(ehrenberg & goodhardt 2000) new brands near instant loyalty

therefore been widely modelled interms of consumers so-called depth ofrepeat, meaning that repeat-buyingwould tend to increase whether aconsumer had already bought the itemonce, twice, three times, etc.27

Despite 25,000 new products a yearbeing launched in the USA alone,8

Hardie and his colleagues,9 Wright andSharp,10 and others have long notedhow there was in fact little systematicknowledge about how loyalty to newbrands develops:

One thing is certain there is no ruleabout the level of repeat-purchasing to beexpected at different levels ofpenetration.11

We expect there will often be a periodof instability during which consumerspreferences for the new product areevolving.12

INTRODUCTIONAn exploratory analysis of 23 successfulnew brands or line extensions has shownan unexpected but clear finding: the newbrands purchase frequency is almostinstantly normal. So are most otherstandard loyalty-related measures. Theexception is quarter-by-quarter repeatbuying which was consistently low.Hence it seems that there were someinitial triers of the new brand whonever became loyal to it at all. Butcustomers who did adopt the brand wereloyal from the start.

Traditionally, however, it seems tohave been thought that a new brandsrepeat-buying loyalty would growslowly over time (see, for example,Franzen1 and many earlier references the authors plan to survey suchprofessional expectations more formally).In the past, the growth of loyalty has

Henry Stewart Publications 0967-3237 (2001) Vol. 10, 1, 9-16 Journal of Targeting, Measurement and Analysis for Marketing 9

Academic Papers

New brands: Near-instant loyaltyReceived (in revised form): 11th May, 2001

Andrew Ehrenberghas been Professor of Marketing at South Bank University since 1993. He was previously 23 years at London BusinessSchool and in industry for 15 years. He has also held academic appointments at Cambridge, Columbia, Durham, London,NYU, Pittsburgh and Warwick.

Gerald Goodhardtis Emeritus Professor, City University; Visiting Professor, Kingston University; Adjunct Professor, University of South Australia;Visiting Research Associate, South Bank University. He was formerly Sir John E. Cohen Professor of Consumer Studies, andDean of the City University Business School. He spent 20 years in industry and commerce prior to 20 years as an academic.

Abstract It is widely thought that loyalty to successful new brands or line extensionsevolves slowly. An unexpected but striking finding therefore is that loyalty to a newbrand is nearly instant in some 20 cases examined so far: the new brands averagepurchase frequency at launch is already normal, ie at the same level as a year or twolater and also as that of competitive established brands. The finding was unexpectedbut now makes much sense with hindsight. More empirical work is in hand.

Andrew EhrenbergSouth Bank BusinessSchool, SouthwarkCampus, London Road,London SE1 0AA, UK.

Tel: 44 (0)20 7815 6169;Fax: 44 (0)20 7815 6166;e-mail: [email protected]

much the smallest sample base and notstatistically significant). Nonetheless, moreextensive follow-up work is now beingpursued. The product categories (newbrand cases) were:

antidepressants (2) cereal bars (1) chocolate biscuits (1) coffee (1) detergents (11), mainly the then new

liquid brand variants fruit drinks (1) shampoos (1) tea (1) toothpaste (4).

RESULTSConsumers purchasing of a given BrandX in an analysis period like a quarter of ayear can be broken into two main factors:

the penetration: the percentage ofconsumers who bought Xat least once in the period

the purchase rate: how often onaverage buyers of X bought in theanalysis period.

The main finding is that virtually fromthe start, a successful new brands averagepurchase rate was at or near its subsequentnormal levels, and equal to the rates forthe established brands in the category.This also occurred for all but one of theother loyalty-related measures analysed.

In contrast, a new brands quarterlypenetration at times increased greatly,especially for a real winner (as in Table1). In many other cases the number ofbuyers in a quarter levelled out quitesoon.

An example: Prozac

The antidepressant Prozac was verysuccessful when it was launched in the

The study

This paper therefore seeks to examinehow newly-launched brands performedon a variety of repeat-buying andbrand-switching performance measures.The approach was explicitly exploratory,since with new brands it was not knownwhat to look for.

Extensive empirical regularities andmatching theory (the NBDDirichletmodel) have however long been availablefor established brands.1318 This providedgrounded benchmarks and a conceptualbasis against which to evaluate theperformance of new brands.

The analysis was for newly-launchedbrands or line extensions which had beenbroadly successful, ie available for retailsale for at least a year or two.19 Thecases were mainly for grocery products,selected for the study by Taylor NelsonSofres (TNS), and two prescription drugsprovided by Dr Philip Stern. New brandfailures (or near-failures) have beenoutside the scope of these early analyses,since detailed in-market data for themare more sparse.

Methods

For simplicity, data suppliers were askedto use their standard quarterly analysisperiods. Sometimes data for three or foursuch quarters from launch were available,sometimes more. The first quarter ineach case was then partially discounted,since the precise timing of the launch inthe quarter was unclear (hence thenear-instant in the title of this paper).

Listed below are the five productcategories prescription drugs, food,drink, personal and household cleaners in which there is a total of 23 new-brandor line-extension cases, as well ascomparison data on almost 100 establishedbrands as benchmarks. All but one ofthese cases gave the same instant-loyaltytype of result. (The exception was on

10 Journal of Targeting, Measurement and Analysis for Marketing Vol. 10, 1, 916 Henry Stewart Publications 0967-3237 (2001)

Ehrenberg and Goodhardt

detergent case is excluded, asmentioned). For conciseness, the twotables show the first three quarters andthe last one available, this result beingtypical also of the intermediate ones.

The observed buying rates in Table 2were broadly steady from the first fullpost-launch Quarter II onwards, at about1.9 on average. They are also close tothe established brands benchmark figures,averaging also 1.9 in the last column.This was the case for really new brandslike Prozac, as well as for line extensionssuch as the new liquid detergents.

The leading established brands had largermarket penetrations than most of the newbrands, as was to be expected. They alsohad somewhat higher purchase rates, as amanifestation of the well-established doublejeopardy (DJ) phenomenon.20,21 But therecent adjustment did not affect theconclusion (eg the average purchase ratefor the detergents cases only reduced from2.2 to 2.0).

Market penetrations of the new brandsalso stabilised soon, as shown in Table 3,other than for the two medical

early 1990s. The percentage of doctorswho prescribed Prozac at least once in agiven quarter rose some 20-fold over thefirst two years, from a penetration of 1per cent in Q.I (the launch quarter) inTable 1 to 21 per cent in Q.VIII twoyears later.

In contrast, the average number ofnew Prozac prescriptions written in aquarter per prescribing doctor was almostat once at a rather steady level of around2.3, ie after the typically mixed-uplaunch quarter. (It is unclear whenprecisely in the quarter Prozac becameavailable to particular doctors). Thesomewhat higher prescription rates at 2.9in Quarters VI and VII was not typicalfor the other data. Overall, Table 2shows that repeat-buying loyalty wasalmost instantly at its longer-term level.

The various new brand cases

The outcome for 22 of the new brandswas similar, as summarised in Table 2 forquarterly purchase rates, and in Table 3for penetration. (One small-sample


New brands: Near-instant loyalty

Table 1: Prozac

Quarters Average(QI) II III IV V VI VII VIII QIIVIII

Percent prescribingAverage prescription rate*

11.0

32.3

82.2

101.7

171.9

182.9

172.9

212.3

142.3

*Average new prescriptions a quarter per prescribing doctorThe brand launched sometime in this quarter

Table 2: New brands quarterly purchase rates (average purchase frequencies per quarterly buyer)

Quarters

(I) II IIILastquarter

Establishedbrands*

Pharmaceutical drugs (2 cases)Mixed products (6 cases)Detergents (10 cases)Toothpaste (4 cases)

Average (22 cases)

1.51.11.41.5

1.4

1.91.81.91.3

1.8

2.11.92.01.4

1.9

2.51.92.01.4

2.0

2.32.22.01.4

1.9

Launch quarter.Cereal bars, chocolate biscuits, coffee, fruit drinks, shampoos, tea.* After DJ adjustment.

attract an exceptional incidence ofonce-only triallists (perhaps for the firstyear or two), because, perhaps, of thebrands newness, its exceptionalconcomitant publicity and retail display.The question of such triallists loyaltydoes not, however, really arise since theywill so far have bought the item onlyonce. The loyalty issue here is thereforethat those consumers who do becomeloyal do so near instantly, as the evidenceso far shows, rather than building upslowly, as is traditionally supposed.

Previous new-brand cases

Instantly normal average purchasefrequency had already been reported in afew isolated earlier new brand cases,2729

but these cases were discounted asunexpected aberrations. The fact thatthe new brand looked like an existingbrand so quickly is a little curious.30 Thereceived wisdom was that new brandloyalty had to develop slowly and theauthors themselves had previouslyaccepted this. Only now, with a range ofsome 20 cases virtually all giving thesame instant-loyalty outcome, do theearlier isolated cases seem propheticrather than aberrations.

Supportive evidence also comes fromthe Unilever/RBL/Research-International Minivan test paneloperation. This was a mobile groceryshop that signed up a panel ofhousewives who could shop there once a

blockbusters which went on increasinggreatly to a penetration of 22 per cent ofthe population (here Prescribing GPs).A possibility is that most new brandssimply settle down quickly in all respects,not just in their loyalty levels.

Other loyalty measures

The average frequency with which abrand is bought by its buyers in a periodsuch as a quarter is one measure relatingto its customers loyalty. It has beenwidely shown for established brands thatthis correlates closely with othermeasures which relate to loyalty, such asthe incidence of 100 per cent-loyalbuyers, the brands shares of categoryrequirements, and generally the levels ofperiod-to-period repeat buying.22,23

Quarter-by-quarter repeat-buyinglevels were, however, consistently lowfor the new brand cases here. At anaverage of 24 per cent they are muchless than the norm for established brands(about 50 per cent). Period-to-periodrepeat-buying measures would beespecially sensitive to the occurrence ofonce-only triallists, who are likely withnew brands. This could be checked asand when more detailed tabulations areavailable, by using so-called conditionaltrend analysis. (CTA analyses the repeatbuying of light and heavier buyersseparately, together with interpretativetheoretical norms.2426

The suggestion is that new brands do



Table 3: New brands quarterly penetrations (percentage buying the item at least once in the quarter)

Averages

I II IIILastquarter

Pharmaceutical drugs %Mixed products %Detergents %Toothpaste %

Average %

1220

1

6471

4

10452

5

22461

8

Launch quarter

image-related) for choosing brand Arather than brand B. Brands are thoughtto possess different levels of equity,either strong or weak. The authorscontrary view has long been that acompetitive advantage is soon copied.Competition therefore consists ofmatching rivals on what matters, ratherthan of being different.35,36 This view isnow strengthened by finding that evenoperational loyalty measures for newbrands are normal, implying that they arealso not seen as different.

New brands having the same degreeof loyalty as established brands also fits inwith the finding that similar brandsappeal to similar kinds of consumersmore generally ie that competitivebrands do not segment the market.37

Attitudinally, for established brands, usersof brand A look at A much as users ofthe competitive brand B look at B.38 Itnow needs to be established whether ornot users of a new brand also view it inthe same way.

It has also been shown that temporaryprice cuts for established brands, ie dealsor price promotions, appeal virtuallyonly to past customers.39,40 Pricepromotions therefore do not recruit newcustomers for established brands. For newbrands, however, the mechanism mustdiffer, since there are no past customers.This needs to be explored further.

Received wisdom is that advertising ismostly persuasive, eg that advertisingmotivates the consumer to become loyalto the brand. The authors have howeverlong been arguing that this is not so.41

They believe that the finding of instantloyalty to new brands further underminesthe traditional persuasive views becausethere is no time to persuade. It supportsinstead the alternative perspective thatadvertising works as publicity.

Unsuccessful new brands have notbeen covered by the exploratory studyhere. The causes of the failure of new

fortnight as one of their retail outlets,under experimentally controlledconditions.31 The Minivan attracted agood deal of research interest frommanufacturers at the time, but wasultimately abandoned. It rather accuratelypredicted a new brands real-life repeatbuying but failed to predict its crucialmarket penetrations. That is in line withthe new instant-loyalty result here.

It is also understood that commercialvolume forecasting of services nowadaystend to use the well-establishedNBD/Dirichlet model or related resultsto predict the test brands repeat buyingaccordingly. The more difficult to predictkey to a new brands success is itsultimate market penetration ie thenumber of new customers, rather thantheir loyalty (which depends on thevariety of marketing-mix inputs).

WIDER IMPLICATIONSThe new brand finding here has widerimplication. Thus the authors havepreviously argued that any loyaltymeasure for a successful new brand mustultimately settle down to the typicalnorm for all brands in the productcategory.32 This has had powerfulplanning implications.33 The newinstant-loyalty finding for new brandsnow greatly strengthens this conclusion.

More generally, the authors haveargued that what matters is whether abrand enters a consumers considerationset, ie whether it stands out, is familiar,and commands assurance, which theauthors call salience-plus.34 Instantloyalty supports this because all that canreally matter to a brands success is tohow many consumers the brand is salientin that way.

The traditional view of competition isthat a brand must be differentiated:consumers are thought to need a reason(functional or emotional/



some weeks, or slowly rotting fabricwith Persil Power), often despiteextensive pre-launch testing andwarnings from friendly competitors.

Case 6: Honeymoon effects wearingoff (eg an attractive flavour whichcloys with extended use, probably likethe new Coke a few years ago).

DISCUSSION

The research outcome of this exploratorystudy, that successful new brands havenormal loyalty virtually from the start,seems sufficiently straightforward forcompanies to incorporate it into theirnew product thinking (planning andevaluation), and for academics toincorporate it in their relevant teachingand research.

Although largely unexpected, thefinding makes sense with hindsight. Thusit is believed that experienced consumersgenerally know that competitive brandsare similar and hence largelysubstitutable, which is why the brandsare in fact competitive. No extensivenew learning about the new brandsattributes is therefore needed: theconsumer can either choose it as anormal repertoire brand, or not.

Possibilities for follow-up research nowinclude more systematic and in-depthstudies of successful new brands in otherproduct or service categories (egtelevision programmes) and in othercountries, including emerging markets.And also studies of unsuccessful newbrands.42,43 The task of collecting suitabledata is eased now that one knows betterwhat to look for.

AcknowledgementsThis paper is based on work in the R&D Initiative, aresearch programme at South Bank University. Theresults were given in a prizewinning paper given at theMarketing Academy Conference, 2000 and recorded assuch in the Journal of Marketing Management.

brands can however be very varied,beyond mere lack of repeat buyingloyalty. This is illustrated below withsix cases from the authors earlierexperience.

Case 1: In the 1960s Beechamslaunched a one-shot shampoo, LeafShampoo, a green, hand-sized andleaf-shaped specialty paperimpregnated with soap to rub on wethair. This sold like a bomb: it seemeda very attractive idea. But it had norepeat buying as the wet paper wasmessy to dispose of after use, whetherin a shower or the bath. This shouldhave been spotted by pre-testing.

Case 2: Another new product faileddespite much well-designedpre-launch product research. But topmanagement was so committed to itsnew venture that it did not take onboard that the research had in factalready given it the thumbs down.

Case 3: Shell launched Four Seasonsin the 1960s. It was the firstall-year-round multi-grade motor oil.But they ignored that having priced itlow to compete with the thenfast-growing DIY Castrol,filling-station managers/owners wouldnot actively sell Four Seasons becauseof its lower margin, compared withthe standard and more expensive ShellOil. In addition the ink on thedisplay posters for Four Seasons ran inthe rain.

Case 4: Tate & Lyle launched asyrup for puddings, ignoring thatthere were already competitivebrands in the market. Hence theirnew syrup did not get the effortlessnear-100 per cent retail distributionto which they were accustomed fortheir other near-monopoly sugarproducts.

Case 5: Delayed after-effects are notuncommon (eg skin irritation after



Society, Vol. 147, pp. 621655.16 Stern, P. (1995) Prescriptions for branded and

generic pharmaceuticals, Journal of BrandManagement, Vol. 2, No. 3, pp. 177183.

17 Stern, P. and Ehrenberg, A. S. C. (1995) The marketperformance of pharmaceutical brands, Marketing andResearch Today, November, pp. 285292.

18 Sharp, B. M. (1999) Evaluating the impact of aloyalty program on brand loyalty, Doctoral Thesis,University of Adelaide, Adelaide.

19 See also Davis, E. J. (1964) Test marketing: Anexamination of sales patterns found during forty-fourrecent tests, The Market Research Society, Researchin Marketing, London.

20 Ehrenberg, A., Goodhardt, G. J. and Barwise, T. P.(1996) Double jeopardy revisited, Journal ofMarketing, Vol. 54, pp. 8291.

21 Ehrenberg, A. S. C., Barnard, N., Kennedy, R. andBloom, H. (2001) Brand advertising as publicity,Journal of Advertising Research, (submitted).

22 Ehrenberg (1972/1988) op. cit.23 Ehrenberg, A. S. C., Goodhardt, G. and Uncles, M.

(2001) Using benchmarks in understanding buyerbehavior, Journal of Marketing, (submitted).

24 Goodhardt, G. J. and Ehrenberg, A. S. C. (1967)Conditional trend analysis a breakdown by initialpurchasing level, Journal of Marketing Research, Vol.4, pp. 155161.

25 Morrison, D. G. (1969) Conditional trend analysis:A model that allows for non-users, Journal ofMarketing Research, Vol. 6, pp. 342345.

26 Ehrenberg, Goodhardt and Uncles (2000) op. cit.27 Ehrenberg, A. S. C. and Goodhardt, G. (1968)

Repeat-buying of a new brand, British Journal ofMarketing, Vol. 2, pp. 200205.

28 Wellan, D. M and Ehrenberg, A. S. C. (1988) Asuccessful new brand: Shield, Journal of the MarketResearch Society, Vol. 30, pp. 3549.

29 Wright and Sharp (1997) op. cit.30 Ibid.31 Charlton, P., Ehrenberg, A. S. C. and Pymont, B.

(1972) Buyer behavior under mini-test conditions,Journal of the Market Research Society, Vol. 14, pp.171183.

32 Ehrenberg, A. S. C. (1991) New brands and theexisting market, Journal of the Market Research Society,Vol. 33, pp. 285299.

33 Ehrenberg, Goodhardt and Uncles (2001) op. cit.34 Ehrenberg et al. (2001) op. cit.35 Dawes, J. and Page, N. (2000) Drivers of service

quality, Working Paper, University of SouthAustralia.

36 Ehrenberg et al. (2001) op. cit.37 Kennedy, R. and Ehrenberg, A. (2000) Competitive

brands user-profiles hardly differ, Proceedings ofthe MRS Conference 2000, The Market ResearchSociety, London.

38 Barnard, N. R. and Ehrenberg, A. S. C. (2000)Consumer attitudes and brand-choice, South BankUniversity, R&D I Report.

39 Hammond, K., Ehrenberg, A. S. C. and Goodhardt,G. (1996) Market segmentation for competitive

The authors were assisted in this work by theircolleagues Dr Rachel Kennedy, Dr Neil Barnard, JohnBound, Dr Carl Watson-Gandy, Jaywant Singh andMaria Clemente, and by very helpful drafting commentsfrom Helen Bloom and an anonymous reviewer.

The authors are indebted to TNSofres for data setsfrom the AGB Superpanel based on some 10,000 UKhouseholds, and to ISIS Ltd. and Dr Philip Stern forthe medical prescriptions data from ISIS Ltd.

The project was funded in large part by the ESRC,ROPA Grant RO22250076.

References1 Franzen, G. and Bouwman, M. (2001) The

mental world of brands: Mind, memory and brandsuccess, World Advertising Center, Oxfordshire,UK.

2 Fourt, L. A. and Woodlock, J. W. (1960) Earlyprediction of market surveys for new groceryproducts, Journal of Marketing, Vol. 25, pp. 3138.

3 Baum, J. and Dennis, K. E. D. (1961) Theestimation of the expected brand choice of a newproduct, 17th ESOMAR/WAPOR Conference,ESOMAR, Amsterdam.

4 Parfitt, J. H. and Collins, B. J. K. (1968) The useof consumer panels for brand-share prediction,Journal of Marketing Research, Vol. 5, pp. 131146.

5 Eskin, G. (1973) Dynamic forecasts of new productdemand using a depth of repeat model, Journal ofMarketing Research, Vol. 10, pp. 115129.

6 Kalwani, M. and Silk, A. J. (1980) Structure ofrepeat-buying for new packaged goods, Journal ofMarketing Research, Vol. 17, pp. 316342.

7 Hardie, B. G. S., Fader, P. S. and Wisniewski, M.(1998) An empirical comparison of new producttrial forecasting models, Journal of Forecasting, Vol.17, pp. 209229.

8 The Economist (1998) Market makers, 19th March,p. 67.

9 Hardie, Fader and Wisniewski (1998) op. cit.10 Wright, M. and Sharp, A. (1999) New brand effects

in a Dirichlet market. Working Paper, University ofSouth Australia, Adelaide.

11 Parfitt and Collins (1968) op. cit.12 Fader, P. and Hardie, B. (1999) Investigating the

properties of the Eskin/Kalwani/Silk model of repeatbuying of new products, in Hildebrandt, L.,Annacker, D. and Klapper, D. (eds), Marketing andcompetition in the Information Age, Proceedings ofthe 28th EMAC Conference, 1114 May, HumboldtUniversity, Berlin.

13 Ehrenberg, A. S. C. (1959) The pattern ofconsumer purchases, Applied Statistics, pp. 2641.

14 Ehrenberg, A. S. C. (1972, 1988) Repeat buying,Griffin, London. Reprinted in Journal of EmpiricalGeneralisations in Marketing Science (2000), Vol. 5,(http://www.marketing.unisa.edu.au/MSC/JEMS_intro.html).

15 Goodhardt, G. J., Ehrenberg, A. S. C. and Chatfield,C. (1984) The Dirichlet: A comprehensive modelof buying behavior Journal of the Royal Statistical



analysis of stimulated brand choice, Journal ofAdvertising Research, Vol. 13, pp. 2133.

43 Lomax, W. , Hammond, K., Clement, M. and East,R. (1996) New entrants in a mature market,Marketing Intelligence and Planning, Vol. 12, pp.281295.

brands, European Journal of Marketing, Vol. 30, No.12, pp. 3044.

40 Hammond, K. and Ehrenberg, A. S. C. (2001) Thecase against price-related promotions, Admap..

41 eg Ehrenberg et al. (2001) op. cit.42 Ehrenberg, A. S. C. and Charlton, P. (1973) An



(ehrenberg & goodhardt 2000) new brands near instant loyalty

Documents