A cost-effectiveness analysis of online, radio and print tobacco control advertisements targeting 25–39 year-old males
The authors have stated they have no conflict of interest
Objective: To assess the relative cost-effectiveness of various non-television advertising media in encouraging 25–39 year-old male smokers to respond to a cessation-related call to action. Information about how new electronic media compare in effectiveness is important to inform the implementation of future tobacco control media campaigns.
Methods: Two testimonial advertisements featuring members of the target group were developed for radio, press and online media. Multiple waves of media activity were scheduled over a period of seven weeks, including an initial integrated period that included all three media and subsequent single media phases that were interspersed with a week of no media activity. The resulting Quit website hits, Quitline telephone calls, and registrations to online and telephone counselling services were compared to advertising costs to determine the relative cost-effectiveness of each media in isolation and the integrated approach.
Results: The online-only campaign phase was substantially more cost-effective than the other phases, including the integrated approach.
Conclusions: This finding is contrary to the current assumption that the use of a consistent message across multiple media simultaneously is the most cost-effective way of reaching and affecting target audiences.
Implications: Online advertising may be a highly cost-effective channel for low-budget tobacco control media campaigns.
The progressive decrease in smoking rates in developed nations has been attributed in part to mass media campaigns that have informed the public of the dangers of tobacco use.1–3 In Australia, for example, a sustained program of public education since the 1970s has seen a reduction in the number of regular smokers from 36% in 1977 to 16% in 2010.4–6 While this trend is encouraging, tobacco use remains high among specific population subgroups. For example, 27% of males aged 25–29 years and 24% of males aged 30–39 years are regular smokers.4,6 This is comparable to Australian smoking rates from 20 years ago,4 indicating that current tobacco control measures are not reaching these smokers. Interventions that cater to the specific needs of this population subgroup are required to address the disparity in usage rates.
The long-held assumption of positive synergistic effects between advertising media continues to dominate the communications literature.7,8 According to this assumption, using an integrated approach involving the combination of multiple media to deliver a message produces greater effects than relying on one medium alone.9 When using an integrated media approach, it is typically recommended that a consistent message is delivered across the selected media to enhance audience comprehension and recall of the message.8
Despite the general belief in the superiority of an integrated communications approach, much of the research examining the effectiveness of messages disseminated via the mass media has focused on television.7 There is a need for more research involving other media, both individually and in combination, to provide data on their relative effectiveness. This is especially the case in the context of health promotion messages, where the higher costs of television advertising relative to advertising in other media can make it difficult for campaigns to be adequately resourced to meet their communication objectives.10,11 Evidence pertaining to the ability of more affordable media to reach and influence key target groups is needed to maximise the cost-effectiveness of future campaigns.
Some work has been done in the tobacco control field to address this need. To date, this work has produced inconsistent outcomes in terms of the relative cost-effectiveness of different media. For example, Durkin and Wakefield12 found that radio advertising outperformed television advertising on campaign awareness and intention to quit outcomes, while Mosbaek et al.13 found that television advertisements were more effective than radio advertisements in stimulating calls to a Quitline. Farrelly et al.11 examined the effects of expenditure on television, radio, and print advertising and concluded that while television advertising produced the greatest overall increase in calls to a Quitline, incremental increases in expenditure on radio advertising yielded proportionally higher increases in the call rate.
These differences in reported cost-effectiveness may be partly due to the different dependent variables tested (e.g. calls to a Quitline versus intention to quit), differing message types (e.g. emotional testimonials versus rational information provision), disparate levels of resourcing between campaigns, varying methodological approaches and different national contexts. In addition, direct comparisons are confounded by the inevitable differences between advertisements run in different media.13 For instance, radio is limited to sound while traditional print media are confined to static pictures. Further, it has proven difficult to isolate the effects of individual media due to the tendency for campaigns to typically involve the simultaneous use of different media to optimise the overall results.10,11 In such circumstances, it is difficult to attribute results to specific media.
Within the integrated communications context, the advent of new media has resulted in interest in the potential for offline and online media to complement and reinforce each other.7 Tobacco control media campaigns now routinely include an online component,8,14 but little is known about the most effective methods of information presentation and response elicitation and how best to combine online and offline advertising to optimise results.15,16 Some evidence suggests that online advertisements may be more effective than traditional media in encouraging smokers to participate in cessation programs, especially in the case of young male adults.16 This is consistent with analyses of internet use behaviours that have found males especially value the ability to quickly navigate to information of interest.17
The aim of the present study was to assess the relative cost-effectiveness of non-television media in encouraging 25–39 year-old male smokers to respond to a call to action relating to smoking cessation. Of interest was whether one medium was more effective than others, and whether a combination of media provided synergistic effects over and above the use of individual media. Testimonial-style advertisements broadcast on radio, printed in newspapers and featured on websites were compared for their ability to generate calls to a Quitline, visits to a Quit website and registrations to smoking cessation services. Production and media expenditures were calculated across the different channels to facilitate cost-effectiveness calculations. Of note is that there was no intention to assess the cost-effectiveness of different interventions on actual quit rates, which has been done elsewhere.18,19 The purpose of the study was to provide information relating to the cost-effectiveness of various non-television media in encouraging a specific population segment to respond to a cessation-related call to action.
The study was approved by the Curtin University Human Research Ethics Committee.
Given that testimonial advertisements have been found to be successful in previous smoking cessation campaigns20 and that the target group was young men, the creative consisted of two male personal testimonies created specifically for this campaign. To maximise consistency, the same two testimonials were used throughout the different media channels (accessible at www.cancerwa.asn.au/MSH-personal-testimony-campaign). Each advertisement promoted both the Quitline telephone number and a Quit website specifically created for the campaign. Those calling the Quitline were offered the opportunity to register for ongoing telephone counselling. Those visiting the website could elect to register with the QuitCoach service, a web-based cessation tool that provides smokers with information and a personalised quit plan.
Over a period of seven weeks, a staged approach was used to stagger the timing of the advertisements across the three media (radio, press, and online) in Western Australia. The time period, 9 July to 26 August 2012, was selected to avoid contamination by national quit campaigns that run intermittently throughout the year. In the first week of the study, an integrated approach was used where advertisements were run in all three media. The integrated approach was used first to minimise the order effects that would have been likely to influence the proportions of responses across the three individual media should one of them have been initially used in isolation. Historical data showed that Quitline calls and website hits resulting from previous campaigns were short-lived, returning to baseline levels within a week of exposure. A week was therefore left between the conclusion of the integrated phase and the commencement of the first round of advertising via individual media. A random allocation was made to determine the order in which the other media would be utilised, resulting in the following order: radio, online, press. In subsequent weeks, one medium at a time was used, with a week break in between each.
The radio, online, and press advertisements were run in the most popular media platforms for 25–39 year-old Western Australian male smokers. Suitable media platforms were established through the use of reputable media research tools including Roy Morgan Single Source Data and Nielsen Media Research that identify the population segments that are reached predominantly through various types of communications channels. On the basis of these data, four commercial radio stations that are popular among this group were selected to broadcast the radio advertisements. Similarly, the selected websites included two news websites, two combined news and entertainment websites (portals), one internet dating site and Facebook. There was less scope to segment by audience characteristics in the print media, resulting in both major state newspapers and 24 community newspapers being selected to publish the press advertisements to ensure adequate reach among the target audience.
Table 1 shows the distribution of campaign costs. There was some variation in both production and media costs for each medium, reflecting different production requirements, the costs of purchasing advertising space and media reach. The radio advertisements were the least expensive to produce, followed by the press ads, with the online ads being the most expensive. The production costs for each medium were halved, with 50% attributed to the single media phase and 50% to the integrated phase. The relative costs of purchasing advertising space in each media constrained cost distribution due to the much higher per unit costs associated with press outlets relative to radio and online. The greater amount spent on press media in the integrated phase relative to the other two media reflects the high minimum cost associated with buying advertising space in the newspapers that would provide adequate geographical coverage.
Table 1. Counts by type of media activity.
|Baseline|| || || ||187||85||14||16||70||0||0|
|Break|| || || ||79||71||13||18||77||1||1|
|Break|| || || ||53||49||15||31||79||1||1|
|Break|| || || ||68||63||16||25||78||0||0|
In terms of potential reach, Roy Morgan data21 indicate that among Western Australian smokers aged 25 to 39 years, around two-thirds were likely to have accessed one of the selected online sites in the previous month, one-half were likely to have read one of the selected newspapers and one-third were likely to have listened to one of the selected radio stations. To partially compensate for these variations, radio received the greatest amount of media expenditure in the individual phases, followed by press and then online. As noted above, it was not possible to replicate this strategy in the integrated phase because of the high minimum spend for press, but the online and radio expenditures were adjusted to accommodate the differing levels of likely reach. The amounts expended in each of the individual phases also reflect the ‘spend’ required to achieve critical mass for a campaign being run in a single media, within the limits of the campaign budget.
As those being exposed to any of the advertisements were provided with both the Quitline phone number and the Quit website address, their possible responses to the calls to action included: (i) calling the Quitline; (ii) accessing the specific web address provided; and/or (iii) conducting a web search to locate the website. As noted above, those who telephoned the Quitline could access immediate counselling from a trained operator and/or register for an ongoing counselling service, while those visiting the website could interrogate the site for information and/or register for the QuitCoach service. It is possible that visitors to the website obtained the Quitline telephone number and called this service rather than progressing to the QuitCoach service, and vice versa.
Continuous monitoring of activity on the Quit website and telephone calls to the Quitline was conducted during the week prior to the campaign period and the following seven weeks that comprised alternating weeks of media activity and no activity. To estimate the effectiveness of the integrated phase and each of the single media phases individually, the calls to the Quitline, visits to the website, and registrations for the QuitCoach and call-back counselling services that could be attributed to the campaign were estimated. This involved calculating the number of events in the week of each phase minus the number expected in that period (i.e. the number of events in the baseline week before the commencement of media activity). If this estimate was negative, it was assumed that the phase had no effect.
These calculations of the effectiveness of each of the campaign phases could lead to different ordering of effectiveness for different outcomes. For example, the online advertisements might be more effective for generating web visits, while the radio advertisements may be more effective for generating calls to the Quitline. Two measures for the overall effectiveness of each phase were therefore also constructed, as recommended by the World Health Organization.22 For the first measure, all the individual events (i.e. calls to the Quitline, visits to the website and registrations for the QuitCoach and call-back counselling services) were summed. To correct for the fact that some events were more likely to result in quitting smoking than others, it was assumed that the QuitCoach and call-back counselling service registrations were equally effective interventions, and that website visits and calls to the Quitline constitute a first step towards registration. The probability of a visit/call leading to registration (i.e. the number of registrations divided by the number of visits/calls) was therefore used as a weight for these events in the second measure.
To determine the cost-effectiveness of each media, the following were calculated for each phase of the campaign:
- costs per additional call to the Quitline
- costs per additional visit to the website
- costs per additional registrations for QuitCoach
- costs per additional registration for the call-back counselling service
- costs per unweighted total of additional events
- costs per weighted total of additional events.
The costs per weighted total of additional events were used to determine the incremental cost-effectiveness of each media type. Those tested media that were more expensive but less effective than others were ruled out because they were dominated by the campaigns that were more effective and less expensive. For the remaining media, the relative performance was measured as the additional cost of a specific strategy divided by its additional overall effectiveness, compared with the next least expensive strategy.
To determine the robustness of the results given the weighting assumptions, the outcomes were also evaluated using the following weights: (i) half the original weight for visits to the website and QuitCoach registrations; (ii) half the original weight for Quitline calls and registrations for the call-back counselling service; (iii) half the original weights for visits to the website and Quitline calls; and (iv) double the original weights for visits to the website and Quitline calls. In addition, the results for total visits were compared to those for unique visits within each week of media activity to assess sensitivity according to type of visit as an outcome measure.
The costs, website visits, Quitline calls, QuitCoach registrations and call-back service registrations for each phase of the campaign, including the initial baseline week, are reported in Table 1. Of note is that the ‘no activity’ weeks exhibited response rates similar to or lower than those recorded at baseline. This demonstrates that the campaign effects receded rapidly once the advertisements were withdrawn and that the media activity weeks ‘soaked up’ smokers who may otherwise have sought assistance at a later point in time.
The online phase was most effective in generating extra visits to the website, Quitline calls and registrations to QuitCoach; while the radio phase was most effective in generating extra registrations to the call-back counselling service (Table 2). Overall, the online phase generated the most events, both unweighted as well as weighted by the presumed success rate of each of the events. In the weighted total, website visits were weighted by 176/2,589, reflecting the total number of QuitCoach registrations from the total number of website visits (Table 1); and Quitline calls by 29/686.
Table 2. Effectiveness by media.
|Radio||No effect||No effect||11||13||24||13|
Because the online phase was both the most effective and the least expensive, it was also the most cost-effective, dominating all other media (Table 3). The costs per unweighted event and weighted event were three to five times lower than those of the other phases. The cost-effectiveness of the online phase was robust for the different weighting assumptions and when using unique visits to the websites instead of total visits. In all cases, the online phase remained both the most effective and least expensive (data not shown).
Table 3. Cost-effectiveness by media.
|Radio||No effect||No effect||5,223||4,420||2,394||4,267|
| || || || || || || |
The present study assessed the relative cost-effectiveness of various non-television media in producing a response to smoking cessation advertisements. It was not designed to test the efficacy of a Quit intervention in terms of actual cessation rates, but rather to provide information to assist those seeking to optimise media allocation with limited advertising budgets.
Reflecting the modest size of the city of Perth (population of around 1.8 million) and a smoking rate that is lower than the national average (13%),23,24 the absolute numbers used in the analysis are not large. However, the pattern of results across the various media provides a clear indication of relative cost-effectiveness of various non-television media for a tobacco control campaign targeting males aged 25–39 years. Specifically, online advertising was identified as being substantially more cost-effective in generating responses relative to advertising in other non-television media. The responses monitored were total website visits, calls to a telephone quitline and registrations to online and telephone counselling services. Radio and print advertisements were much less cost-effective, and even the integrated approach that used all three media in combination produced a considerably lower rate of response per dollar expended.
These results are counter-intuitive given the focus in the communications literature on the benefits of integrated approaches to message dissemination and the reported lack of evidence of negative synergies of multiple media usage.7,9,25 In the present study, spending $47,557 on online advertising alone substantially outperformed an investment of $131,296 across all three media. It appears that the shifting media landscape may require a change in assumptions relating to integrated marketing communications. Particularly for target audiences who are heavy users of online media, placing advertisements in the online environments they visit frequently may constitute an especially cost-effective strategy. The superior performance of online advertising is likely to be the result of the greater reach of this medium and the ability for smokers to immediately respond to the advertisement by clicking through to the Quit website, thereby overcoming the inertia problems associated with other media that require message recipients to be adequately motivated to remember to act at a later point in time (such as when hearing a radio advertisement while driving).
Of note is the extent to which responses receded in the no media activity weeks. These results support previous research that has found tobacco control campaigns to be effective while they run, but for lag effects to be minimal.26 This reinforces the need for investment in sustained campaigns that continue to resonate with smokers. Such campaigns have the potential to be very expensive, hence the need to identify the most cost-effective means of communicating with target audiences.
There are several notable limitations of the present study that may be addressed in future research. In the first instance, due to the logistical difficulties of tracking website hits, there was no assessment of the demographic profiles of those responding to the advertisements. It is therefore not possible to determine whether it was members of the target group (males aged 25–39 years) who responded the most. However, the selection of specific media outlets that have relevant audience demographics would have ensured that substantial numbers of the target group were exposed to the advertisements. Second, there would have been some degree of contamination across the media bursts, although this would likely have occurred in two opposing directions, potentially cancelling each other out. For example, the press advertisements in the last media phase would have benefited from the accumulative effects of ongoing dissemination of the messages through multiple media over the campaign period. However, they would also have been disadvantaged by the previous phases picking up many of those in the contemplation and preparation stages of behaviour change (as per the Transtheoretical Model of the stages of change).27 Such effects are inevitable where different media are used sequentially to assess relative cost-effectiveness, but this is likely to be preferable to the more complex problems associated with disentangling effects resulting from simultaneous exposure to multiple media. Third, the use of a ‘natural experiment’ precluded identical apportionment of costs across the media within the integrated phase and across the individual phases. This is an unavoidable limitation given the varying cost structures associated with different media and the limited budget available to fund the campaign.
A further limitation is the exclusion of television advertising from the communications mix. Given the results of previous research,11,13 it is likely that television advertising remains one of the most cost-effective means of reaching smokers, including those in lower income groups. However, more than half (55%) of the lowest quintile households in Australia and two-thirds (66%) of the second-lowest quintile households now have Internet access,28 indicating that Internet advertising will become increasingly effective with this group over time. Future research could include a television advertising phase to provide further insight into the relative cost-effectiveness of online media relative to traditional media, especially among more disadvantaged groups. Finally, any changes in awareness of and attitudes to the advertisements among both the target audience and secondary audiences were not measured. Such outcomes are important elements of any media campaign because of their role as antecedents to ultimate behaviour change.29 Future research could include these variables to obtain a more comprehensive understanding of campaign outcomes and to facilitate inclusion of their effects in cost-effectiveness analyses.
To conclude, this study demonstrates the potential for online Quit advertisements to out-perform advertisements in other non-television media in encouraging a response from a specific target segment. This information is likely to be useful to those designing low-budget tobacco control media campaigns. In particular, the results may assist those attempting to achieve the best value allocation of limited funds (or ‘best bang for the buck’) to maximise tobacco control outcomes.
The authors acknowledge Tony Scampoli, Emma Lambert, Kelly Dienaar, Adam Barker, Lorena Chapman, Seng Sengsourinho and the staff at Quitline and QuitCoach for their contributions.