diffusion of innovations: successful adoption needs more ... · down to badly judged marketing...

early adopters) instead of targeting theentire market at once, when introducinga new product/innovation. In order toimprove the odds on a successfulintroduction he states that it is necessaryto ‘single out’ those most interested inthe new product/service (the innovators),learn what drives them (which featuresof my new product make it interestingfor them?) and design communicationsspecifically to recruit and convince them.What he is not saying is how thatcommunication needs to be designed; orhow the introduction strategy has to berolled out. The purpose of this paper isto initiate a way of answering this

INTRODUCTION

‘We can also now focus marketing effort ontargeting innovators. Once we have singledthem out and understood what drives them,we can write and design ourcommunications specifically to recruit them.We can also choose whatever media are bestto reach them with greatest efficiency. Inshort, in the late 1990s, we have thecapability to focus on innovators. But westill have to know who they are.’1

The above citation is from an article inAdmap by John Carter (Media Solutions),in which he explains why it is initiallybetter to focus on the innovators (and

140 Journal of Targeting, Measurement and Analysis for Marketing Vol. 13, 2, 140–155 � Henry Stewart Publications 1479-1862 (2005)

Diffusion of innovations:Successful adoption needs moreeffective soft-DSS driven targetingReceived (in revised form): 7th April, 2004

Gino Verleyehas a Masters degree in economic psychology and wrote a PhD on the performance of missing data solutions for structuralequations models. He teaches research methodology and statistics at the University of Ghent in Belgium. His currentresearch deals with customer segmentation and forecasting of new technology adoption and diffusion.

Lieven De Marezhas a Masters degree in communication research. As a researcher at the Department of Communication Sciences, Universityof Ghent, he deals with models for technology adoption.

Abstract Despite the promising prophecies that usually go hand-in-hand with theintroduction of new technologies or any other innovation, they often turn out to be a bigdisappointment due to a bad introduction and communication strategy. In order toimprove these strategies a soft Decision Support System (DSS) was developed bycombining a measurement and segmentation tool, the Delphi method-based productspecific innovativeness (PSI) scale with the bowling pin marketing communicationmodel. This new combined approach enables a researcher to detect in advance (iebefore the actual introduction) the innovators, early adopters, early majority, late majorityand laggards for a certain innovation. Based on the resulting segments and profiles, it ispossible to develop an introduction and communication strategy that has a betterchance of successful customer adoption.

Lieven De MarezUniversity of Gent,Department ofCommunications Sciences,Korte Meer 7-9-11, B 9000,Ghent, Belgium.

Tel: �32 9 264 6885;Fax: �32 9 264 6992;e-mail:[email protected]

overload: too much too soon’.Companies try too fast to change thecustomer from a passive television-userto an active and interactive user of amultimedia platform. In the euphoria ofwhat is technologically possible, theysuddenly harass the market with lots ofapplications and possibilities. Would notit be better then to approach the marketfirst with a more basic product, with a‘basic digital package’? A package knownnot to frighten the larger part of themarket.

DIFFUSIONS OF INNOVATIONS:THEORYIf an attempt is made to link all this totheory, one soon ends up with‘innovation theories’. The basic work parexcellence within this tradition is ‘Diffusionof innovations’ by Everett Rogers(1962). Since 1962, many authors haveworked further on this theory, but thecore remained the same: when a newproduct/new technology is introduced,the target market can be divided intofive segments along an axis of riskaversion: in the beginning there are theinnovators, followed by the earlyadopters, the early majority, the latemajority and the risk-allergic laggards.

The ‘innovators’ and ‘early adopters’can be seen as kind of trendsetters. Theyhave an important ‘opinion leader-role’to play towards the rest of the market.One of the most important principles ofthese innovation theories is the principleof ‘copy-behaviour’:8 early adopters copythe behaviour and product usage of theinnovators, early majority copies theearly adopters, and in turn is copied bythe late majority . . . The innovators andearly adopters are by far the mostimportant segments: not only becausethey are the first generating cash flow,but also, and even more, because theyare the first that have to be copied.

question that was left open by Carter(and many others).

After a short sketch of the situation,the main methods of innovationforecasting2 that have been used untiltoday will be mentioned briefly. Thiswill be followed by a description of theauthors’ measurement and assignmentmethodology. All this is illustrated withthe case of digital television (dTV) inFlanders.3

dTV: A FAILING INNOVATION?Several reactions from players on marketswhere dTV has already been introduced,teach us that, again despite the promisingprophecies, the acceptance of digitaltelevision is not going as smoothly asexpected. At the presentation of theGerman digital zdf-platform, for example,the following remark was made: ‘Overmuch of Europe, digital pay-TV remainscharacterised by poor take-up . . . partlydown to badly judged marketingdecisions.’4

Apparently, the adoption of dTVdisappoints somewhat in Europe.According to the commentator above,Stuart Thomson (Cable & Satellite), thisis mainly due to badly judged marketingdecisions.

At the beginning of 2002 OnDigitalchanged its name to ITVDigital. Thereason for this change of name was thedisappointing introduction of OnDigital.5

In the Netherlands, the acceptance ofdTV is even more disappointing. Henkde Goede (of the cable companyCasema), had to dismiss 25 per cent ofhis personnel (300 out of 1,200), due tothe disappointing take up results ofdigital television. There were only twosolutions, according to de Goede: a‘drastic increase of the price or . . . anational innovation fund’.6

Fritha Sutherland7 described the badlyjudged marketing decisions as ‘remote

� Henry Stewart Publications 1479-1862 (2005) Vol. 13, 2, 140–155 Journal of Targeting, Measurement and Analysis for Marketing 141

Diffusion of innovations

digital television is coming sooner thanexpected (cf. Figure 1). It is not thatsimple to make the step from those firstgroups (innovators and early adopters) tothe mass market. A clear strategy isneeded to make the jump towards thatmass market. Geoffrey Moore10 comparesit with a chasm (cf. Figure 1 verticaldotted line) which has to be bridged —a chasm that has to be crossed.

HOW TO CROSS THE CHASM?HOW TO REACH THATMASS MARKET?Geoffrey Moore developed his bowlingpin model.11 In this model for theintroduction of innovations he states thatin a lot of cases a gradual introduction isbetter than an introduction by which thewhole market is offered the wholeproduct (with all its applications) at once.Instead of haphazardly trying to gaindifferent niches by promoting theproduct from the start with a wholebunch of applications, he is convincedthat it is better, in the initial stage, toconcentrate on a few very interestedsegments. By offering that one or thosefew segment(s) a product that istailormade for them (but can also be

Thinking back to Fritha Sutherland’sremark (too much too soon), this couldbe another reason to start approachingthe market with a basic package (insteadof the whole product at once). It doesnot seem very likely that the earlymajority will start copying the earlyadopters in their usage of sophisticatedapplications. For this reason, it seemsbetter to start with a basic product,which the mass market is known to beinterested in, and which stands a betterchance of starting the copy mechanism.In order to know what this basic productlooks like, to know which are the basicapplications most of the market isinterested in, a research tool is requiredthat will enable the marketer to make adistinction between the five innovationsegments for a particular new product inadvance.

The critical point in this adoptioncurve is somewhere around the place (ormoment) the early adopters have to startadopting (double vertical dotted line inFigure 1). The diffusion or adoptionpattern as presented by Everett Rogers9

is used to illustrate this.Cases like ITV Digital (UK) and

Casema (The Netherlands) illustrate thatthe turning point for the adoption of


Verleye and De Marez

Figure 1: Critical point in adoption or innovation curve

ITV

Casema

adopts the basic A1). At the same timeS1 is offered a second, new application(A2) — an application that will beadopted by S2 at the moment S3 adoptsthe basic A1, and when the firstsegment S1 is offered a third applicationA3, etc. . . .

As a consequence, the whole market isreached anyway after some time, offeringthe total product with all its applications,but in a way that has left more‘adjustment time’ for the market tobecome familiar with the newtechnology. A bit more time foradjustment can be crucial for a newtechnology because ‘uncertainty’ and ‘notbeing familiar’ with a new conceptappears to be one of the biggestthresholds for acceptance of newtechnologies.13,14

INNOVATION FORECASTING?There are three main methods of‘innovation forecasting’ that are beingused: bibliometrics, econometrics anddomain specific innovativeness (DSI)scales.15 Each of these traditions isinadequate to give an answer to thequestions above.

Bibliometrics is a research tradition in

situated within the domain of interest ofsome ‘adjacent niches’), the marketer canbe practically sure of their adoption.Once they have purchased the product,sales towards those adjacent niches willbe easier, because of the already existingword-of-mouth relations between thoseniche segments, allowing so-called‘references’. What Moore means withthis word-of-mouth advertisingmechanism is comparable with the ideabehind Rogers’ copy mechanism.

In Figure 2 the bowling pin model isschematically clarified. Initially the focusis on one segment (S1) which is knownto be very interested in the new product,because of one or a few application(s)(A1). The condition for the offeredapplication A1, or the application whichis emphasised in communications, is thatit has the potential to appeal to othersegments after a while. Once the firstsegment has adopted the application akind of bowling game starts. A secondsegment (S2) bearing a close resemblanceto the first one, sees howacquaintances/relatives from the firstsegment are using the new product, theword-of-mouth advertising occurs, andpeople start to show some kind of copybehaviour (and as a consequence S2



Figure 2: Bowling pin model12

S1 / A1

S2 / A1S1 / A2

S3 / A1S2 / A2S1 / A3

� � � � � �

communication strategy has yet to bedrawn up;

— from the resulting segmentation, amarketing decision support systemmust be developed, integrating thesegments with communication,product offer and pricing.

The first problem is that the marketerwants to question the consumer aboutsomething the consumer does not yetknow very well, or even at all. Thisimplies that, before being questioned theconsumer has to be informed completelyand objectively about the new concept.To do this efficiently, the interviewersneed to be well trained: they must allhave in mind the same, correct pictureof the innovation. Once the respondentis acquainted with the innovation, withthe new technology, the interviewerstarts with the questionnaire.

The PSI scale consists of threequestions, for which there are fivepossible answers (the same five for eachof the three questions). After reading thetext, watching the video or testing theinnovation, the respondent receives thefollowing question:

‘Suppose digital television was already on themarket. As you see dTV now, howinterested would you be in subscribing to adTV package?’

This first question is a very general one,asking about a general interest in digitaltelevision. There is no specification ofprice, content offer or applications. Theresponses are on a 5-point Likert scalewith answers from 1 (I would subscribeimmediately) to 5 (I certainly won’tsubscribe). The answers to this questionalone give a general indication of agroup of interested people, a group ofthose not interested and a very big groupof those in between. To split up thesegroups more specifically, the followingtwo questions are needed.

which research and forecasts are based onliterature research: people search theliterature for results and informationon/of existing tests, patents, substitutes,introductions in other markets andcountries, . . . and integrate them intoone document, on which conclusions arebased concerning the innovations whichare to be brought to the market. Theauthors do not believe that informationon previous cases can yield the properstrategy for a new innovative product.

The biggest problem the econometricstradition confronts is that it does notallow forecasts to be made (beforeintroduction, even when the product isstill at the R&D stage) concerning theadoption curve. The tradition allows veryaccurate forecasts to be made about thepattern of the curve but it requires atleast six to 12 months of data.

DSI scales do allow forecasts to bemade, but the authors do not belive theyare specific enough. Different authors16,17

have developed scales to measure thedegree of innovativeness of a person.Innovativeness is seen as a personalitytrait and is measured by a series ofLikert-items. One of the most frequentlyused DSI-scales is the one developed byGoldsmith and Hofacker.18 On thevalidity and reliability of these DSI-scales,there have been several remarks.19

PSI APPROACHIn order to overcome these deficienciesthe authors set out a new approach, theproduct specific innovativeness (PSI)scale. The scale has to meet certainconditions, it must be:

— simple (the questions must be clearenough to be used in a telephonesurvey, as well as a personal interviewor a street interview);

— applicable when the product still hasto be launched, even when the



product-specific innovation segmentationis obtained. In order to produce areliable segmentation categoryassignment, five experts were recruitedand asked to label each combination ofanswers on the three PSI questions. Thefive evaluations constitute ordinalindicators and are combined into onefinal assignment (average inter-expertcorrelation � 0.76; CronbachAlpha � 0.94). This methodology isinspired by the Delphi method whichhas proven added value in the context ofadoption forecasting.21–23 The dotted lineshows the innovation curve as describedin theory. The full line shows the curveas forecast in the authors’ research, usingthe PSI scale. Although this line followsmore or less the theoretical pattern,fewer laggards and a bigger mass marketare found. As for most other innovationsor new technologies, there seems to be asmall group of people in Flanders willingto adopt digital television immediately.But there is also a large ‘waiting and still

‘If you were offered an optimaldTV-package20 (only the channels you areinterested in, the services you want to use,and all that for a price that isn’t higher thanthe price you are willing to pay for it), howinterested would you be in subscribing?’

In the foregoing discussion with theinterviewer the main interests of therespondent will have become clear (egsomeone only interested in an electronicprogramme guide (epg), e-mail, andmore thematic channels (sport andnature)). Knowing this, the interviewercan formulate the third question:

‘If you were offered a dTV-package that isnot your optimal package, but deviatingfrom it on a certain point (higher price, notall your favourite features . . .), to whatdegree would you be interested insubscribing?’

Using a gradual or stepwise allocation(see Figure 3) on a sample of 621respondents (see below), a



if (question1 � 1) adopcat � 1....if (question1 � 2 and question2 � 1 and question3 � 1) adopcat � 2.if (question1 � 2 and question2 � 1 and question3 � 2) adopcat � 2.if (question1 � 2 and question2 � 1 and question3 � 3) adopcat � 2.if (question1 � 2 and question2 � 2 and question3 � 1) adopcat � 2.if (question1 � 2 and question2 � 2 and question3 � 2) adopcat � 2....if (question1 � 2 and question2 � 1 and question3 � 4) adopcat � 3.if (question1 � 2 and question2 � 1 and question3 � 5) adopcat � 3.if (question1 � 2 and question2 � 2 and question3 � 3) adopcat � 3.if (question1 � 3 and question2 � 2 and question3 � 5) adopcat � 3....if (question1 � 3 and question2 � 3 and question3 � 3) adopcat � 4.if (question1 � 4 and question2 � 3 and question3 � 4) adopcat � 4.if (question1 � 4 and question2 � 3 and question3 � 5) adopcat � 4....if (question1 � 4 and question2 � 4) adopcat � 5.if (question1 � 4 and question2 � 5) adopcat � 5.if (question1 � 5) adopcat � 5.

Figure 3: Allocation to the five innovation segments based on the three questions (1: innovators; 2: earlyadopters; 3: early majority; 4: late majority; 5: laggards)

3 Compared to my friends I own a fewof (a lot of ) . . . . . . . . . .

4 In general, I am the last (first) in mycircle of friends to know thetitles/brands of the latest . . . . . . . . . .

5 I will not buy a new .. . . . . . . . . if Ihaven’t heard/tried it yet. (I will buy anew .. . . . . . . . . if I haven’t heard/triedit yet.)

6 I (do not) like to buy . . . . . . . . . .before other people do.

Based on this scale the followingsegmentation is obtained (based on thepercentiles on the sum scale with the sixLikert items): five segments (innovators,early adopters, early majority, latemajority and laggards) according to theirinnovativeness concerning the domain‘ict’ (information and communicationtechnologies). It speaks for itself that theresult of both segmentations partlycorrelates. The significant chi-square test

hesitating’ majority: people who onlyadopt when digital television meetscertain conditions (price, applications . . .).The next step is to find out what thoseconditions are, who those people are andhow to reach them.

PSYCHOMETRIC QUALITIESNext to the high inter-experthomogeneity, the concurrent validity ofthe PSI scale is assessed by integratingthe DSI scale of Goldsmith andHofacker.24

This scale consists of the following sixitems:

1 In general, I am among the first (last)in my circle of friends to buy a new.. . . . . . . . . when it appears.

2 If I heard that a new .. . . . . . . . . wasavailable in the store I would (not) beinterested enough to buy it.



Figure 4: Innovation curve: theory (or results based on DSI) vs. dTV-Flanders (based on PSI)

Adoption curve DTV Flanders

0

5

10

15

20

25

30

35

40

45

50

Adoption category

Per

cent

age

of p

opul

atio

n

theory 2,5 13,5 34 34 16research 4 15 45,9 28,2 6,9

innovators early adopters early majority late majority laggards

the reality of making business decisions.Softer DSS also copes with‘arithmo-morphism’, the belief that thereis a unique, mathematically-defined,optimal decision where there is no placefor multiple measurements orsubjectivity.27 This softer decision aid isexactly what this paper is all about:integrating a new measurement technique(PSI) within marketing managementdecision aids with room for thesubjective.

THE SAMPLEFor this research the authors contactedover 1,000 Flemish households, andasked them if the decision makers for thepurchase of (new) communicationtechnologies could be interviewed. 621of them agreed to this (N � 621). Theyhad to fill out a ten-page questionnaire,including questions on sociodemographicsituation, media possession and usage,interactive buying behaviour, lifestyle,DSI scale of Goldsmith and Hofacker(six Likert items), PSI scale, price-settingand interest in dTV applications. Table 2describes the sample.

Because most people did not knowexactly what digital television was, orthey had too positive an image of it, therespondents first had to read anintroductory text (two pages) on digitaltelevision. Afterwards, this text wasdiscussed with the interviewer. This tookabout ten to 15 minutes. Only when theinterviewer was convinced therespondent had a precise and correctimage of digital television did they go tothe actual questionnaire.

THE FIVE INNOVATIONSEGMENTS FOR DIGITALTELEVISION IN FLANDERSBased on the answers to the PSI scale itwas possible to split up the Flemish

for Table 1 and the nature of thediagonal reflects this (chi-square � 72.7;df � 12, p � 0.000). Treating PSI valuesas an ordinal variable, a LISRELmeasurement model using all the Likertitems behind the DSI scale in one latentvariable was fitted to the data. In thismodel there is a 0.34 correlationbetween the DSI and the PSI values.The LISREL model has excellentgoodness-of-fit: delta2 � 0.979,rho2 � 0.966. Psychometric MokkenScalability analysis of the three items,assuming a cumulative order in the threeitems, yields H � 0.73 and scalereliability rho � 0.85 — indicating areliable cumulative phenomenon.Predictive validity of the method hasbeen tested in a GPRS study (cellularphones) and was found accurate.25

In Table 1 the innovation segments forthe domain ict (DSI scale) are in therows, and the innovation segments forthe product digital television (PSI scale)are in the columns. Although these twosegmentations correlate to a certaindegree, it is clear that these twosegmentations are not completely thesame. Besides the already mentioned 0.34correlation, there is also a dense diagonal(figures in bold), illustrating the linearrelationship between the twosegmentations.

TOWARDS A SOFT MARKETINGDSS FOR DIGITAL TELEVISIONEver since the Second World War,strategic decisions have been driven andsupported by quantitative models, usuallyknown as operations research models ordecision support models. Such modelsare usually purely quantitative in naturealthough some authors see the emergenceof more qualitative approaches that stillrely on mathematical or statisticaltechnology.26 Such ‘softer’ DSS are moreflexible and more suited to coping with



communication technologies of thosehouseholds): 4 per cent innovators and15 per cent early adopters. The ‘waitingmass market’ consists of 45.9 per centearly majority and 28.2 per cent latemajority. The not interested group, thelaggards (6.9 per cent) appears to besmaller than predicted by theory.

For the five categories as a whole apositive correlation with age (Pearsonr � 0.125, p � 0.02) and the number ofchildren is found (r � 0.105, p � 0.012),and a negative correlation with income(r � –0.234, p � 0.000). This means thatthe older someone is, or the morechildren they have, the bigger the chanceof being located somewhere at the rearof the innovation curve. Higher incomeearners on the other hand are moreinnovative for digital television. So, thosemost interested in digital television areyoung people, without children and withan income that is higher than average.Apart from this, these five segmentsdiffer, of course, on a lot of other

sample into five innovation segmentsaccording to their interest or degree ofinnovativeness towards digital television.

In his article ‘why settle for the earlyadopters?’, Carter said ‘in this way, wecan see that 2.5 per cent should be takenas an indicator of the numbers one mightfind in any particular market, rather thana hard and fast rule’.28 According to him,it is wrong to stick to those fixedpercentages of 2.5 per cent innovators,13.5 per cent early adopters . . .. Thistheoretical percentage distribution onlyhas to be seen as a global indicator. It isimportant to remember that there isalways a small group that adopts and asegment showing no interest at all. Inbetween there is the big mass market. Byusing a segmentation, as described above,this is taken into account. So, whatRogers and Schoemaker and Watkins29

called ‘earlier adopters’,30 is for digitaltelevision in Flanders a segment of 19per cent of Flemish households (or atleast the decision makers for new



Table 1: PSI vs DSI segmentation

PSI innovation segments (specifically dTV)

InnovatorsEarlyadopters

Earlymajority

Latemajority Laggards Total

DSI innovationsegments (ict)

InnovatorsEarly adoptersEarly majorityLate majorityLaggardsTotal

34

1161

25

33136212

93

945

1099428

285

414557032

175

14

10141443

2098

22120577

621

Table 2: Distribution of education, income and age

Education LO:1.4%

Lsec:8.5%

Hsec:31.4%

HOKT:28.8%

HOLT:12.1%

University17.7%

Income <900e:13.4%

900-1250e

22%1250–1750e

28.4%1750–2500e

21.8%>2500e

14.4%

Age <202.1%

2028.4%

3018.4%

4032.4%

5015.2%

>603.5%

Notes: LO: Lager Onderwijs (school for children aged under 12); Lsec: school for children aged 12–16; Hsec:school for children aged 16–18; HOKT: a three-year higher education course; HOLT: a four-year highereducation course

This group, however, is somewhat older(38 years), more environment consciousand social minded compared to the othersegments. (‘Social minded’ meansspending more time with family andfriends.) This segment, with the biggestshare of non-married couples livingtogether, surfs the most (over one hour aday) and is medium viewer (1.75 hours aday). Surfing is something they do forleisure as well as for professionalpurposes.

The early majority have an averageage of 40 years and are willing to payabout 20e a month for digital television.It is a segment with significantly morelower/mid-level employees, reading a lotof newspapers and using PCs mainly forprofessional purposes. Socially, they havea ‘lower profile’ (less social activities andspending less time with friends andfamily) than the other segments. Theyare medium surfers (about 45 minutes aday), but they do this mainly for work,not in their leisure time. They watchabout 1.75 hours television a day.

The late majority want to pay 15e fordigital television and have, except for thebigger share of lower incomes (<900e

and 900–1250e), a similarsociodemographic profile to the earlymajority (average age also around 40).This late majority segment especiallydistinguishes itself by its media usage: itreads fewer newspapers and magazines,but watches a lot of television (over twohours a day). They are less experiencedwith the internet (30 to 40 minutes aday).

For the laggards the monthly price fordTV cannot be more than 7.5–9.5e amonth. It is an older (>45 years),low-educated segment with a lowincome. As the late majority, they alsoshow a low media usage (for newspapersas well as magazines, internet andtelevision). Although almost everyone hasa computer at home, it is rarely used by

variables. It can be summarised asfollows: the innovators for digitaltelevision in Flanders are young(mid-30s), well-educated people, nothaving much leisure time (a lot ofself-employed people and executives).They describe themselves as activePC-users, and ‘heavy readers’ ofnewspapers and magazines. In theirleisure time they like to go out, andshow who they are by their choice ofmusic, clothing and the following oftrends.

Within this category there is also aremarkable share of students, singles andlower incomes (<900e). The latter is notthat surprising if account is taken of thephenomenon of ‘credit-buyers’. Thisshare of lower incomes within theinnovators was also seen for such thingsas wide-screen television, WAP andgames computers. They may be buyingon credit, but by being among the firstto have these technologies they create asort of identity.

The monthly amount these innovatorsare willing to pay for dTV is, onaverage, about 30e. This is an averagefor the whole segment of innovators. Ifstudents (no income) are not includedthis ‘willing to pay’-average increases to40e a month. If only the self-employedare considered an average of 50e amonth is reached.

Besides this, the segment of innovatorsdeviate slightly from the 60/40male/female distribution. Within theseinnovators there are proportionally moremen (70/30). Concerning media usageinnovators can be considered as asegment of ‘heavy surfers’ and ‘heavyviewers’ — they surf on average 50minutes a day and watch television fortwo hours a day.

Although the Flemish early adoptershave a similar profile to the innovators,the price they are willing to pay issignificantly less: about 22.5e a month.



— more channels;— more specific/thematic channels;— subtitling in different languages;— additional background information;— choosing films/programmes according

to taste/choice;— ability to start/stop/pause a

programme;— epg;— home shopping;— home banking;— downloads for delayed watch;— proton (loading pay card for small

expenses);— interactive courses;— browsing the net on the television;— send and receive e-mail;— sms function.

Looking at Figure 5, it is clear there aresignificant differences between thedifferent segments. The five ‘interestlines’ of the innovation segments runmore or less parallel, and in a logical

these respondents (decision makers fornew communication technologies). Thesecomputers are probably used by thechildren.

INTEREST IN APPLICATIONSIn order not to restrict the adoption tothat small segment of innovators, and toget digital television adopted by thebiggest possible share of the market, it iscrucial to have a thorough insight in theinterest these innovation segments havein several dTV applications. Also, in thesearch for so-called ‘killer applications’, itwould be useful to know whichapplications the global market is more orless interested in, and in which theinterest is more segment concentrated.The 17 dTV applications that weretested in this study were:

— better image quality;— better sound quality;



Figure 5: Interest in dTV applications (segments based on PSI scale); 5 � very interesting

1

1,5

2

2,5

3

3,5

4

4,5

5

better image

better sound

morechannels

spec/them

atic

subtitling

background

owntaste

start/stop

epg

home-shop

home-bank

download

proton

course

browsing

e-mailsms

innovators

early adopters

early majority

late majority

laggards

efficient marketing strategy is still opento debate. In the authors’ opinion,Geoffrey Moore’s ‘bowling pin model’reasoning is the best way to do this. Thismeans that a plan of communicationscould be as presented in Figure 7.

For example, a basic package could beoffered for a price of 7.5e up to 10e amonth. This package could containapplications such as epg, the possibility ofstarting and stopping programmes andasking for additional backgroundinformation with news, journal andcurrent affairs programmes. Besides thecontent focus on news and currentaffairs, the documentary genre could alsobe included in the basic package. Theseare the things that appeal to almost theentire market32 and deserve emphasis inthe initial mass media campaign. Also,the emphasis on ‘more channels’ cannotbe excluded from this campaign. Thisinitial mass media campaign is illustratedby the long dotted line at the right

sequence. (‘Logical’ meaning that interestin the applications decreases towards theend of the adoption curve. Innovatorsare for almost all applications the mostinterested; early adopters are generallyless, but still quite, interested, andlaggards are the least interested.Kruskal-Wallis and Anova tests yieldsignificant differences between the fivesegments. Except for subtitling(significant at the 0.05 level), alldifferences are significant at the 0.01level.

Further statistical analysis shows thatthe PSI approach is much more sensitivethan the DSI approach.31 A comparisonof Figures 5 and 6 reveals this.

BOWLING PIN DSS MODELFOR DIGITAL TELEVISIONIN FLANDERSThe best way to implement all thisinformation in order to obtain the most



Figure 6: Interest in dTV applications (segments based on DSI scale); 5 � very interesting

1

1,5

2

2,5

3

3,5

4

4,5

5

better image

better sound

morechannels

spec/them

atic

subtitling

background

owntaste

start/stop

epg

home-shop

home-bank

download

proton

course

browsing

e-mail

sms

innovators

early adopters

early majority

late majority

laggards

watching TV’ will be accompanied fromnow on with a certain degree ofinteractivity and a certain degree ofcontrolling what is on the screen. A lotof people would immediately give up ifthey were to be offered ‘the totalpackage’ at once (at the full price, tooexpensive for a lot of people, and withall of its applications, a lot of them notinteresting enough for a lot of people),but they now have the time to learn toappreciate the concept. By having thistime, there is also more time for themarketers to prepare and to prepare themarket for more interactivity and moresophisticated applications. Applicationssuch as sms, sending and receivinge-mails, VOD will be better offered laterfor this reason. Once sales of the basic

pointing to the right upper corner, andgoing from S1 to S5. By choosing thisbasic package at this stage not muchwrong can be done. Practically thewhole market is more or less interestedin it, and the price is not higher thanthe price the least interested segment(laggards) is willing to pay for it. Also, toget everybody acquainted with thisunknown product, the mass media arethe best channel to choose in this initialstage.

By starting with an offer of such abasic package (which the greater part ofthe market is interested in and willing topay for), the market is also given sometime to grow familiar with the digitaltelevision concept: time to get used tothe fact that the ‘traditional event of



Figure 7: BPM DSS for digital television in Flanders

S2/A1

Mass mediaSpecif. Media

S1/A1

S2/A1

S3/A1

S4/A1

S5/A1

S1/A2

S2/A2

S3/A2

S4/A2

S1/A3

S2/A3

S3/A3

S1/A4

S2/A4S1/A5

S1: Innovators (7.5e)A1: Basic package: more channels,

epg, start/stop, news + background

information, documentaries

S2: Early adopters

S3: Early majority

S4: Late Majority

S5: Laggards

S2: + sms, film

7.5e

S3: + HI, proton, download

± 17.5e

S4: + courses,

brows., e-mail,

For.soap

± 25e

S5: + home-

bank

>30e

7.5e>40e

changing the emphasis of the mass mediacampaign (instead of promoting only thebasic package, the emphasis also toucheson sms and VOD). The same goes forA3, also an application still appealing toa large part of the market (64.9 percent): once the specific segment-targetedcommunication has done its work andthe innovators have adopted thisapplication, this application can also beincluded in the mass media campaign.By offering these applications a bit later(not immediately on introduction), andby working initially with targetedcommunication towards S1 (and S2), andonly then communicating and promotingthrough mass media channels towards thewhole market, the necessary copyingbehaviour is given more time to develop.This also reduces the odds onoverwhelming and frightening the‘initially less interested segments’. It evenaccounts for the fact that somehabituation to a new concept is needed,if people are to take up the product.

Therefore, the basic package ispromoted first through the mass media,and additional (and more sophisticated)applications are promoted later. Eachadditional offer will imply a gradualincrease of the subscription rate. Thisadditional offer will initially be promotedby targeted (segment-tailored)communication, and only afterwardsthrough mass media campaigns. Thiscommunication strategy of ‘first targeted,delayed mass media’ is only useful inpromoting additional applications thathave a relatively big potential. This ‘firsttargeted, second mass media’ strategy isillustrated in Figure 7 by the areabetween the three dotted arrows pointingto the upper right corner. To the right ofthis area communications can be donethrough the mass media. To the left ofthis area mass media will no longer beuseful because the interest is no longerthat big. S4 and S5 only appeal to the

package begin to roll — or once thepins on the right side of the bowlinggame begin to fall (copy behaviour33) —more sophisticated applications (A2, A3,A4 . . .) can be offered.

By using targeted communication (fullline pointed towards the left uppercorner) towards these innovators andearly adopters (their specific media usageis known and they can be reached bydTV because they have already adoptedit for the basic package A1), thesesegments can be prepared for additionaldTV applications (with higher monthlyfees). In the first expansion of theproduct, people can be offered an smsapplication and a film-channel/VODapplication (Application 2). In the nextstage Application 3 can be offered: thereloading possibility for pay card anddownload applications, etc. ForApplication 2 and at the price forA1�A2 (about 17.5e a month) there isstill big potential in the market: it isassumed that only the laggards (16.9 percent) will no longer be interested whichstill leaves 93.1 per cent of the market aspotential customers for A2. ForApplication 3, there will still be broadinterest (and willingness to pay):innovators � early adopters and earlymajority (together 64.9 per cent of themarket). Once the innovators and earlyadopters have become aware of, andfamiliar with, the new additionalapplications through specific targetedcommunication towards them, there is agood chance these very innovativesegments will adopt the product. Oncethey have adopted it, the emphasis in themass media campaign (dotted linespointed to the upper right corner) canchange. Once the innovators (S1) haveadopted the second application A2, theywill be copied by the rest of the market.It is known that the potential forcopying the use of this application is big(93.1 per cent), thus there is little risk in



added value was recognised by marketingcommunication professionals. The nextstage of development includes multiplesuboptimal propositions and thedevelopment of a statistical model thatcomputes and describes the innovationadoption segmentation whereinnovativeness is modelled as a latentvariable with declining intensity.

References1 Carter, J. (1998) ‘Why settle for ‘‘early adopters’’?’,

Admap, Vol. 33, No. 3, pp. 41–44.2 Here ‘innovation forecasting’ means the prediction

(in advance) of innovation categories for a specificnew product, the detection of those segments andthe making of a profile of those segments, beforethe product is launched.

3 At the time of writing (February 2004) digitaltelevision still has to be introduced in Belgium.Within the near future, several players have plans todo so, which makes dTV an ideal test case.

4 Thomson, S. (2001a) ‘Free interaction’, Cable &Satellite Europe, June, p. 2.

5 Thomson, S. (2001b) ‘Lifting the gloom’, Cable &Satellite Europe, July/August, p. 2.

6 Giesen, P. (2004) ‘De val van digitale televisie’, DeMorgen, 7th February, p. 22.

7 Sutherland, F. (1999) ‘Remote overload: Too muchtoo soon?’, Cable & Satellite Europe, July, pp. 46–49.

8 Carter (1998) op. cit.9 Rogers, E. M. (1995) ‘Diffusion of innovations’, 4th

edn, The Free Press, New York, p. 262.10 Moore, G. A. (1999) ‘Inside the tornado’,

HarperCollins, New York, p. 13.11 Moore, G. A. (1994) ‘Crossing the chasm’,

HarperCollins, New York.12 Ibid.13 Flaherty, N. (2001) ‘TV gets personal’, Cable &

Satellite Europe, June, pp. 10–16.14 Professor Parasuraman’s ‘Technology Readiness

Group’ (2001) also emphasises the importance of the‘comfort level’ factor when it comes to theacceptation of innovations. They describe it as ‘afeeling of control (or lack thereof) over technology’.It is a kind of ‘uncertainty factor’ coveringconsumer feelings as ‘the technology is toocomplex’, ‘a lack of trust in tech support’, ‘gettingoverwhelmed by the technology’.

15 Watts, R. J. and Porter, A. L. (1997) ‘Innovationforecasting’, Technological Forecasting and Social Change,Vol. 56, pp. 25–47.

16 Leavitt, C. and Walton, J. R. (1988) ‘Openness ofinformation processing as a moderator of messageeffects on behavior’, faculty working paper, Collegeof Business Administration, Ohio State University.

17 Price, L. L. and Ridgeway, N. M. (1983)‘Development of a scale to measure use

innovators and early adopters: a marketshare that is too small to approachthrough the mass media. So, to the leftof this area only specific media(specifically targeted) are used. Withinthis area a combination of specific andmass media is used. So, as soon as theinterest in the additional applications islimited to the innovators and earlyadopters (for A4 and A5 this is the case),only a specifically targeted strategy ofcommunications is used. A4 and A5 arealso the applications that will need themost persuasive power to yield successfuladoption. Vague mass media messageswill no longer be sufficient here, whichmakes a strategy of specificallysegment-tailored communication,34 thebest. A strategy in which the marketercan go deeper into things and throughwhich the applications and their use canbe situated in line with the lifestyle ofthose segments.

In this way, the whole market can bereached as much as with the totalproduct (with all its applications) — withthe difference that not all applicationswere offered from the start to everybody,only to those who were interested andwilling to pay for them. This results indifferent packages and different forms ofsubscriptions to digital television in aprice range of 7.5e up to more than 40e

a month.

CONCLUSION ANDFUTURE RESEARCHThe combination of the new Delphimethod-based PSI scale with the bowlingpin model within the framework of asoft decision system allows for a bettermarketing and communication strategybefore the launch of new technology.This methodology has been applied inseveral commercial studies (eg universalmobile telecommunications system(UMTS) and broadband) where the



26 Rivett, P. (1994) ‘The craft of decision modelling’,John Wiley & Sons, New York.

27 Roy, B. (1999) ‘Decision aiding today: What shouldwe expect’, in Gal, T., Stewart, T. J. and Hanne, T.(eds) ‘Multiple criteria decision making, advances inMCDM models, algorithms, theory andapplications’, Kluwer’s International Series, Boston.

28 Carter (1998) op. cit.29 Watkins, T. (1985) ‘The economics of the brand’,

McGraw-Hill, London.30 With ‘earlier adopters’ they meant the collection of

innovators and early adopters. Because the segmentof innovators is often too small for effectivetargeting and marketing, they thought it was betterin some cases to consider those two segments as onegroup.

31 De Marez, L. and Verleye, G. (2002) ‘Diffusion ofinnovations’, extended working paper, University ofGhent, Dept. of Communication Sciences.

32 Not only the innovators are interested in it, but alsothe rest of the market, by which the package has abigger chance of being adopted by a larger share ofthe market.

33 There is a good chance this copy behaviour willtake place, because it is a package in which they areinterested, they have to copy.

34 This strategy of only using specific targeting doesnot only have to be used to promote A4 and A5towards innovators and early adopters. It can alsobe used to promote other applications with aspecific segment-related interest: for example topromote home shopping to those young coupleswithin the early adopters, or to promote protonto students.

innovativeness’, Advances in Consumer Research, Vol.10, pp. 679–684.

18 Bearden, W. O., Netemeyer, R. G. and Mobley, M.F. (1993) ‘Handbook of marketing scales. Multi-itemmeasures for marketing and consumer behaviourresearch’, Sage Publications, New York, p. 59.

19 Flynn, L. R. and Goldsmith, R. E. (1993) ‘Avalidation of the Goldsmith and Hofackerinnovativeness scale’, Educational & PsychologicalMeasurement, Vol. 53, No. 4.

20 An optimal dTV package is discussed with theinterviewer, for example, what price, whichapplications.

21 Wright, D. (1998) ‘Analysis of the market for accessto broadband telecommunications in the year 2000’,Computer Operations and Research, Vol. 25, No. 2, pp.127–138.

22 Ward, P. and Davis, B. J. (1999) ‘The diffusion ofinteractive technology at the customer interface’,International Journal of Technology Management, Vol. 17,Nos. 1 and 2.

23 McBurney, P., Parsons, S. and Green, J. (2002)‘Forecasting market demand for newtelecommunications services: An introduction’,Telematics and Informatics, Vol. 19, pp. 225–249.

24 Goldsmith, R. E. and Hofacker, C. (1991)‘Measuring consumer innovativeness’, Journal of theAcademy of Marketing Science, Vol. 19, Summer, pp.209–221.

25 De Marez, L. and Verleye, G. (2004) ‘ICTinnovations today: making traditional diffusionpatterns obsolete, and preliminary insight ofincreased importance’, Telematics & Informatics,forthcoming.



diffusion of innovations: successful adoption needs more ... · down to badly judged marketing...

Documents