Is Once Really Enough

download Is Once Really Enough

of 4

Transcript of Is Once Really Enough

  • 7/28/2019 Is Once Really Enough

    1/4

    Is Once Really Enough?Making Generalizations about Advertising'sConvex Sales Response Function

    JENNIFER TAYLOREhrenberg-Bass Institute

    for Marketing ScienceUniversity of South

    AustraliaJennifer. taylor@

    marketingscience.infoRACHEL KENNEDYEhrenberg-Bass Institutefor Marketing ScienceUniversity of South

    Australiarachel.kennedy@

    marketingscience.infoBYRON SHARPEhrenberg-Bass Institute

    for Marketing Scienceuniversity of South

    Australiabyron.sharp@

    marketingscience.info

    This article examines thegeneralizability of convex-shaped advertising responsefunctions. Using single-source data, we anaiyzed the response functions of brands infour consumer goods categories. This study supports the prior finding that convexresponse functions are typical, but not universal. W hile the convex response functiois found to apply across a range of conditions, more work is needed to understandmeasurement issues, exceptions, and boundary conditions.

    D O E S AN ADVERTISER get a better return by reach-ing twocustomers once, or one customer twice?

    It is an extremely important issue for mediaplanners. Single-source analysis has delivered anempirical generalization (EG): the immediate ad-vertising sa les-response function is convex, whichmeans it is better to reach more consumers once.This empirical law appears to generalize for tele-vision advertising for established consum er goo dsbrands inmature markets. It is such an importantdiscovery, however, that it deserves furtherinvestigation.

    In this article, we present new findings fromsingle-source analysis in four consumer goodscategories. They support the generalization byshowing a dominance of convex-response func-tions. The convex functions are not universal, how-ever, and we have yet to identify the boundaryconditions associated with deviations from thelaw. Much research needs to be done before thisimportant law can be further refined.HOW ADVERTISING EXPOSURES AFFECTBUYING RATESThe advertising-response function can be definedas "the quantitative relationship between someinput of advertising and some output of pre-sumed value" (Simon and Arndt, 1980). The out-puts can he measured by salesthe change in anindividual's propensity to buyor through suchintermediate variables as advertising recall, atti-tudes, or intention to buy. Our focus, in this arti-

    cle, is on the immediate sales impact measuredlongitudinal, individual-level single-source da

    O ne of the most exciting and important EGcome from single-source analysis is that the sresponse to advertising exposures is convex (Figure 1).

    Convex and S-shaped response curves imthat, after a certain level of exposure, consumbecome saturated by advertising, and there isfurther increase in their propensity to purchasbrand. And anS-shaped curve supposes that this a threshold level of exposure, beneath whthe advertising is ineffective. By contrast, a linshape would imply that repeated advertisingposure produces a proportionately increasing ppensity to purchase a brand (Roberts, 1996).

    The convex response function was summedby Jones (1992) with his "once is enough" caphrase: within a short window before category pchase, an advertiser gaia s greatest return from reing the consumer with a single advertising exposThis pattern has significant implications for meplanning, suggesting buying com mercial spots wan emphasis on maxim izing reach, and for any gilevel of frequency of advertising exposure.

    Although authors like Broadbent (1999), Rerts (1996), and Jones (1992) have found that thmay be exceptions to convex-response functiothers (such as Vakratsas and Ambler, 1999)gest that theweight of empirical findings makpossible to generalize that the advertising-respofunction is convex.

  • 7/28/2019 Is Once Really Enough

    2/4

    ADVERTISING'S CONVEX SALES RESPONSE FUNCTIO

    EMPIRICAL GENERALIZATIONThe advertising response curve is "convex"the greatest marginal response is fromthe first exposures. As the number of cunnulative exposures in a period increases, themarginal effect of the advertising drops.

    in

    dPurcas

    e ftC DIner

    o Convex170.0152.5135.0117.5inn n70

    o Linear

    / ^

    1 2 3

    ' S-Shaped

    4 5 6Increased OTS

    ^

    7+

    Figure 1 HypotheticalShapes of the Advert is ingResponse Function

    NEW EXPLORATION: TESTING THEEMPIRICAL GENERALIZATIONUsing TNS's TVSpan single-source paneldata and examining all brands in each offour categories as a whole, the Short-Term Advertising Strength (STAS) resultsdisplayed convex-response functions (seeFigure 2).

    Within that generalization, however,there were brand-by-brand exceptions (seeTable 1). Do these exceptions imply thatthere are conditions under which the lawdoes not hold? Clearly there are measure-

    Figure 2 28-Day Advert is ing Response Functions, acrossProduct Categories, for All Brands

    ment issues (i.e., sampling errors) that caaffect the apparent shape of the responfunction for a brand. In particular, STAscores for small brands can be misleadinbecause they represent small changes ovea ba seline. For example, B ertolli spread hafrequency distributions of 687, 66, 44, 3and 6 3 . The STAS scores varied widely btween one and four or more opportunitieto see (OTS) (see Table 1) while the frequency d istribution show s that, in fact, Betolli's results were quite static.

    We may be able to generalize abouthe shape of the advertising-responsfunction based on a large body of exising research work. But there still arunresolved issues in the accurate measurement of the response function osales.

    CONCLUSIONTo be an empirical law, an empirical finding must hold approximately, but consistently, across some known conditionDecades of single-source analysis on TVcomm ercial sa I es-response functions keereporting convex sa les-response fLinctionseven though the analysis has been don e idifferent ways by different researchers, idifferent countries, on data collected diferent ways, on different product categories and brands.

    The body of research would suggesthat the convex sales-response functiois law-like. But empirical laws are nomerely discovered; there is always ongoing work on the generalizability of different laws. And this article demonstratethat there still is a great deal of researcto be done to fully understand the convex sa les-response function. Non ethelesswe propose:

    EGl: For television advertising foestablished consumer goodbrands in mature markets, wheadvertising has a positive sale

  • 7/28/2019 Is Once Really Enough

    3/4

    ADVERTISING'S CONVEX SALES RESPONSE FUNCTION

    BvRON SHARP is a professor and thedirector of theButter andButter Substitutes 28-Day Response Functions of E h r e n b e r ^ - B a s s i n s t it u t e a t t h e u n i v e r s i ty o f s o u t hBrands with Positive STAS Scores: Ordered by Market Share A u s t r a l ia . T h e i n s t i tu t e s f u n d a m e n t a l r e s e a r c h snpcppnHnt used and financially supported by many of the worl

    leading corporations including Coca-Cola, Kraft,OT S Kellogg's, British Airways, Procter & Gambie, Nielse

    N 1 0 3 1 4 4 5 585 3788 2,744 8,069 TN S , Turner Broadcasting. Network Ten, Simplot,1-' '.' '" ]. 2 3 A Mars, and many others. Dr. Sharp has published o

    100 academic papers and ison the editorial boardFlora 100 102 99 104 98 ,five journals.Anchor 100 109 100 99 87Olivio 100 110 104 96 72

    REFERENCESGolden Churn 100 146 __65 125 126Bertolli 199 h'^??. h l hl .??9 BROADBENT, S. Wlwn to Advertise. Hen

    Tliames, U.K.: Admap Publications in assoc , , , . , ^ - , .,, tion with Incorporated Society of Brieffect, the short-term adverts- ofOTS to correspond with an exposure.__ Advertisers, Institute of Practitioners in Adme response function 15 typi- fm) ^ , tising, 1999.cally, but not always, convex.

    . . . . JENNIFEB TAYLOR is a research associate at the , , ,, . . J i- 1 i . The law as itstands today is important, JONES, ]. P. How Mucit is Enough? Gettin, ,, , , , EhrenbergBass Institute for Marketing Science. Uni- ., , ,.. AJ^- nii Mpractical, useful knowledge. But there are Most from Your Advertising Dollar. NewY, , , . versity of South Australia. Herarea of research in- , , ._exceptions tha t require further inves tiga- Lexington tiooks, iyy..

    , ^ , eludes media planning and advertising. She istion to look for subpa t tems .^ , , , , 1 currently completing a PhO examining the advertising , ,r-, ir L- UL ^-To bett er understand the boundary J K& e ROBERTS, A. "What Do We Know about Adv, ,. , response function and its measurement. ... , , ,_ ^ -, . r- .

    conditions, th e role that schedulmg has tismg's Short-Term Effects? Admap, Februon th e measurement of the sales effects 1996of advertising should beinvestigated. RACHEL KENNEDY is an associate professor and theWill this relationship hold with new head of the Advertising Research Groupat the S I M O N, J. L., and j . ARNDT. "The Shape oftechnologies that pu t viewers incon- Ehrenberg-Bass institute for Marketing Science, Uni- Advertising Response Function."/tnrnd/o/trol of what they watch and when? Re- versity of South Australia. She isinvolved in commer- vertising Research 20, 4 (1980): 11-28.search reexaniining the relationsh ip cial market research, to understand buyer or socialbetween OTS and actual exposures may behavior, and marketing interventions iike advertising, VAKRATSAS, D., and T. AMBLER. "HOW Adalso be needed to clarify whether adjust- aswetlasacademc research into advertising and the tising Works: What Do We Really Know?" ments need to be made tothe numbe r dynamcsofmarket structure. nal of Marketing 63, 1 (1999):

  • 7/28/2019 Is Once Really Enough

    4/4