Social Advertising Catherine Tucker? February 15, 2012 Abstract In social advertising, ads are targeted based on underlying social networks and their content is tailored with information that pertains to the social relationship. This paper explores the e? ectiveness of social advertising using data from ? eld tests of di? erent ads on Facebook. We ? nd evidence that social advertising is e? ective, and that this e? cacy seems to stem mainly from the ability of targeting based on social networks to uncover similarly responsive consumers.
However, social advertising is less e? ective if the advertiser explicitly states they are trying to promote social in? uence in the text of their ad. This suggests that advertisers must avoid being overt in their attempts to exploit social networks in their advertising. Catherine Tucker is Associate Professor of Marketing at MIT Sloan School of Management, Cambridge, MA. and Faculty Research Fellow at the NBER. Thank-you to Google for ? nancial support and to an anonymous non-pro? t for their cooperation.
Thank-you to Jon Baker, Ann Kronrod, Preston Mcafee, and seminar participants at the George Mason University Roundtable on the Law and Economics of Internet Search, the University of Rochester, UCLA and Wharton for valuable comments. All errors are my own. ? 1 Electronic copy available at: http://ssrn. com/abstract=1975897 1 Introduction Recent advances on the internet have allowed consumers to interact across digital social networks. This is taking place at unprecedented levels: Facebook was the most visited website in the US in 2010, accounting for 20% of all time spent on the internet, a higher proportion than Google or Yahoo! ComScore, 2011). However, it is striking that traditional marketing communications have been at the periphery of this explosion of social data despite the documented power of social in? uence on purchasing behavior. Much of the emphasis on marketing in social media, so far, has been on the achievement of ‘earned reach,’ whereby a brand builds its subscriber base organically and also hopes that this will in? uence others organically through sharing links with their social networks (Corcoran, 2009). However, recent research by Bakshy et al. 2011) has emphasized that this kind of organic sharing is far rarer than previously supposed, and that there are very few examples of a commercial message being consistently transmitted across social networks. Further, Tucker (2011a) shows that in order to achieve virality, an advertiser may have to sacri? ce the commercial e? ectiveness of their message. This means that advertisers may need to use paid advertising to facilitate the sharing of their commercial message through social networks. Both Facebook and LinkedIn have recently introduced a new form of advertising called ‘social advertising. A social ad is an online ad that ‘incorporates user interactions that the consumer has agreed to display and be shared. The resulting ad displays these interactions along with the user’s persona (picture and/or name) within the ad content’ (IAB, 2009). This represents a radical technological development for advertisers, because it means that potentially they can co-opt the power of an individual’s social network to target advertising and engage their audience. This paper asks whether social advertising is e? ective, and what active steps advertisers themselves should take in their ads to promote social in? ence. 2 Electronic copy available at: http://ssrn. com/abstract=1975897 We explore the e? ectiveness of social ads using data from a ? eld experiment conducted on Facebook by a non-pro? t. This ? eld experiment compared the performance of social ads with conventionally targeted and untargeted ads. The social ads were targeted to the friends of ‘fans’ of the charity on Facebook. The ads featured that fan’s name and the fact that they had become a fan of this charity. We ? nd that on average these social ads were more e? ective than demographically targeted or untargeted ads.
Further, this technique is useful for improving both the performance of demographically targeted and untargeted campaigns. Comparing the performance of these ads that contained the name of the fan and were targeted towards the fan’s friends with those that were simply targeted to that fan’s friends suggests that their e? ectiveness stems predominantly from the ability of social targeting to uncover similarly responsive consumers. We present results that suggest that as well as being more e? ective at gathering clicks, social advertising is also more e? ective at promoting actual subscriptions to the newsfeed and is more cost-e? ctive. We then turn to investigate how advertisers should word their social advertising. Through randomized ? eld tests, we investigate the e? ectiveness of advertisers deliberately promoting social in? uence in their advertising copy through including a statement that encourages the viewer to, for example, ‘be like their friend. ’ We ? nd that consumers reject attempts by advertisers to explicitly harness or refer to a friend’s actions in their ad copy. This result contrasts with previous empirical research that ? nds consistent bene? ts to ? rms from highlighting previous consumer actions to positively in? ence the consumers’ response (Algesheimer et al. , 2010; Tucker and Zhang, 2011). This rejection is reasonably uniform across di? erent wording, though slightly less severe for ads that make a less explicit reference to friendship. We then present additional evidence to rule out two potential explanations for our ? ndings. First, we rule out that the overt mention of social in? uence simply made people aware they were seeing an ad rather than something organic to the site. We do this by comparing an ad that states it is an ad with an ad that does not, and ? nding no di? rence. 3 Second, to investigate whether it was simply bad advertising copy, we examined how the ads perform for a group of Facebook users who have shown a visible propensity for social in? uence. We identify such users by whether or not they have a stated attachment to a ‘Fashion Brand’ on their Facebook pro? le. These users, in contrast to our earlier results, react more positively to the advertiser explicitly co-opting social in? uence than to a message that did not. This suggests that it was not simply that the message was badly communicated, but instead re? cts a taste (or more accurately distaste) for explicit references to social in? uence among most, though not all, consumers. This research builds on a literature that has studied the interplay between social networks and word of mouth. Zubcsek and Sarvary (2011) present a theoretical model that examines the e? ects of advertising to a social network, but assume that a ? rm cannot directly use the social network for marketing purposes. Instead, ? rms have to rely on consumers to organically pass their advertising message within the social networks. There has been little work on advertising in social networks.
Previous studies in marketing about social network sites have questioned how such sites can use advertising to obtain members (Trusov et al. , 2009), and also how makers of applications designed to be used on social network sites can best advertise their products (Aral and Walker, 2011) through viral marketing. Hill et al. (2006) show that phone communications data can be used to predict who is more likely to adopt a service, Bagherjeiran et al. (2010) present a practical application where they use data from instant messaging logs at Yahoo! to improve online advertising targeting, and similarly Provost et al. 2009) show how to use browsing data to match groups of users who are socially similar. Tucker (2011b) explores how privacy controls mediate the e? ectiveness of advertising on Facebook. However, to our knowledge this is the ? rst academic study of the e? ectiveness of social advertising. Managerially, our results have important implications. Social advertising and the use of online social networks is e? ective. However, when advertisers attempt to reinforce this social 4 in? uence in ad copy, consumers appear less likely to respond positively to the ad. This is, to our knowledge, the ? st piece of empirical support for emerging managerial theories that emphasize the need for ? rms to not appear too obviously commercial when exploiting social media (Gossieaux and Moran, 2010). 5 2 Field Experiment The ? eld experiment was run by a small non-pro? t that provides educational scholarships for girls to attend high school in East Africa. Without the intervention of this non-pro? t, and other non-pro? ts like them, girls do not attend secondary school because their families prioritize the education of sons. Though the non-pro? t’s main mission is funding these educational scholarships, the non-pro? has a secondary mission which is to inform young people in the US about the state of education for African girls. It was in aid of this secondary mission that the non-pro? t set up a Facebook page. This page serves as a repository of interviews with girls where they describe the challenges they have faced. To launch the ? eld experiment, the non-pro? t followed the procedure described in ‘A/B Testing your Facebook Ads: Getting better results through experimentation’ (Facebook, 2010) which involved setting up multiple competing campaigns. These ad campaigns was targeted to three di? erent groups as shown in Table 1. The ? st group was a broad untargeted campaign for all Facebook users aged 18 and older in the US. The second group were people who had already expressed interest in other charities. These people were identi? ed using Facebook’s ‘broad category targeting’ of ‘Charity + Causes. ’ The third group were people who had already expressed an interest in ‘Education + Teaching. ’ Previously, the charity had tried such reasonably broad targeting with little success and was hopeful that social advertising would improve the ads’ performance (Tucker, 2011b). In all cases, the charity explicitly excluded current fans from seeing its ads.
For each of these groups of Facebook users, the non-pro? t launched a socially targeted variant. These ads employed the Facebook ad option that meant that they were targeted only to users who were friends of existing fans of the charity. This also meant that when the fan had not opted-out on Facebook, the ad also displayed a ‘social endorsement’ where the name of the friend was shown at the bottom of the ad as shown in Figure 1. 6 Table 1: Di? erent Groups Targeted Condition Untargeted Baseline: Only Shown Baseline text All people in US over age of 18 who are not fans of the non-pro? t already.
All people in US over age of 18 who state a? nity with charities on their Facebook pro? le who are not fans of the non-pro? t already. All people in US over age of 18 who state a? nity with education on their Facebook pro? le who are not fans of the non-pro? t already. Social Variant: Shown all 5 texts from Table 2 All people in US over age of 18 who are friends of the non-pro? t’s supporters who are not fans of the non-pro? t already. All people in US over age of 18 who state a? nity with charities on their Facebook pro? le who are friends of the non-pro? t’s supporters who are not fans of the nonpro? already. All people in US over age of 18 who state a? nity with education on their Facebook pro? le who are friends of the non-pro? t’s supporters who are not fans of the nonpro? t already. Charity Education The non-pro? t varied whether the campaign was demographically targeted and whether the campaign was socially targeted, and also explored di? erent ad-text conditions. Table 2 describes the di? erent ad-copy for each condition. Each di? erent type of ad-copy was accompanied by the same picture of an appealing secondary-school student who had bene? ted from their program.
The socially targeted ads displayed all ? ve variants of the advertising message depicted in Table 2. For each of the non-socially-targeted campaigns, we ran the baseline variant of the ad text which, as shown in Table 2, simply says ‘Help girls in East Africa change their lives through education. ’ The non-pro? t could not run the other four conditions that refer to others’ actions, because federal regulations require ads to be truthful and they did not want to mislead potential supporters. The di? erent ad conditions were broadly designed to cover the kinds of normative and informational social in? ence described by Deutsch and Gerard (1955); Burnkrant and Cousineau (1975). 1 We want to be clear that we do not argue that these advertising measures 1 Other forms of social in? uence studied in the literature involve network externalities where there is a performance bene? t to multiple people adopting (Tucker, 2008). However, that does not seem to be relevant 7 Table 2: Di? erent Ad-Text Conditions Condition Baseline Be like your friend Ad-Text Help girls in East Africa change their lives through education. Be like your friend.
Help girls in East Africa change their lives through education. Don’t be left out. Help girls in East Africa change their lives through education. Your friend knows this is a good cause. Help girls in East Africa change their lives through education. Learn from your friend. Help girls in East Africa change their lives through education. Don’t be left out. Your friend knows Learn from your friend. capture all types of social in? uence or are necessarily successful at distinguishing between the di? erent types of social in? uence that are possible. The literature on social in? ence has emphasized that the underlying mechanism is nuanced and complex. Obviously, di? erent types of social in? uence relate and interact in ways that cannot be teased apart simply with di? erent wording. However, the variation in messages does allow us to study whether explicit advertising messages that attempt to use di? erent types of wording to evoke social in? uence are e? ective in general. Figure 1: Sample Ad Figure 1 displays an anonymized sample ad for a social ad in the ‘be like your friend’ condition. The blacked-out top of the ad contained the non-pro? t’s name. The grayedhere. out bottom of the ad contained a supporter’s name, who had ‘liked’ the charity and was a Facebook friend of the person who was being advertised to. It is only with developments in technology and the development of automated algorithms that such individualized display of the friend’s name when pertinent is possible. Table 3 describes the demographics of the roughly 1,500 fans at the beginning of the campaign. Though the initial fans were reasonably spread out across di? erent age cohorts, they were more female than the average population, which makes sense given the nature of the charity.
At the end of the experiment, the fans were slightly more likely to be male than before. The way that Facebook reports data means that we have access to the demographics only of the fans of the charity, not of those who were advertised to. Table 3: Demographics of the non-pro? t’s fans before and after the ? eld experiment Age 18-24 25-34 35-44 45-54 55+ Total Before Male 5 5 6 3 3 22 Experiment After Experiment Female Male Female 13 8 14 14 6 14 17 6 16 13 3 13 10 4 10 67 27 67 The ‘Total’ row does not add up to 100% because fans who are below 18 years of age are omitted. 9 3 Data
The data that Facebook shares with advertisers is both anonymous and aggregate. This means that we cannot trace the e? ects of social advertising on the friends of any one individual. It also means that we cannot examine heterogeneity in the degrees of in? uence across individuals, as is studied, for example, by Godes and Mayzlin (2009) in their study of o? ine ? rm-sponsored communications. However, given that the central research question of the study is whether, on average, di? erent types of social advertising are more e? ective, the aggregate nature of the data is su? cient.
Table 4 reports daily summary statistics for the campaigns in our data. Over a 5-week period, there were 630 observations. There were 18 campaigns in total that consisted of a) The three baseline conditions that were demographically targeted to everyone, charity-lovers and education-supporters and used the baseline text, and b) The ? fteen social ad conditions that had all the ? ve di? erent types of text, and socially targeted separately to everyone, charity-lovers and education-supporters. Table A2 in the appendix provides a summary of these campaigns. Table 4: Summary Statistics Mean Std Dev Min Max Average Impressions 13815. 13898. 6 1 98037 Average Clicks 5. 06 5. 17 0 37 Connections 2. 70 3. 52 0 24 Unique Clicks 5. 04 5. 14 0 36 Daily Click Rate 0. 11 0. 10 0 1. 27 Impression Click Rate 0. 045 0. 047 0 0. 50 Cost Per Click (USD) 0. 98 0. 40 0. 31 3. 90 Cost Per 1000 views (USD) 0. 52 1. 37 0 24. 5 Ad-Reach 6165. 7 6185. 0 1 60981 Frequency 2. 32 0. 82 1 9. 70 18 ad variants at the daily level for 5 weeks (630 observations) There are two click-through rates reported in Table 4. The ? rst click-through rate is the proportion of people who clicked on an ad that day. The denominator here is the 10
Ad-Reach measure that captures the number of people exposed to an ad each day. The second click-through rate is per ad impression. We focus on the former in our econometric analysis, because impressions can be a function of person refreshing their page or using the back button on the browser or other actions which do not necessarily lead to increased exposure to the ad. We show robustness subsequently to using this click-through rate per impression measure. Due to the relatively small number of clicks, these click through rates are expressed as percentage points or sometimes as fractions of a percentage point.
In our regression analysis we also use this scaling in order to make our coe? cients more easily readable. 2 The data also contains an alternative means of measuring advertising success. The connection rate measures the number of people who liked a Facebook page within 24 hours of seeing a sponsored ad, where the denominator is the ad’s reach that day. We compare this measure to clicks in subsequent analysis to check that the click-through rate is capturing something meaningful. We also use the cost data about how much the advertiser paid for each of these ads in a robustness check.
The data reassuringly suggests that there were only ? ve occasions where someone clicked twice on the ads. Therefore, 99. 8% of the click-through rate we measure captures a single individual clicking on the ad. 2 11 Figure 2: Social advertising is e? ective 4 4. 1 Results Does Social Advertising Work? First, we present some simple evidence about whether social advertising is more e? ective than regular display advertising. Figure 2 displays the basic comparison of aggregate (that is, across the whole ? ve-week period) click-through rates between non-socially-targeted ads and ads that were socially targeted.
Since these are aggregate click-through rates they di? er from the daily click-through rates reported in Table 4. These are expressed as fractions of a percentage point. It is clear that social advertising earned far larger click-through rates. The di? erence between the two bars is quite striking. To check the robustness and statistical signi? cance of this relationship, we turn to econometrics. The econometric analysis is relatively straightforward because of the randomization induced by the ? eld tests. We model the click-through rate of campaign j on day t targeted to demographic group k as: 2 ClickRatejt = ? SocialT argeting Endorsementj + ? k + ? t + j (1) SocialT argeting Endorsementj is an indicator for whether or not this campaign variance was socially targeted and displayed the endorsement. Since Facebook does not allow the testing of these di? erent features separately, this is a combined (rather than separable) indicator. ?k is a ? xed e? ect that captures whether this was the untargeted variant of the ad. This controls for underlying systematic di? erences in how likely people within that target and untargeted segment were to respond to this charity.
We include a vector of date dummies ? t . Because the ads are randomized, ? t and ? k should primarily improve e? ciency. We estimate the speci? cation using ordinary least squares. Though we recognize that theoretically a click-through rate is bounded at one hundred since it is measured in percentage points, click-through rates in our data are never close to this upper bound or lower bound. 3 Table 5 reports our initial results. Column (1) presents results for the simple speci? cation implied by equation (1) but without the date and demographic controls.
The point estimates suggest that social targeting and a friend’s endorsement increased the average daily clickthrough rate by around half. Column (2) repeats the analysis with the controls for date. It suggests that after controlling for date, the result holds. This is reassuring and suggests that any unevenness in how ads were served across days does not drive our results. It also suggests that our result is not an artifact of a failure of randomization. Column (3) adds an extra coe? cient that indicates whether that campaign was untargeted rather than being targeted to one of the customer groups identi? d as being likely ‘targets’ by the non-pro? t We also tried alternative speci? cations where we use the unbounded clicks measure (rather than a rate) as the dependent variable and show that our results are robust to such a speci? cation in Table A1, in the appendix. 3 13 – Educational and Charity supporters. It suggests that indeed, as expected, an untargeted campaign was weakly ine? ective, though the estimate is not signi? cant at conventional levels. We speculate that the apparent weakness of demographic targeting may be because target markets of charity and educational supporters is reasonably broad, and consequently may have ontained many individuals who would not support an international charity. An obvious question is what explains the success of social advertising. One explanation is that the endorsement of a friend is informative. Another explanation is that social targeting uncovers people who will be more likely to be interested in their charity as they are similar, in unobserved ways, to their friends who are already fans of the charity. Manski (1993) pointed out that this particular issue of distinguishing homophily (unobserved characteristics that make friends behave in a similar way) from the explicit in? ence of friends on each other is empirically problematic. Ideally, to address this we would simply randomize whether users saw the endorsement or not. However, Facebook’s advertiser interface does not allow that. What we can do is take advantage of the fact that sometimes ads are shown to people without the endorsement if that fan has selected a privacy setting which restricts the use of their image and name. The interface which users use to do this is displayed in Figure A1; all users do is simply select the ‘No One’ rather than the ‘Only my friends’ option.
Of course, this will not represent perfect randomization. It is likely that the fans who select stricter privacy settings di? er in unobserved ways from those who do not, and that therefore their social networks may di? er as well. However, despite this potential for bias, this does represent a useful opportunity to try to disentangle the power of social targeting to enable homophily and the power of personal endorsements. Column (4) displays the results of a speci? cation for equation (1) where the dependent variable is the conversion rate for these socially targeted but not socially endorsed ads.
Here for ads that were being shown to friends, the click-through rate was only calculated for occasions when the endorsement was not shown. A comparison of Column 14 (3) and Column (4) in Table 5 makes it clear the ads that were displayed to friends of fans but lacked a clear endorsement were less e? ective than those that had a clear endorsement. However, they were still measurably more e? ective than non-socially-targeted ads. It appears that, roughly, the endorsement accounted for less than half of the persuasive e? ect and the ability to use social networks to target the ad accounted for slightly more than half of such ads’ e? acy. Columns (5) and (6) of Table 5 estimate the speci? cation separately by whether the campaign was targeted or untargeted. Though the point estimate for the targeted campaigns is higher, it is notable that social advertising improved the performance of both targeted and untargeted campaigns. Given the widely reported lack of e? cacy of untargeted campaigns (Reiley and Lewis, 2009), the increase in e? ectiveness allowed by social advertising appears large for untargeted campaigns. 15 Table 5: Social Targeting and Endorsement is E? ective (4) No Endorsement Click Rate SocialTargeting Endorsement
All (1) Click Rate 0. 0386??? (0. 0123) (2) Click Rate 0. 0385??? (0. 0108) 0. 0287?? (0. 0143) -0. 000275 (0. 0122) 0. 0794??? (0. 0116) 0. 0132 (0. 0166) (3) Click Rate 0. 0386??? (0. 0125) Untargeted (5) Click Rate 0. 0297??? (0. 00755) Targeted (6) Click Rate 0. 0376??? (0. 00927) SocialTargeting Untargeted Constant 16 Date Controls No Yes Yes Yes Yes Yes Observations 630 630 630 630 210 420 Log-Likelihood 542. 1 610. 3 610. 3 427. 8 187. 7 452. 3 R-Squared 0. 0221 0. 212 0. 212 0. 119 0. 317 0. 228 OLS Estimates. Dependent variable is the percentage point of people who click on the ad.
Dependent variable in Columns (4) for social ads is the percentage point daily click-through rate of ads that did not display the endorsement. Robust standard errors. * p < 0. 10, ** p < 0. 05, *** p < 0. 01 4. 2 Robustness Table 6 checks the robustness of the ? nding that social targeting and endorsement are effective, to di? erent de? nitions of the dependent variable. Column (1) reports the results of using a dependent measure which is the percentage click-through per impression. Again, we ? nd that social advertising is more e? ective, though the e? ectiveness is less pronounced and less precisely estimated than before.
This suggests that the appeal of social advertising is not necessarily enhanced by multiple exposure. It could also, of course, merely re? ect noise introduced into the process by someone refreshing their browser multiple times. The results so far suggest that consumer privacy concerns or the intrusiveness of such ads do not seem to outweigh the appeal of social advertising for consumers. 4 There is always the possibility of course that people clicked on the ads because they were annoyed or wanted to understand more the extent of privacy intrusion rather than because the ads were actually e? ective.
To explore this, we estimate a speci? cation where the dependent measure was the proportion of clicks that became subscribers of the newsfeed. The results are reported in Column (2). We see that again social advertising appears to be more e? ective at encouraging Facebook users to take the intended action as well as simply clicking. This is evidence that people are not clicking on social ads due to annoyance at their intrusiveness but instead are clicking on them and taking the action the ads intend to encourage them to take. Untargeted ads are less likely to lead to conversions than those targeted at appropriate demographics.
This makes sense – these people are being targeted precisely because they are the kind of people who have signed up for such news feeds in the past. A ? nal question is whether ads that are socially targeted and display endorsements are more expensive for advertisers, thereby wiping out their relative e? ectiveness in terms of return on advertising investment. We explore this in Column (3) of Table 6. There are This may be because Facebook users ? nd it reassuring that these ads, though narrowly targeted, are not overly visually intrusive (Goldfarb and Tucker, 2011). 4 17 everal missing observations where there were no clicks that day and consequently there was no price recorded. In Column (3), we report the results of a speci? cation where our explanatory variables is the relative price per click. The results suggest that advertisers pay less for these clicks that are socially targeted. This suggests that Facebook is not charging a premium for this kind of advertising. Though Facebook shrouds in secrecy the precise pricing and auction mechanism underlying their advertising pricing, this result would be consistent with a mechanism whereby advertisers pay less for clicks if they have higher clickthrough rates.
In other words, prices paid bene? t from an improved ‘quality-score’ (Athey and Nekipelov, 2011). The results also suggest that advertisers pay less for demographically untargeted clicks which is in line with previous studies such as Beales (2010). Table 6: Social Advertising is E? ective: Checking robustness to di? erent dependent variables SocialTargeting Endorsement (1) Click Rate (Multiple) 0. 0108?? (0. 00501) 0. 00526 (0. 00582) Yes 630 1086. 5 0. 150 (2) Clicks to Connections Rate 0. 433??? (0. 0997) -0. 321??? (0. 0768) Yes 554 -467. 5 0. 163 (3) Cost Per Click (USD) -0. 95??? (0. 0480) -0. 177??? (0. 0520) Yes 559 -129. 0 0. 426 Untargeted Date Controls Observations Log-Likelihood R-Squared OLS Estimates. Dependent variable is the click-through rate (expressed as a fraction of a percentage point) for impressions in Column (1). Dependent variable in Column (2) is the clicks to conversions rate. Dependent variable in Column (3) is cost per click. Robust standard errors. * p < 0. 10, ** p < 0. 05, *** p < 0. 01 4. 3 What Kind of Social Advertising Messages Work? We then go on to explore what kind of advertising message works in social ads.
We distinguish between ads that rely simply on the Facebook algorithm to promote social in? uence by featuring the automated endorsement at the bottom of their ad, and ads that explicitly refer to this endorsement in their ad copy. 18 Table 7: Social Advertising is Less E? ective if an Advertiser is Too Explicit (3) No Endorsement Click Rate SocialTargeting Endorsement All (1) Click Rate 0. 0577??? (0. 0139) (2) Click Rate 0. 0571??? (0. 0113) 0. 0333?? (0. 0168) -0. 0287??? (0. 00886) -0. 000463 (0. 0122) -0. 0136 (0. 0115) -0. 0189? (0. 01000) -0. 0378??? (0. 0115) -0. 0429??? (0. 0144) -0. 101 (0. 0124) Yes 630 615. 4 0. 225 Yes 630 618. 1 0. 232 Yes 630 429. 5 0. 124 Yes 210 189. 6 0. 329 Yes 420 461. 0 0. 260 -0. 000281 (0. 0177) 0. 0161 (0. 0169) -0. 0303? (0. 0167) -0. 0284?? (0. 0124) Untargeted (4) Click Rate 0. 0498?? (0. 0245) Targeted (5) Click Rate 0. 0527??? (0. 0130) SocialTargeting SocialTargeting Endorsement ? Explicit Untargeted SocialTargeting Endorsement ? Don’t be left out SocialTargeting Endorsement ? Be like your friend SocialTargeting Endorsement ? Learn from your friend 19 SocialTargeting Endorsement ? Your friend knows SocialTargeting ? Explicit
Date Controls Observations Log-Likelihood R-Squared OLS Estimates. Dependent variable is the percentage points of people who click on the ad. Dependent variable in Columns (3) adjusted for social ads so that is the percentage point daily click-through rate of ads that did not display the endorsement. Robust standard errors. * p < 0. 10, ** p < 0. 05, *** p < 0. 01 We use the additional binary indicator variable Explicitj to indicate when the advertiser uses a message that evokes social in? uence explicitly in their ad copy, in addition to the social endorsement automated by the Facebook algorithm.
This covers all the non-baseline conditions described in Table 2. We interact this with the SocialT argeting Endorsementj , meaning that SocialT argeting Endorsementj now measures the e? ect of the baseline effect, and the interacted variable measures the incremental advantage or disadvantage of mentioning the friend or the potential for social in? uence in the ad. Column (1) of Table 7 reports the results. The negative coe? cient on the interaction between Explicit and SocialT argeting Endorsementj suggests that explicit reference to a social in? uence mechanism in the ad a? ected the performance of the ad negatively.
That is, when the advertiser themselves were explicit about their intention to harness social in? uence, it back? res. Further, the large point estimate for SocialT argeting Endorsementj suggests that the base
Do you similar assignment and would want someone to complete it for you? Click on the ORDER NOW option to get instant services at essayloop.com