Evaluating Travel Advertising Impact

EVALUATING TRAVEL ADVERTISING IMPACT

Ilkka A. Ronkainen, Georgetown University

Arch G. Woodside, University of South Carolina

ABSTRACT -

Tourism is the number one retail business in the United States and has become increasingly competitive to attract the approximately 30 million frequent travelers. Attractions are using various methods to get their share of traveling population.

The development of techniques measuring the effectiveness of such campaigns is analyzed in this paper. Conversion studies in travel and tourism are reviewed in terms of applications and issues critical to achieving reliable results over a ten-year period, with emphasis on developmental improvements.

Methodological topics include sampling, response rates, measuring expenditures, and sponsor identification.

Applications beyond estimating rates of conversion of inquirers into visitors include comparisons of the relative performance of ads, advertising campaigns, as well as media vehicles.

INTRODUCTION

With more states and specific locations awakening to tourism's potential as an economic accelerator, the advertising budgets to attract tourists have increased . During the 1982/83 fiscal year, the average state ad budget was $937,000, with New York in the forefront with $6,800,000 earmarked for advertising production and media costs. A number of other states, such as Alaska, Florida, North Carolina and Hawaii, are spending well in excess of $1,000,000 (U.S. Travel Data Center 1983; Advertising Age 1982).

These figures are not surprising when per person per visit expenditures are considered. Alaska ranks first with $1,667 per person per visit (1977 data), followed by Hawaii with $524 (1977 data) and Florida with $474 (1979 data) (Goeldner and Dicke 1981).

As both the costs and the benefits of these programs have increased, more research has been directed at the effectiveness of the advertising campaigns conducted. Most state travel offices conduct promotional campaigns which include coupons, addresses, and/or toll-free telephone numbers through which those interested can obtain brochures and travel information on the particular state's attractions and features.

The effectiveness of such campaigns is measured by using advertising conversion research. This research is conducted to answer the question of how many inquirers from travel ads convert to visitors and what their (travel) behavioral characteristics ( e.g . , expenditures) are. The information generated from returned questionnaires (or responses obtained by phone) permits the computation of total revenue, average revenue per inquiry (RPI) , and average cost per inquiry (CPI). Since this technique was presented by Woodside and Reid (1974), a number of conversion studies have been completed. Conversion studies have been used to compare the relative performances of one ad versus another ad, and one ad campaign versus another ad campaign (Woodside and Motes 1980; 1981), as well as one media vehicle versus another media vehicle (Woodside and Ronkainen 1982). Furthermore, several studies have addressed methodological issues in conversion research (Woodside 1981; Ellerbrock 1981; Ballman et. al. 1983; Woodside and Ronkainen 1983; 1984).

The purpose of this paper is to discuss the major issues pertaining to conversion research as well as to address some of the problems inherent to the approach. Data from conversion studies are used to highlight the major points raised.

THE RESEARCH PROCESS

Data Collection Modes

Three primary modes are available: personal interviews, telephone interviews, and mail interviews. Personal interviews have not been reported as a primary mode mainly because of its cost and staff requirements. Typically, a questionnaire is mailed to (or phone call made to) a sample of inquirers six to ten months after receiving the inquiries.

In comparing and contrasting the telephone and mail methods, the advantages of one are often the disadvantages of the other. Wh i.1 e a general comparison can be made, a major caveat should be emphasized. The choice of a mode is situation specific; i.e., factors such as time constraints, budget, sample size, and information requirements determine a particular mode's advantages and disadvantages in given situation (Peterson 1982; Churchill 1983).

The per-contact cost of the mail questionnaire is generally low; e.g., Hunt and Dalton (1983) found that the cost per usable questionnaire was estimated to run approximately 40% more for the telephone method ($1.33 versus $2.05). However, if nonresponse is substantial, the cost per return can end up being high.

The introduction of computer-controlled cathoderay terminals (CRT) has had a significant impact on the conduct of telephone interviews. In addition to assisting in sequencing of questions, results are available at a moment's notice since all replies are stored in memory: Depending on the number of follow-up mailings required, the total time needed to conduct a good mail survey can be substantial. However, as the number of required contacts increases, the attractiveness of the telephone mode decreases (with additional staff requirements and escalating costs).

Although the mail- questionnaire mode allows little speed control, it represents a standardized stimulus. With telephone interviews, interviewer- induced variance can be reduced through proper selection, training and supervision during the data collection.

The types of questions that can be asked as well as the amount and the accuracy of the information needed will influence the choice of the mode. Telephone interviews, although they usually do not allow the same quantity of data collection as mail, will enable flexibility in terms of probing respondents' comments further as well as the use of open-ended questions to a greater extent. since respondents to a mail questionnaire are able to work at their own pace, a better thought-out response may result. The appropriateness of these responses, however, is a function of the adequacy of the questionnaire.

Mail questionnaires permit data collection from a wide geographic dispersion of individuals since the mode is relatively time and cost insensitive. With sample sizes of 5,000-10,000 (e.g., Woodside and Reid 1974; Woodside and Motes 1980), mail interviews become the only feasible mode of data collection.

Sampling

Since the number of inquirers in usually quite high, e.g. , 73,831 inquirers from black-and-white ads placed in magazines as reported by Woodside and Ronkainen (1982), sampling is one of the major problem areas to be encountered. Random sampling in some form is the generally accepted method. Ballman et. al. (1983) provide a through discussion on incorporating sampling precision to improve the quality of management decisions based on conversion studies.

Response Rate

Typically, the response rates produced from an initial mailing range from 20 to 40%. If a second mailing to nonrespondents to the first mailing is used, the response rates usually range between 30 and 50%; response rates after a third mailing have been reported at 46 to 80% . Only one study reported by Wettstein ( 1982) has a single mailing. The results of this study after a 40% response rate should be treated with suspicion, especially when no further sampling of nonrespondents was reported.

Ellerbrock (1981) points out that people who actually visit are more likely to respond to a survey due to the feeling that their answers are more valuable to the state than those who did not visit. It is, therefore, clear that failure to correct for non-response bias leads to inflated conversion estimates. Hunt and Dalton (1983) calculated that not correcting for non-response bias would have resulted in a 44%overstatement of conversion. Woodsideand Ronkainen (1983) found that conversion rates of inquirers were substantially lower for second versus first mailing respondents, while rates were similar for third versus second mailing respondents. While demographics were found to be similar across respondents categories, travel behavior was found to differ between first, second, and third mailing respondents.

Two approaches have been proposed to correct for the non-response problem. Response rates between 70 and 80% should be achieved if mailings are to be used. Hunt and Dalton (1983) suggest threat telephone survey methods (probably) give more accurate results than mail surveys comprised of only one or two mailings. Telephone surveys are, however, subject to bias resulting from the exclusion of households without telephones. For instance, of the 498 households selected by Hunt and Dalton (1983), 100 did not list a telephone number. Furthermore, and additional 62 could not be located and another 81 had been disconnected or would not cooperate. The proportion of households with telephones increases each year, and thus the problem of bias due to the exclusion of no-telephone households should diminish in the future.

A number of studies report the use of incentives to increase response rates. These incentives have usually been posters of the region sponsoring the study or a promise to send the respondent the summary results of the study once completed.

Expenditures

Respondents are typically asked as to how much they spent while in the location under study. Muha (1974) and Mak et.al. (1977) have attacked conversion studies with respect to the accuracy of expenditures reported by respondents. Mak et,al. (1977) empirically tested the relationship between data from recall surveys and diaries and found that survey respondents tend to underestimate their expenditures. Ellerbrock (1981) proposes the use of diary techniques at the minimum as a control group from which the weight the expenditure figures reported by a large group of survey respondents. The diary technique is not recommended as the primary data collection method. Less than 20% of the persons approached agree to participate in such endavors. Of these, less than 50% return their diaries. Also, daily and accurate entries are doubtful (Woodside 1981). The additional costs incurred by using a diary panel would not be acceptable for the most state tourism offices.

Travelers do accurately respond to questions on expenditures if broad expenditure categories are used. Typically, nine to twelve categories are used (e.g., $0, $1-50, $51-100 etc.). Vehicles, media, ads, and campaigns can be evaluated and decisions made based on revenue projections using such expenditure categories.

Sponsor Identity

The identification of a state or location as the sponsor of the study should be avoided. To desensitize respondents, the questionnaire should be used to collect data on travel behavior with respect to competing travel destinations. Furthermore, the letter accompanying the questionnaire should not refer to the fact that the respondents is known to have written for travel information. Recent conversion studies have been conducted by independent research agencies identified as the sponsor of the study. However, no direct test has been reported to substantiate this proposition.

DATA ANALYSIS

Revenues

In calculating the average revenue per inquiry (RPI), the total number of parties actually visiting has to be estimated. The total number of inquires, e.g., 4,135, is multiplied by the percentage of those respondents who visited; e.g. , 44% for 1,836 estimated inquirers who visited as a result of the advertisement. Woodside and Reid (1974) have noted that this estimate may be inflated since the calculations are based on the assumption that the percentage of inquirers who visited is identical to that indicated by the results. This, of course, can be remedied by a high response rate.

Ellerbrock (1981) and Ballman et. al. (1983) point out to the need to use net versus gross conversion rates. They propose that a true conversion rate is one in which inquirers who decided to visit the destination before ad exposure are not included. Ballman et. al. (1983) suggest that gross conversion rates may be overestimating conversion by 50%; in their study the gross conversion rate was 33%, while the net rate was 22%. Their results are based on answers to the following questions: "Did you request information from us before or after you decided to vacation in X?" The literature received may in some cases cause the final decision, although the respondent will most likely indicate that the decision had been made before the request. Woodside (1981) calls for the use of a true experiment to meaningfully answer the question of whether or not the advertising and the literature influenced visits. Cities could be used for such research along with regional editions of magazines. For a specific magazine, some editions would include and advertisement (i.e., treatment) , while some would not (i.e. , control) . Random samples of all subscribers in all cities could be drawn and information on their travel behavior collected. The travel behavior of the treatment versus the control groups could be compared.

The average expenditure per party, e.g. , $432, is then multiplied by the estimated number of visitors ( 1 836) to arrive at an estimate for total revenue ($793,000). Estimated total revenue divided by the total number of inquirers results in RPI: $793,000/4,135 = $192.

The more specific the destination for which the conversion study is performed (e.g., state versus specific location), the more careful the definition of the barriers of the region to the respondents needs to be. Misunderstanding or misconception may lead to unnecessary inflation of the results.

Costs

Average cost per inquiry (CPI) is obtained by dividing the total cost of the advertising campaign (e.g., $128,000) by the number of inquiries (4,135) for $31.00. It is of utmost importance that all costs be included beyond production and insertion charges; i.e., including inquiry handling costs (e.g., postage, brochure printing, etc.) . It is even appropriate to add conversion study costs to the overall costs of the campaign.

The ratio of CPI as a percentage of RPI ($31.00/$192.00 = 16%) provides a means of comparing the effectiveness of different campaigns, media vehicles or ads. The interesting criterion for a state agency is to see if taxes as a percent of revenue per inquiry (CPI/RPI). In a recent study on South Carolina's advertising tourism campaign, CPI as a part of RPI for five advertising campaigns was less than the estimated 9.9% tax rate (Woodside and Motes 1981).

APPLICATIONS

Although conversion studies are designed primarily to estimate rates of conversions of inquirers into visitors, applications can be extended well beyond this purpose. Woodside and Motes 0 981 ) studied f ive distinct advertising strategies with different creative approaches, media schedules, and direct mail literature which were used to affect vacation behavior of five market segments. Substantial differences in estimated total revenues, costs, and net revenues produced f rom each market and advertising strategy were detected. The findings resulted in the advertiser's reallocation of his advertising efforts for the following year.

Woodside and Motes (1980) compared image versus direct-response advertising and found no significant profile differences between direct-response and image-ad inquirers, but data showed significant differences in cost and net revenue. The findings resulted in the reduction of the advertiser's use of image advertisements.

Woodside and Ronkainen (1982) compared the performance of black-and-white and color ads placed in newspapers versus magazines. Although substantially more national, city and state tourism advertising expenditures are allocated to magazines, newspapers were found to out perform magazines in revenue-generating power.

SUMMARY

Conversion studies, since their introduction, have been developed with respect to controlling factors which may inflate the results. The following improvements have been made:

( 1 ) Controlling for non-response bias by insuring response rates of 70% or better. This can be achieved by following standard methods of increasing response propensity (e.g., Heberlein and Baumgartner 1978; Kanuk and Berenson 1975), such as individualizing the letters, having a postage stamp on the envelopes, and providing incentives.

(2) Not identifying the real sponsor (nor the reason) of the study by using an independent research agency. This avoids unnecessary sensitizing of the respondents and allows for collection of data on competing vacation destinations.

(3) Using net versus gross rates of conversion by factoring out inquires who had decided to visit the destination before ad exposure.

(4) Includiniz all relevant costs, not only inquiry-generating costs, that can be directly allocated to the project such as postage for mailing material to inquirers.

(5) Clearly defining the area(s) of vacation especially in the case of individual locations to avoid possibility of misinterpretation and subsequent inflation of results.

(6) Striving for an accurate estimate of the expenditures incurred by the respondents. Remedial actions include using broad expenditure categories as well as diary techniques as a complementary data-collection method.

Given the approach's intuitive appeal and relative simplicity of the methodology, conversion studies enjoy increasing popularity. However, if the conversion study is not conducted properly, the results can lead into interpretive errors based on inflated results. If executed as an integral part of an advertising campaign (i.e., planning for a conversion study starts with the planning of a campaign), conversion studies can provide invaluable input in improving managerial decision making.

REFERENCES

Advertising Age (1982), "Alaska Goes After Tourists in a Big Way," (4 January), 10.

Churchill, Gilbert A. (1983), Marketing Research, Hinsdale, IL: The Dryden Press.

Ellerbrock, Michael J. (1981), "Improving Coupon Conversion Studies, "Journal of Travel Research, 19 (Spring), 37-38.

Goeldner, Charles R. and Karen P. Dicke (1981), Travel Trends in the UnitedStates and Canada, Boulder, CO: Business Research Division, University of Colorado, 44-50.

Heberlein, T.A. and R.A. Baumgartner (1978), "Factors Affecting Response Rates to Mailed Questionnaires: A Quantitative Analysis of the Published Literature," American Sociological Review, 43 (August), 447-462.

Hunt, John D. and Michael J. Dalton (1983), "Comparing Mail and Telephone for Conducting Coupon Conversion Studies," Journal of Travel Research, 21 (Winter), 16-18.

Mak, James James Moncur, and David Yonamine (1977), "How or How Not to Measure Visitor Expenditures," Journal of Travel Research, 16 (Summer), 1-4.

Muha, S.L. (1976), "Evaluating Travel Advertising"A Survey of Existing Studies," paper presented at the Fifth Annual Educational Seminar for State Travel Officials, Lincoln, NE, (4 October).

Petersonk, Robert A. (1982), Marketing Research, Plano, TX: Business Publications, Inc.

U.S. Travel Data Center (1983), Survey of State Travel Offices, Washington, D.C.: U.S. Travel Data Center.

Wettstein, Earl (1982), "Agency's Business Reply Card Survey Elicits 40% Response; Results Valuable," Marketing News, 15 (22 January), 20.

Woodside, Arch G. ( 19 81 ) , "Measuring the Conversion of Advertising Coupon Inquirers into Visitors," journal of Travel Research, 19 (Spring), 38-41.

Woodside, Arch G. and William H. Motes (1980), "Image versus Direct-Response Advertising," Journal of Advertising Research, 20 (April/May), 31-39.

Woodside, Arch G. and William H. Motes (1981), "Sensitivities of Market Segments to Separate Advertising Strategies," Journal of Marketing, 45 (Winter), 63-73.

Woodside, Arch G. and Ilkka A. Ronkainen (1982), "Travel Advertising: Newspapers versus Magazines,: Journal of Advertising Research, 22 (June/July), 39-43.

Woodside, Arch G. and Ilkka A. Ronkainen (1983), "How Serious is Non-Response Bias in Advertising Conversion Research?" Working paper, College of Business Administration, University of South Carolina.

Woodside, Arch G. and Ilkka A. Ronkainen (1984), "Principles of Pretesting Travel Advertising," Proceedings of the XIV Annual Meeting of the TTRA, Salt Lake City, UT: TTRA.