Email Marketing: Success Factors

Ruth Rettie [HREF1], Senior Lecturer, Kington Business School [HREF2], Kingston University [HREF3],
Kingston,KT12 1JZ. Email:r.rettie@kingston.ac.uk

Abstract

Email marketing is increasingly recognised as an effective Internet marketing tool. Our paper reviews the email marketing literature which highlights the importance of obtaining recipients' permission. Email marketing is compared with other forms of direct and Internet marketing, identifying its key advantages. We identify the factors that have been found to increase response rate in direct marketing and direct mail.

Following exploratory qualitative research among industry experts we analysed 30 email marketing campaigns to identify factors associated with higher response rates. We found the following factors were associated with increased response rate: subject line, email length, incentive, number of images. For nine of these campaigns we were able to link demographic and lifestyle data to response. Analysis of these campaigns suggests that recipients who have previously bought on the Internet have higher response rates to email marketing. These finding are used to create an email marketing process model based on the Vriens et al (1998) direct mail process model.

Introduction

Email marketing is being increasingly recognised as a cost-effective marketing tool. Forrester (Niall 2000) describes email marketing as one of the most effective online marketing tools because of its high response rate, and expects email marketing to be worth 5 billion US dollars by 2004. eMarketer (2000) estimate that 61% of all medium and large US companies use email marketing on a regular basis. Jupiter, (Pastore 2001), predicts that spending on digital marketing initiatives such as coupons, promotions and e-mail will surpass that of internet advertising (advertisements placed within website content such as banners, interstitials, rich media, pop ups, etc).

This paper explores the factors that affect the response rate of email marketing. In the next section we review the literature on email and permission marketing.

Email Marketing

The advantages of email marketing have been recognised by a number of authors. Jackson and DeCormier (1999) recognised that email provided marketers with communication that permitted relationship building and real-time interaction with customers. Wreden (1999, p3) described email marketing as the 'Internet's killer application' because of the precision with which email can be tailored, targeted and tracked. Low costs and digital processing allows companies to send out huge numbers of emails. The medium is push rather than pull, the consumer does not have to instigate the interaction, and currently response rates are high (Di Ianni 2000; Rosenspan 2000). Peppers and Rodgers (2000, p 4), claim that 'clear benefits, including high response rates and low costs are rapidly turning email marketing into an invaluable tool'. Email marketing can be used for acquisition or retention; this paper focuses on acquisition email marketing, marketing designed to win rather than retain customers.

Table 1 Comparison of Direct and Internet Marketing Techniques


Table 1 compares email to other forms of direct and Internet marketing. The basic characteristics of email marketing are low costs, shorter turnaround (in the time involved to prepare, send the messages and receive the responses), high response rates and customisable campaigns. The advent of HTML, audio and video email improves the scope for creativity in email marketing. Ultimately it is conversion, rather than response rate, that will determine the cost efficiency of acquisition email marketing; this will depend on the targeting, the message, and the receptivity of the recipient. Briggs and Stipp (2000) have argued that the 'lean-forward' nature of the Internet increases involvement in streamed Internet advertising, this could equally apply to email marketing.

Email is a relatively new medium, in the future, consumer response is likely to be adversely affected by increasing traffic volume (Rosenspan 2000; Di Ianni 2000). Mehta and Sividas (1995) suggest that spam messages are unwanted, untargeted and therefore negatively perceived. Turban et al, (2000, p360) define spam 'as the practice of indiscriminate distribution of messages without permission of the receiver and without consideration for the messages' appropriateness.' Jupiter Communications (2000) estimate that the average US surfer will receive up to 1,600 unsolicited emails every year by 2005. Windham (2000) believes that unsolicited email is considered an invasion of privacy, and has already become a serious problem for some customers; spam taints the reputation of email marketing. To avoid being perceived as spam, several authors recommend that companies should restrict the messages they send (Wreden 1999; Wright and Bolfing 2001); in addition, marketers should obtain recipients' permission.

Permission Marketing

Godin, (1999, p 43) coined the term 'permission marketing' which is based on consumers giving their consent to receive marketing information. Permission marketing 'offers the consumer an opportunity to volunteer to be marketed to' and it is therefore 'anticipated, personal, and relevant'. The idea of consent is not new; customer permission had been introduced in the context of privacy issues in direct marketing (Milne and Gordon 1993). The key to permission marketing is knowing customer interests and knowing their information needs (Sterne and Priore 2000). It is particularly relevant to Internet marketing because the low marginal cost of messages creates a potential volume problem for both consumers and marketers. Permission marketing improves the targeting and relevance of promotional messages, thus improving response and conversion rates. The interactivity of the Internet facilitates communication of consumer permission and preferences.

A survey by IMT Strategies (1999) found that permission email has a higher response rate than non-permission email; more than half of their respondents felt positive about receiving permission email. Successful permission marketing is about building long-term relationships with customers once the initial permission has been granted. The consent, trust and two-way exchange of information develop the relationship between the consumer and the company.

Hagel and Singer, (1999) discuss the emergence of 'infomediaries' or information brokers who elicit the permission of consumers and preserve their privacy. In effect, these companies are 'permission' brokers; an example is yesmail.com.

Krishnamurthy (2001) presents a model where consumer interest in a permission-marketing program depends on five factors: message relevance; monetary benefit; personal information entry costs; message processing costs, and privacy costs. He also introduces the concept of permission intensity, which he defines as 'the degree to which a consumer empowers a marketer in the context of a communicative relationship'. In addition the permission may be more or less explicit. In-order to obtain as many permission email addresses as possible marketers sometimes provide options that are unclear with a default 'opt-in'. A study by Cyber Dialogue found that 69% of US Internet users did not know they had given their consent to inclusion on email distribution lists (Bellman et al 2001).

The effectiveness of direct marketing depends on the targeting, the nature of the offer, the creative, the timing and the volume of communication (Fraser-Robinson 1989; Stone 1996; Roberts and Berger 1989). Vriens et al., (1998, page 325) develop a theoretical framework for the response process in direct mail, see Figure 1, distinguishing between factors that affect the three stages: opening the envelope, paying attention to the contents and the response.

Figure 1 Direct Mail Response Process


Source: Vriens et al. (1998)

Figure 2 adapts Vriens et als' model to form a basic email marketing response process. There are three stages: opening the email, paying attention to the email and the response.

Figure 2 Basic Email Marketing Response Process

Research Agenda

The literature review, the email response model and our exploratory qualitative research suggest the hypotheses below.

Opening the Email

Due to the prevalence of spam, the first hurdle is to persuade the recipient to open the message rather than delete it, so the subject line promise is important, hence:
H1 Higher response rates correlate with more appealing email subject lines

Paying Attention to the Email

If the email is too long recipients will not be bothered to read it:
H2 Response rate is inversely related to the length of the email

If the email is attractive respondents are more likely to read it; more attractive emails will have more images, hence:
H3 Higher response rate correlate with more images

Click Through URL Link

In line with other direct marketing, response will depend on the appeal of the offer, therefore:
H4 Higher response rates correlate with more appealing incentives

Methodology

Preliminary qualitative research consisted of four depth-interviews with industry experts: an interactive agency consultant, an email list sales director, an email transmission bureau consultant and a director of a leading UK direct mail company. Respondents were asked to identify key factors for effective email marketing campaigns.

The hypotheses above, derived from these interviews and the literature review, were then evaluated in quantitative analysis of 30 acquisition permission email-marketing campaigns run by Claritas (UK) Ltd., a leading UK direct marketing agency. The email addresses were derived from partner companies' registration sites where consumers had opted-in to third party mailings (permission pool model, Krishnamurthy 2001).

The data consisted of two separate data sets.
1. Examples of each email dispatched, together with delivery and response rates.
From these a consolidated data set was created including the email and response characteristics, i.e. subject line, incentive, email length (in terms of the number of scrolls on a 15" screen), number of links, number of images, response rate, delivery rate, and click-through rate. The incentive in each email was coded from 1 - 4 as follows: '4' - competition with a major prize e.g. a holiday abroad, '3' - competition with a small prize e.g. a weekend in the U.K., '2' - a voucher redeemable against purchase or a special discount, '1' - no incentive. The subject line was coded similarly according to the incentive, if any, suggested by the subject line of the email.

2. The Claritas file of 371,072 individuals with email addresses, appended with demographics, lifestyle and email response (each URL is tagged so that responses can be identified).

Analysis

Our analysis was in two stages:
1. Analysis of 30 campaigns examining the correlation between email type (length of email, number of links, number of images, nature of incentive, subject line and click-through rate. The summary statistics for these campaigns are in Table 2.

Table 2 Summary of Statistics

2. For nine campaigns we had sufficient data to combine demographic and lifestyle with email and response data. These nine campaigns included 136,189 delivered emails with a 3% average click-through rate. Demographic and lifestyle data consisted of 16 attributes: gender, marital status, number of children, occupation, home ownership, age, lifestage, income, bingo, eating out, fishing, foreign travel, quality news readership, mid-market news readership, popular news readership, and buy products online. The demographic and lifestyle response profiles were biased by the original targeting applied in selecting recipients for the campaign, for example, a campaign for mobile phone services is biased by the selection criterion, e.g. 'age under 35, income over £15,000'. It is therefore not meaningful to combine responses for the nine campaigns. We used an index to normalise the effect of targeting for each campaign by calculating the response rate for each attribute (e.g. male) as an index of the average response rate. Thus, an index of 176 reflects a 76% above average incidence of response for that profile.

Results

1. Analysis of 30 Campaigns.

Upon first review the data was checked for skewness (a statistic which measures the lack of symmetry in a distribution of data); this showed a high level of skewness at 2.173; removal of three outliers reduced skewness to 0.629, which is within the acceptable range. Using Kendall's Tau non-parametric correlation test (because we had a small data set with a large number of tied ranks), we found that click through percentage was significantly related to subject line, length of email (inverse relationship), number of images and number of click-throughs, as shown in Table 3.

Table 3 Kendall's Tau Correlation

The following hypotheses are supported:
H1 Higher response rates correlate with more attractive email subject lines
H2 Higher response rates correlate with more attractive incentives
H3 Response rate is inversely related to the length of the email.
H4 Higher response rate correlate with more images.

We also found a significant correlation between the number of links in an email and the click-through rate; this is probably because the number of links is correlated with the number of images. A partial correlation, controlling for number of images, found no independent significant relationship.

To take the analysis one stage further the metric variables: length of email, number of images and number of links were entered into stepwise linear regression with click-through rate as the dependent variable. This regression produced a model explaining 54% of the variance in click-through (R2 = 0.537); the number of images accounted for 29% of the variance, with the number of scrolls accounting for a further 25%. Adding the number of links to this model did not produce a significant improvement. These results support hypotheses H2 and H3.

2. Demographic and Lifestyle Analysis of Nine Campaigns.

Examination of the indices for the nine campaigns shows that 'having bought online' is well above average for all campaigns with an average index of 174; the indices for age 30 -34 and income over £35,000 were also significantly above average. Table 4 shows the attributes that were high or low for the majority of campaigns and their average indices. Gender shows a slight tendency to be more male than female. 'Bingo', and ' read popular newspapers' show the majority of indices below 100, compared with 'foreign travel' and 'eating-out' with most indices slightly over 100. This, together with the bias to higher incomes suggests, an upmarket bias in response rate.

Table 4 Indices of Demographic and Lifestyle Variables

Research Limitations

This research is limited by the number of campaigns, and their variation in terms of targeting and email contents. Demographic and lifestyle profiles are biased for each campaign in three ways: 1) the original targeting applied in selecting recipients for the campaign, 2) the nature of the product category and 3) the incentive offered. For example, a campaign for mobile phone services is biased by the selection criterion, e.g. 'income over £15,000' and then further bias is introduced both by the offer, i.e. mobile phone services and the incentive e.g. a free phone, which have a different appeal to different segments. The indices we used normalise the bias of targeting. However, it was not possible to normalise the differential appeal of product category and incentive.

Further Research

The literature, the email response framework and our exploratory qualitative research suggest several hypotheses that were not testable from our data.

Opening the Email

Recipients are more likely to open emails from known senders, hence:
H5 Higher response rates correlate with email from well-known brands
Recipients are more likely to open emails where they have given permission, so:
H6 Permission email has higher response rates
Unfortunately,nearly all of our campaigns were from well know brands and all were subject to third party permission, so we were not able to test these hypotheses.
Spam and the volume of email received is a situational characteristic that may affect response, therefore
H7 Response rates are higher for respondents who receive less email.

Paying Attention to the Email

More attractive emails will be in HTML, therefore
H8 Response rate is higher for HTML than text messages.
As nearly all our emails were sent in HTML we were unable to test this. Investigation of this issue is also complicated, because some emails are sent using both HTML and text, allowing the recipient to choose, while others use 'sniffer' technology, which detects whether the recipient email client can read HTML and then sends the appropriate version.

Click Through To URL Link

The higher response rate of those who buy online, and the upmarket bias, may reflect longer use of the Internet so a further hypothesis is:
H9 Respondents who have used the Internet for longer are more likely to respond.
These hypotheses together with the research findings are incorporated into the email marketing response process shown in Figure 3.

Figure 3 Email Marketing Response Process

Conclusions

Our qualitative research suggests that email marketing is growing rapidly and should be integrated into the overall communication mix. Interviewees also believed that email marketing would be more effective as a retention rather than as an acquisition tool, because its interactivity facilitates two-way communication. The experts interviewed also stressed the importance of targeting and the use of 'permissioned' lists. As the latter varies from 'double opt in' company-specific lists to pooled 'opt out' lists, it is likely that the specificity and intensity of permission will also affect response rate.

Our response process model suggests that there are three stages in effective email marketing: getting the recipient to open the email, getting them to pay attention to the email and persuading them to click through the URL link, hence response rate should depend on the email header as shown in the in-box, the email contents and the recipient. Our quantitative research supports this model, with a significant correlation between response rate and subject line, email length, incentive, and number of images. Regression on email length and number of images accounted for 54% of the variance in response rate.

Analysis of demographic and lifestyle data for nine campaigns found a significantly higher response for respondents who had bought online, who were aged 30 -34 or who had incomes over £35,000. These may relate to length of Internet use and early Internet demographics, but we were unfortunately unable to test this. Whilst this would auger well for the industry as email marketing response rate would increase with length of Internet use, the dramatic growth in 'non-permissioned' email marketing or spam may undermine the development of acquisition email marketing.

References

Bellman, S., Johnson, E. J. and Lohse, G. L. (2001), "To Opt-In or Opt-Out? It Depends on the Question", in Communications of the ACM, v.. 44, n.2, p. 25-27.

Briggs, R., Stipp H.(2000), "How Internet Advertising Works: New Evidence and Directions for Future Study", Esomar Pre-Congress Workshop.

Di Ianni, A. (2000), "The E-business Enterprise and the 'Web-first' Principle of E-marketing", Journal of Interactive Marketing, v.2, n.2, p. 158-170.

Egan, D. E., Remde, J. R., Gomez, L. M., Landauer, T. K., Eberhardt, J. and Lochbaum, C. C (1989). " Formative Design-Evaluation of SuperBook Research Contributions" in ACM Transactions on Information Systems v.7 n.1 p.30-57.

eMarketer (2000), eMail Marketing Report, eMarketer, New York, N.Y.

Fleming, J. (1998). Web Navigation: Designing the User Experience. O'Reilly, Beijing.

Hagel, John and Singer, M. (1999), Net Worth, Harvard Business School Press, Boston, MA.

Fraser-Robinson, J. (1989), The Secrets of Effective Direct Mail, McGraw-Hill, London.

Godin, S. (1999), Permission Marketing: Turning Strangers Into Friends, and Friends into Customers, Simon and Schuster Publishing Company,New York, N.Y.

IMT Strategies (1999), "eMail Marketing Report", IMT Strategies, Stamford, CT.

Jackson, A. and DeCormier, R. (1999), "E-mail Survey Response Rates: Targeting Increases Response", Journal of Marketing Intelligence and Planning , v.17, n.3, p. 135-139.

Jupiter Communications (2000), "Media Metrix Executive Summary", Jupiter MMXI.

Krishnamurthy, S. (2001), "A Comprehensive Analysis of Permission Marketing", Journal of Computer Mediated Communication, v.6, n.2, Available online [HREF4]

Leventhal, L. M., Teasley, B. M., Instone, K., Rohlman, D. S., and Farhat, J. (1993). "Sleuthing in HyperHolmes: An Evaluation of Using Hypertext vs. a Book to Answer Questions" in Behaviour and Information Technology 1993 v.12 n.3 p.149-164.

Mehta, R. and Sivadas, E. (1995), "Direct Marketing on the Internet: An Empirical Assessment of Consumer Attitudes", Journal of Direct Marketing, v.9, n.3, p. 21 -31.

McKnight, C., Dillon, A. Richardson, J. (1991). Hypertext in Context. Cambridge University Press, Cambridge.

Milne, G.R. and Gordon M.E. (1993), "Direct Mail Privacy-efficiency Tradeoffs Within an Implied Social Contract Framework', Journal of Public Policy and Marketing, v.12, n.2., p. 45 - 53.

Niall, J. (2000), The Email Marketing Dialogue, Forrester, Cambridge, M.A.

Pastore, M. (2001), "Ad Spending to Rebound, Digital Marketing to Soar", Cyberatlas, Available online [HREF5]

Peppers, D. and Rogers, M. (2000),Email Marketing Maximized, Peppers,Stamford, C.A.

Rask, M. and Dholakia, N. (2000), "Next to the Customer's Heart and Wallet: Frameworks for Exploring the Emerging M-commerce Arena", working paper. Available at [HREF6]

Roberts, M.L. and Berger, P.D. (1989), Direct Marketing Management , Prentice-Hall, Englewood Cliffs, N. J.

Rosenspan, A. (2000), "Permission is Not Enough", Journal of Interactive Marketing, v.2, n.3, p. 215-218.

Sterne, J. and A. Priore (2000), Email Marketing: Using Email to Reach your Target Audience and Build Customer Relationship, John Wiley and Sons, Inc, New York, N.Y.

Stone, B. (1996), Successful Direct Marketing Methods, NTC Publishing Group, Chicago.

Turban, E., Lee, J., King, D. and Chung, H M. (2000), Electronic Commerce: A Managerial Perspective, Prentice-Hall, New Jersey, N.Y.

Vriens, M., van der Scheer, H.R., Hoekstra, J. C. and Bult, J. R. (1998), "Conjoint Experiments for Direct Mail Response Optimisation", European Journal of Marketing, v.32, n.3, p. 323- 339.

Windham, L. (2000), The Soul of the New Consumer, Allworth Press,New York, N.Y.

Wreden, N. (1999), "Mapping the Frontiers on Email Marketing", Harvard Management Communication Letter , 9 January.

Wright, N.l D. and Bolfing, C. P. (2001), "Marketing via Email: Maximizing its Effectiveness Without Resorting to Spam", working paper, James Madison University.

Yeshin, T. (1998), Integrated Marketing Communications, Butterworth-Heinemann,Oxford,U.K.

Hypertext References

[HREF1]
http://www.kingston.ac.uk/~bs_s520/
[HREF2]
http://business.king.ac.uk/facbus.html
[HREF3]
http://www.kingston.ac.uk/
[HREF4]
http://www.ascusc.org/jcmc/vol6/issue2/krishnamurthy.html/
[HREF5]
http://cyberatlas.internet.com/markets/advertising/article/0,,5941_862241,00.html
[HREF6]
http://ritim.cba.uri.edu/wp2001/wpdone1/wpdone1.htm/Copyright

Ruth Rettie© 2002. The authors assign to Southern Cross University and other educational and non-profit institutions a non-exclusive licence to use this document for personal use and in courses of instruction provided that the article is used in full and this copyright statement is reproduced. The authors also grant a non-exclusive licence to Southern Cross University to publish this document in full on the World Wide Web and on CD-ROM and in printed form with the conference papers and for the document to be published on mirrors on the World Wide Web.