Conversions on the rise – modernizing e-mail marketing practices by utilizing volunteered data

Purpose –The purpose of the study is to examine how utilizing volunteered data influences the response and unsubscribe rates of e-mail marketing to consumers. Design/methodology/approach – In three longitudinal field experiments conducted among 1,864 applicants of a higher education institution, the study compares customized marketing e-mails based on volunteered consumer data to e-mails that are personalized based on observed consumer data and to control emails that are not tailored by the marketer at all. Findings – The results indicate that marketers should make consumers active participants in the communication process, as response rates are higher in those e-mails where volunteered data are utilized. However, the unsubscribe rate is the highest in customized e-mails. Research limitations/implications – The authors demonstrate that e-mails displaying empowering aspects influence consumers’ behaviors and lead to outcomes that mostly outperform non-empowered e-mails. Practical implications –Compared to other forms of interactivemarketing, e-mail has lagged behind in both popularity and customer-friendly implementation. However, it has the potential to succeed if marketers pay more attention to consumer empowerment. As over 306 billion e-mails are sent worldwide daily and 75% of marketers use e-mail when contacting customers, the increase in response rates can have a significant influence on their returns. Originality/value –Unlike prior research the focus was on the process of tailoring, this perspective supports customer advocacy and emphasizes consumers’ important role in creating engaging, empowering e-mail marketing communication.

Introduction "Dear Professor, Greetings. I hope this message finds you well. As you are a respected professor in your field of expertise, I would like to take this opportunity and introduce our professional copy editing and proofreading services to aid in the accuracy, clarity, and readability of the documents before and during the publishing process." Probably, all of us have received these messages sometimes: e-mails often sent without permission, selling a product or service that might be relevant for us, but usually, however, it is not. Marketers are relying on collected or purchased data about us and, in many cases, try to tailor the message based on this information (Yun et al., 2020). The example was sent to a doctoral student; thus, being titled as Volunteered data influencing e-mail marketing "a respected professor" raised a laugh and some shares on the social media but did not convince the recipient about the offering of the marketer, even though the service might have been beneficial. The end result was wasted time and effort for both the marketer and the recipient and even for the e-mail marketing industry as a whole (Pavlov et al., 2008).
Even though e-mail is a legitimate, profitable and widely used interactive marketing communication channel, its effectiveness has suffered from spamming. From the consumer's perspective, marketing e-mails are often irritating and irrelevant. This is paradoxical in that marketers today have better opportunities than ever before to tailor and target messages according to consumer expectations by utilizing consumer data (Line et al., 2020). Driven by computation, marketers can personalize marketing communication to individual recipients (Yun et al., 2020) and create an engaging, active connection with them in real time (Wang, 2021). This, in turn, helps build favorable attitudes toward marketing (Bhatia, 2020;Bacile et al., 2014;Oh and Sundar, 2015), increase the response and conversion rates (Line et al., 2020;Kemp et al., 2021) and advance customer brand advocacy (Bhati and Verma, 2020).
Interactivity is an integral part of contemporary marketing practice (Wang, 2021) especially on social media (Qin, 2020) but also on other digital platforms. While new interactive marketing practices, such as live streaming, mobile apps or gamification, have rapidly evolved toward a more participative and engaging direction, most marketing e-mails still appear to be based on one-way persuasive communication (Wang, 2021;Brandon, 2015). However, e-mail marketers are collecting, processing and analyzing increasing amounts of data that they could use to tailor their marketing.
In general, data have become one of the world's most valuable resources, and much of it relates to individuals either directly or indirectly (Abrams, 2014). OECD (2019) divides the origin of data in two: volunteered and observed data. Volunteered data are information that individuals actively and purposefully share about themselves or others. Observed data are information about the actions of individuals, such as opens and clicks on e-mails, search terms used or purchases made (Malthouse et al., 2018). Observed data are captured implicitly without the participation of the data subject. Today this "big data" can be generated through several sources, analyzed immediately (Line et al., 2020;Hofacker et al., 2016) and used, for example, in behavioral advertising (Bhatia, 2020). Malthouse et al. (2018) use another typology to categorize consumer data: first-party data are data gathered from the marketer's internal databases and thus owned by the marketer. Second-party data are another organization's first-party data that are bought or traded and used, for example, to reach new audiences. Third-party data can be purchased from data aggregators who gather it from various platforms and websites and pay data owners for their first-party data (Lotame, 2019). A later category of data is "zero-party data," which is any data that a customer proactively and deliberately share in exchange for personalized experience (Britt, 2020;Gilliland, 2020;Yun et al., 2020). Zero-party data are also known as "little data" (Hofacker et al., 2016) describing data that the marketer gather directly from consumers, for example, through questionnaires (Yun et al., 2020). In this study, we use the term "volunteered data" to cover voluntarily given zero-party data.
When marketers only keep the data that consumers grant the right to use for specific purposes and provide benefits in exchange of the use of that data, the consumer's role changes from a mere target of offerings to an active participant in the value creation process (Line et al., 2020;Britt, 2020). The change is well suited to our time because the information revolution, especially the internet, has improved the skills and knowledge of consumers and balanced the power between marketers and consumers (Hartemo, 2016). This is called consumer empowerment, which is used as a specific context of e-mail marketing in this study. As Prahalad and Ramaswamy (2004) phrase in their article, "informed, networked, empowered, and active consumers are increasingly co-creating value with the firm". This thinking is in line with the service-dominant logic of marketing in which consumers are viewed as operant resources that can create value as co-producers of services or partners to be "marketing with" (Lusch and Vargo, 2009;Line et al., 2020). Here, the firm-centric and company-driven view is replaced with personalized and interactive consumer experience (Wang, 2021) in which consumers are expecting engagement instead of interruptions (Ho et al., 2020).
To the best of our understanding, scholars have not so far rigorously tested whether volunteered data are useful in interactive marketing. Conducting such a study helps marketers to find an optimal way to implement marketing communications. This is especially important in e-mail marketing, which is fighting for its place in the midst of new and effective interactive marketing practices (e.g. social media marketing or influencer marketing described in Wang, 2021). If e-mail marketing is developed to work with the same principles as these practices, it may be better able to defend its existence in the future. E-mail as a means of communication has managed to hang on: 86% of business professionals prefer e-mails, millennials spend about 6.4 h daily scrolling through their inbox and even Generation Z has started using e-mails to complete daily tasks during the coronavirus disease  pandemic (Dziak, 2021;Campaign Monitor, 2020). Furthermore, e-mail marketing is still performing strongly in terms of return on investment (ROI) and has a good chance to take advantage of future advancements such as artificial intelligence (Dziak, 2021).
Answering the call for modernizing e-mail marketing, this research investigates how content tailoring influences the results of e-mail newsletters when the tailoring is based on (1) the data the recipients have provided to the marketer voluntarily (i.e. customization) or (2) the customer data derived from the databases of the marketer (i.e. personalization). Customization is user initiated, which is done by the customer and personalization is system-initiated and done for the customer (see, e.g. Sundar and Marathe, 2010). Table 1 summarizes the terms utilized in the theoretical development regarding tailoring and the types of data. Additionally, a non-tailored e-mail is used as a baseline in the controlled experiments to provide empirical evidence in support of the cause-and-effect relationship of the differences in the outcomes of the two aforementioned conditions (Vargas et al., 2017).
Personalization and customization are recognized and widely used interactive marketing strategies that have been researched extensively (in e-mail marketing, e.g. Ansari and Mela, 2003;Micheaux, 2011;Sahni et al., 2018). However, previous academic research in interactive marketing has not paid enough attention to who contributes to the tailoring. Similarly to the research by Sundar and Marathe (2010), this study focuses on the process of tailoring but examines e-mails instead of websites. Their study used questionnaires to evaluate undergraduates' attitudes to different website contents, while the current one collects behavioral results unobtrusively in a real-life situation. Bacile et al. (2014) studied co-production in a mobile-coupon campaign by letting consumers to decide a message's delivery time. According to their results, co-producing communications leads to a strong purchase-redemption rate.
This paper assumes that consumers that are activated to disclose volunteered data about themselves provide a better response to e-mail marketing when measured with behavioral data (open, click-to-open and conversion rate). Previous studies suggest that developing personal connections with consumers is positively related to engagement and attitudes toward the marketer, which can lead to purchase intentions and revenue generation Data used (Kemp et al., 2021;Qin, 2020). When consumers are activated, they get engaging and personally relevant content and are positive about receiving e-mails and act on them.
The remainder of this paper is organized as follows. The next section reviews the literature and describes the development of four hypotheses. Next, the research methodology and the field experiment are described, which is followed by the results of the research. Finally, implications for research and practice are discussed and limitations and avenues for future research suggested.

Literature review and hypotheses
Empowering e-mail marketing approach E-mail marketing is very popular among marketers but has often a negative image among consumers. One suggestion for overcoming this discrepancy is to embrace an approach that empowers consumers instead of exploiting them. This study builds on the systematic literature review of Hartemo (2016), which suggests that e-mail marketing can be used to empower consumers in three aspects as follows: (1) by obtaining permission before sending the e-mail, (2) by making consumers active participants in the communication process and (3) by making e-mails relevant for the recipients.
Obtaining permission is the first step in a successful, empowering e-mail approach, as consumers who have expressed their willingness to receive marketing messages consider them to be important and relevant (Bhatia, 2020). For today's consumer, it is important to be able to control the information flow (Krafft et al., 2017) and the frequency, time and content of the messages (Bhatia, 2020). An explicit consent to send direct marketing is also required by law in many countries. In the European Union, the introduction of the General Data Protection Regulation (GDPR) in 2018 further strengthened and unified data protection for individuals, stating that consent must not only be explicit, but also specified and informed (European Union, 2019). The study at hand considers permission as a necessity, as the field experiment is conducted in a member state of the European Union.
Furthermore, obtaining permission is an important part of an empowering e-mail marketing approach because it creates commitment and trust. The commitment-trust theory of Morgan and Hunt (1994) indicates that successful relationship marketing requires relationship commitment and trust, whichwhen presentpromote efficiency, productivity and effectiveness. Trust exists when one party has confidence in the other party's reliability and integrity, as may be assumed when the consumer is giving permission for an organization to send e-mail marketing. That said, consumers might also have other reasons to grant permission. For example, they may perceive e-mails useful or entertaining or benefit from monetary incentives when subscribing (Bhatia, 2020;Krafft et al., 2017).
The second and third aspects in the empowering e-mail approach are intertwined. Making consumers active participants in the communication process is central to making e-mails relevant to them. An active relationship requires interactivity, two-way communication through active connection, engagement and participation (Wang, 2021). The revolution of digital platforms has greatly facilitated the bi-directional value creation. Today's consumers are not only able to tell their interests to the marketer, but also expect these preferences to be considered in communication addressed to them (Sundar and Marathe, 2010;Prahalad and Ramaswamy, 2004;Kemp et al., 2021). In these customer-managed relationships consumers are empowered to determine, for e.g. what kind of information they want and how often (Pires et al., 2006). In the study of Bacile et al. (2014), consumers who were activated by letting them contribute, improved the attitude toward the marketing communication, purchase intent and purchase activity.
The theory of psychological ownership (Pierce et al., 2001) gives an explanation as to why consumer contribution is the key in making e-mail marketing more engaging and effective.
Psychological ownership occurs when individuals have possessive feelings about tangible or intangible objects, ideas, persons, technology (Kirk et al., 2015) or brands (Chang et al., 2015) regardless of actual legal ownership. Fuchs et al. (2010) propose that in the new product development, consumers develop a sense of psychological ownership when they are empowered to contribute to the process. Likewise, in e-mail marketing, making consumers active participants in the communication process enhances consumer engagement behavior. They feel as though the e-mail is "theirs" and adopt a more favorable attitude toward e-mail communication. The agency model of customization of Sundar (2008) similarly argues that higher interactivity leads to a greater sense of being gatekeeper to one's own information universe. Message interactivity enhances message elaboration, leading to more positive attitudes to the message topic (Oh and Sundar, 2015).
Relevance requires customer-oriented information captured and processed in a way that the dialog and timing of the messages can build up on the individual's preferences and needs. We argue that the best source for personal needs and wants is the individual itself. Especially, the individual non-behavioral data relating to interests, beliefs, values and other lifestyle characteristics would be more accurate when collected voluntarily from consumers (Robertshaw and Marr, 2006). An increase in personally relevant messages, in turn, improves attitudes toward marketing (Bhatia, 2020). According to the theory of reasoned action (Fishbein and Ajzen, 1975) and its extension, the theory of planned behavior (Ajzen, 1991), attitude influences behavior.
Hypotheses' development Based on the literature review described in the previous section, we assume the following: emails that display empowering aspects (permission, active participation and relevancy) influence consumers' behaviors, leading to outcomes that outperform non-empowered e-mails.
We are manipulating the process of tailoring (i.e. the aspect of empowerment), assuming to find differences in behavioral outcomes: opening the e-mail, clicking on a link and giving a response. Each of these steps is an indication of interest that can be measured with behavioral clickstream data. The first outcome we are interested in is opening the e-mail, which can be measured with open rate (OR) and is an indication of the recipient's attention (Micheaux, 2011;Hartemo et al., 2016). We hypothesize as follows: H1. E-mails that utilize volunteered consumer data to tailor the e-mail contents lead to a higher OR compared to e-mails where observed consumer data are used.
After opening, the recipient is expected to read the newsletter. Clicking on a link on the newsletter indicates that the recipient is interested in the content and wants to learn more. By measuring the click-to-open rate (CTOR), we can demonstrate how much recipients are engaging with the content of the e-mail and thus form a perception of the relevancy of the email (Micheaux, 2011). Therefore, our second hypothesis is as follows: H2. E-mails that utilize volunteered consumer data to tailor the e-mail contents lead to a higher CTOR compared to e-mails where observed consumer data are used.
The behavioral metrics of customer interactions such as OR and CTOR help to evaluate the performance of e-mails but do not always result in revenue or profitability for the marketer. When measuring business impact, conversion rate is relevant (Yun et al., 2020). Conversion rate describes the number of e-mail recipients that complete a desired goal set by the marketer, such as a purchase, enrollment or filling in a questionnaire. Conversion is typically made on the linked landing page. Sometimes it may be difficult to determine if the conversion results from the e-mail or, for example, some other marketing or business effort Volunteered data influencing e-mail marketing (Yun et al., 2020). Nevertheless, from the business perspective it is important to formally test the hypothesis as follows: H3. E-mails that utilize volunteered consumer data to tailor the e-mail contents lead to a higher conversion rate compared to e-mails where observed consumer data are used. Krafft et al. (2017) studied reasons that caused consumers not to grant permission for marketing. They identified drivers related to benefit and cost to the consumer, which are described in various theories explaining the basics of human interaction (Krafft et al., 2017). According to these theories (e.g. the social exchange theory of Homans, 1961), interaction with an organization happens when consumers feel that the benefits outweigh the costs of the exchange. As soon as the costs feel higher than the benefits, the relationship may be terminated. In e-mail marketing, this is done by unsubscribing the newsletter. The unsubscribe rate is thus useful in understanding the recipient's engagement. In their research, Sahni et al. (2018) found that unsubscribe rate decreased when the e-mail was personalized with the recipient's name due to increasing elaboration of the communication. In our research context, this would mean that the unsubscribe rate is lower in the observed than the control condition. However, our interest lies in the volunteered condition, where recipients have explicitly expressed a will to engage in conversation with the marketer. Therefore, we assume that they want to remain subscribers more than the recipients in the observed condition. It is postulated as follows: H4. E-mails that utilize volunteered consumer data to tailor the e-mail contents lead to a lower unsubscribe rate compared to e-mails where observed consumer data are used.

Data and methodology Methodology
Our controlled between-subject experiment examined the effect of a random allocation of email recipients to two different conditions: one that utilized volunteered data (customization) and another one that utilized observed data (personalization). In addition to these experimental treatments, a control group was set up to be used as a baseline to verify the differences in outcomes (Vargas et al., 2017). The conditions were identical except that the independent variable of customization/personalization (or process of tailoring) was changed for the treatments but was held constant in the control condition. Dependent variables were OR, CTOR, conversion rate and unsubscribe rate. We also monitored delivery rate, which was needed to calculate OR, conversion rate and unsubscribe rate. The clicks were proportional to the number of messages opened (thus, we measured CTOR), which enabled us to control possible unequal ORs (Lewis et al., 2013). Dependent variables got either a value of 1 or 0, depending on whether there was a response (1) or not (0). The results were aggregated in the analysis and did not review individual recipients. The research involves three experiments on the same research subjects during a fivemonth period. The benefit of a longitudinal study is that reliability and statistical power increase, often leading to more accurate conclusions (Ployhart and Vandenberg, 2009).

Research context
The experiment tested permission-based e-mail marketing sent to individuals who had applied to a Finnish higher education institution (hereafter: university) in autumn 2019 but were not admitted. The university is regularly leveraging e-mail marketing to reach potential students and has identified it as a viable way to increase their interest in applying. A similar e-mail newsletter was sent to applicants in summer 2019 with good results, but the university wanted to develop their communications to be more engaging. Thus, they suggested testing our hypotheses with their following batch of e-mails.
As studying is produced and consumed simultaneously, it is difficult to evaluate the service in advance. In this regard, higher education is a service industry (Musa et al., 2011). A study place is also a high-involvement product, requiring a lot of information before selection (Kurian, 2013), even in Finland, where university studies are free. A total of 67% of Finnish university applicants were under 30 years old in 2020 (Education Statistics Finland, 2020); thus, this group of 67% represents digital natives expected to be familiar with different Internet technologies and active communication (Prensky, 2001). Both the product (a study place in the university) and customer (digitally talented consumer/applicant) represent a contemporary empirical context for studying e-mail marketing.
The university has three joint application periods each year: in January, March and September. It is also possible to apply to the university through open university between the joint application periods. In general, competition for entry is fierce in Finland, as two-thirds of applicants are left without a study place every year (Ministry of Education and Culture, 2019). The Ministry of Education defines degree objectives of the universities as well as the appropriations allocated on the basis of these, which in practice mean that universities cannot freely adapt the number of their study places. Therefore, they need to look for ways to fill in also the less popular degree programs. In the case university, there were, on average, 3.4 primary applicants per study place in 2019, but the interest varied a lot depending on the degree program, being between 0.6 and 25.1 (Education Statistics Finland, 2020). By participating in the experiment, the university had three goals: to maintain the applicants' interest in the university despite the fact they were not selected at the first place, to get them to apply again in the next application period and, therefore, to fill in the study places that lacked applicants.

Sampling and data collection
The research subjects had given an approval of handling their personal data for purposes of direct educational marketing when enrolling to the admission register (Finnish National Agency for Education, 2019). The following data concerning the applicant were used: e-mail address, preferred application options and the first language. Only those who indicated Finnish as their first language were selected as research subjects.
In their application, the research subjects were able to choose from 20 different degree programs, which the university had grouped in four fields of study: art, business, technology and health care and social services. The art applicants were filtered out from this study, since there were not any study options to offer them in the time of the first experiment. The total number of research subjects was 1,864.

Field experiments
The research subjects were divided into three different conditions. There were two treatment groups: (1) activated: those who were activated to give volunteered data and received customized e-mails based on these preferences and (2) observed: those who received personalized e-mails based on the observations the researchers made about their preferred application options. In addition to these, there was a (3) control group with a non-tailored content.
The experiment consisted of an activation phase and three experiments as indicated in the research design shown in Figure 1. The applicants received three or four e-mails during the five-month research period, depending on which condition they belonged to.
The three conditions were formed during the activation phase. Based on random selection, half of the research subjects received an e-mail in which they were asked to answer to a In each phase, those who unsubscribed filtered out 15 10 11 11 Those who did not select future application options filtered out 9 Figure 1. Research design survey ("Choose one or more fields of study and types of content you are interested in"). The e-mails were sent, and responses recorded in Apsis e-mail marketing software. In total, 118 individuals (12.7%) answered to the survey but 9 individuals were filtered out on the grounds that they did not want information about the future application opportunities, which they would have received in the experiments. After these procedures, 109 individuals were left in the activated condition. Based on clickstream data, those 799 individuals who had not unsubscribed or answered to the activation e-mail were moved back to the control group. The research subjects in the control group were then randomly assigned on either observed (854 individuals) or control (877) condition. Next, the data of the observed treatment group were reviewed, and recipients were divided in three different subgroups based on the preferred application options they had indicated when applying to the university: technology, business or health care and social services. In the activated condition, the recipients were divided in the aforementioned subgroups or combinations of these based on their selections during the activation phase (seven subgroups in total).
In the first experiment, the subject line, the main image and all the text in the e-mail were modified based on the subgroup. E-mails for activated and observed conditions were identical so that, for example, applicants interested in/applying for business studies received an e-mail with a similarly tailored content in both conditions ("In the field of business, you can now apply for five degree programs. There are a total of 20 different degree programs during this application period."). The control group received a general, non-tailored version of the newsletter ("You can now apply for 20 different degree programs"). Each e-mail contained information about applying to the open university and mentioned that the next joint application period starts in January.
After two weeks, OR, CTOR and unsubscribe rates were documented. These are shown at condition level during different phases in Table 2. Conversions indicate how many actually applied during the application period. This information was received from the university after the application period ended.
In the second experiment, tailoring the e-mails based on the field of study was not feasible, as there were only four study options to be offered. Instead, the subgroups were formed based on the level of studies: a bachelor's or master's degree. In the e-mail, the main image and news were the same to everybody, containing basic information about the application period. That was because the principles of the joint application period changed in Finland in 2020 and the Volunteered data influencing e-mail marketing university needed to inform the applicants about the change. However, the subject line of the e-mails as well as two other pieces of news and their images were tailored based on the level of studies. The tailored news provided information about study options and reference cases of some current/alumni students.
The activated and observed conditions received identical e-mails: either bachelors or masters (e.g. "There are a total of 80 study places in our bachelor's programs."). The control group received a general, non-tailored version of the newsletter (e.g. "There are a total of 125 study places in our programs."). Again, OR, CTOR and unsubscribe rate were documented two weeks after the sending and conversion rate after the application period ended.
The third experiment took place during the second joint application period, which offers many study options. Therefore, the subgroups were formed at the most detailed level based on the field of study and the level of degree (e.g. "There are eight degree programs leading to a bachelor's degree in business administration."). The e-mail contained five different news texts and five images, of which three were tailored. The subject line emphasized timing and was the same for all the e-mails. The control group received a non-tailored newsletter (e.g. "There are over 608 programs in the fields of technology, business, health care and social services."). OR, CTOR and unsubscribe rate were documented two weeks after the sending, and the conversion rate after the joint application period ended. After all the phases, there were 1,806 remaining recipients, and thus, 3.2% had unsubscribed.

Results
In all the experiments, those e-mails that were customized based on volunteered data led to the highest OR, when evaluated on aggregated level (rather than separately for each condition in each phase), as shown in Table 3. The collective OR for the activated condition was 70.2%, whereas for the observed condition it was 51.0%. The results of a chi-squared test showed statistically significant differences in the ORs between the activated and observed condition (X 2 5 41.44, p < 0.001). Interestingly, the OR in all the experiments was lowest in the e-mails personalized based on observed consumer data and not in the control e-mails. In the three experiments, aggregated OR ranged from 56.3 to 85.3% in activated, from 44.5 to 61.1% in observed and 46.0-69.5% in the control condition. There was considerable variation across different treatments, as e-mail metrics ranged from 0 to 100%. Thus, H1 is supported.
Also, in all the experiments, those e-mails that were customized based on volunteered data led to the highest CTOR. The collective CTOR for the activated condition was 35.5%, whereas for the observed condition it was 20.5%. The difference is significant (X 2 5 24.18, p < 0.001). In two of the three experiments, CTOR was lowest in the control e-mails. In the three experiments, aggregated CTOR varied from 32.4 to 38.7% in activated, from 14.3 to 25.8% in observed and from 8.6 to 26.9% in the control condition. Again, e-mail metrics ranged from 0 to 100%. H2 is supported.
Conversions were highest in the condition where volunteered data were used, except for the second experiment where the number of conversions was very low overall (13 in total). The collective conversion rate for the activated condition was 19.9%, whereas for the observed condition it was 14.4%. The difference is significant (X 2 5 6.71, p < 0.01). Conversions across treatments varied from 0 to 100%. The aggregated responses in experiment 1 and 3 ranged from 20.2 to 38.5% in activated, 8.9-33.4% in observed and 10.4-33.1% in the control condition. Thus, H3 is supported. As expected, the conversion rate was the highest during the second joint application period (experiment 3), which is the most popular one in Finland. Unsubscribes were very low throughout the experiments. Contrary to our expectations, in two of the three experiments, the unsubscribe rate was highest in e-mails that were customized based on volunteered data. The collective unsubscribe rate for the activated condition was 1.3%, whereas for the observed condition it was 0.6% and for the control condition 0.5%. The difference is not significant (X 2 5 2.24, p < 0.13). The unsubscribes across treatments varied from 0 to 9.1%. Our evidence did not support H4. Instead, e-mails that utilized volunteered consumer data to tailor the e-mail content led to a higherunsubscribe rate compared to e-mails where observed consumer data were used. Both led to a higher-unsubscribe rate than in the control e-mails where consumer data were not used at all.

Discussion and implications
The purpose of the study was to examine the consequences of utilizing volunteered data in email marketing to consumers. The research postulated that e-mails displaying empowering aspects (permission, active participation and relevancy) lead to outcomes that outperform non-empowered e-mails. By manipulating the process of tailoring, we were able to find differences in behavioral outcomes: e-mails customized based on volunteered consumer data are more effective in terms of e-mail metrics than e-mails personalized based on observed consumer data. Unlike prior research (e.g. Sahni et al., 2018;Lewis et al., 2013), the focus was on the process of tailoring rather than, for example, different executional tactics of tailoring. This perspective supports customer advocacy (Bhati and Verma, 2020) and emphasizes consumers' important role in creating engaging, empowering e-mail marketing communication.
The results indicate that activating consumers is worth the effort, as it has a significant positive effect on openings, clicks and conversions. Following the theory of psychological ownership (Kirk et al., 2015), our experiment indicated that consumers reacted more favorably when they were encouraged to contribute to the process of customization. The results are also in line with the behavioral results of Bacile et al. (2014) and support the findings in Oh and Sundar (2015), although their measures were attitudinal instead of behavioral.
The unsubscribe rate differs from the other results of this study. The recipients in the activated condition decided to unsubscribe more often than in the other two conditions. This behavior was surprising, as we expected them to be engaged in the relationship due to the sense of psychological ownership. The results were not significant but give reason to seek an explanation. First, the contents of the e-mails may have not fulfilled the needs of the recipients in the activated group. They might have expected a more customized experience, and due to disappointment, they unsubscribed. Second, privacy concerns may have caused consumers to unsubscribe at the point when they noticed their e-mails are personalized (Krafft et al., 2017), which might happen earlier when the consumer has been activated. Alternatively, these recipients were not willing to be profiled online for tailored e-mail marketing at all. Third, during the five-month period, some of the research subjects may have found another place to study or given up study plans and, therefore, unsubscribed. As the comparison groups were different in size, the activated being the smallest one, this behavior may have had a greater percentage effect in this group. These explanations remain to be studied further.

Volunteered data influencing e-mail marketing
We decided to divide the research subjects in three groups; thus, include also the control group in our experiments in order to be able to confirm the differences in outcomes as guided in Vargas et al. (2017). That turned out to be a valuable decision, as our aggregated results suggest that, in fact, personalization does not lead to better open, conversion or unsubscribe rates than in the control condition. Although personalization was not the main focus in this study, this is an interesting result that contradicts previous research findings (e.g. Sahni et al., 2018;Munz et al., 2020). However, personalization seems to increase CTOR, which is consistent with, for example, the study of Ansari and Mela (2003).
As only the process of tailoring differed in two treatments (i.e. the contents of the e-mails were similar regardless of the activated/observed treatment), we reason that tailoring itself cannot explain the differences in outcomes. OR, CTOR and conversion rate were notably higher (37.7, 73.2 and 38.2%, respectively) in activated condition compared to observed condition. Compared to the control condition, OR of the activated condition was 27.9%, CTOR was 99.4% and conversion rate was 37.2% higher, whereas in the observed condition OR was 7.6% lower, CTOR was 15.2% higher and conversion rate was 0.7% lower. Furthermore, the e-mails of each experiment were sent at the same time; thus, timing does not explain the results. As research subjects were divided in groups based on random selection, differences between the comparison groups are not the key either. Therefore, we assume that the differences are due to the activation of recipients.

Implications for practice
In 2020, over four billion people around the world used e-mail, sending approximately 306 billion e-mails daily (Tankovska, 2021). In total, 75% of marketers use e-mail when contacting customers (Data and Marketing Association, 2020). Taken the volume of e-mail marketing, an increase in response rates could have a significant influence on the returns of marketers. It is noteworthy that in the present study, there was 37.2-38.2% improvement in the conversion rate of the activated condition compared to the two other conditions. Even though consumer activation was worthwhile in this study, collecting data from consumers may require a balance and it might be wise to ask preferences incrementally as suggested in Britt (2020). On the other hand, even very surprising issues can become relevant to the consumer or seemingly relevant issues can be non-relevant to the consumer. Constant dialog would ensure that the volunteered data keep up to date. Previous research has found that continued engagement demands marketers to prioritize managing interactions with consumers (Kemp et al., 2021). We advise marketers to take this aspect into account and plan a process as well as a suitable technological solution to give consumers this possibility.
Tightening privacy regulation may prohibit using cookie data for marketing in the future (Yun et al., 2020). With this trend, voluntarily given data become even more important, as it is explicitly given for communication purposes (Britt, 2020). However, marketers need to pay attention to fostering privacy and safe data handling. It is advisable to avoid collecting data that are not needed and make consumers feel it is worthwhile to share their data in exchange for more relevant e-mails. It might be wise to inform the recipients why they get certain emails and why and how their personal data has been used to tailor the e-mails. Furthermore, consumers want their consent to apply to only in the given context and expect marketers not to pass their data to other organizations (Dev et al., 2020).

Limitations and directions for future research
This study has several limitations. First, the study context may affect the results. The experiments were conducted among applicants of a European university. It is not certain whether the results apply to a different culture, digitally less talented research subjects or other (e.g. low-involvement) products. In this study, the individuals were highly motivated to have a relationship with the university because of the high demand for entry. Instead, lowinvolvement products are often cheap and purchases impulsive. Research has shown that perceived registration effort decreases consumers' attitude toward permission basedmarketing (Bhatia, 2020;Krafft et al., 2017). It is possible that consumers would not feel the increased effort of giving volunteered data worthwhile when the question is about lowinvolvement products. More research is needed in different settings to generalize the results.
Second, although the study utilized randomized experiments, we must acknowledge that differences among recipients may have affected the results. For example, activation might be more important for the recipients who received e-mails tailored with business content than for the other subgroups. These aspects are beyond our research but might be worth studying.
Third, although controlled experiments provide a reliable methodology to detect causal relationships (Vargas et al., 2017) and our data indicated that activation increased response, the mechanism behind this should be studied in more detail. If complemented by attitudinal data through a survey or interview, one could get more information about why activation yields to better results or whether attitudes correlate with behavioral results. For example, differences in information-seeking or self-identity motivations of consumers may be the underlying mechanism (Qin, 2020) in addition to the psychological ownership.
Fourth, we examined tailoring only based on the field of study and the type of content the applicants were interested in. There are several other, possibly more relevant, preferences to customize and to be studied. Furthermore, based on our results, it is not clear if the recipient wants that every single text and image are customized based on their preferences or just some of them. Notwithstanding these limitations, this article provides evidence that customization rather than personalization pays off. We highly recommend e-mail marketers to implement tailoring with consumers not just for consumers.