florida Lottery?: Advocacy Example: Advertising Spending In much stronger position to defend advertising funding Example: Retailer Density Gained legislative authority to increase number of terminals How is this Study Helping the Florida Lottery?


Further Information: Florida Lottery Dennis Harmon harmond@dol.state.fl.us Battelle Mike Huffenberger huffenbm@BATTELLE.ORG Further Information