Implications and Lessons Learned While Using Social Media Advertisements to Promote Longitudinal Social Network Study Participation in Latino Men Who Have Sex with Men (LMSM): A Brief Report

Abstract

We assess the effectiveness of paid ads on social media platforms as a research recruitment tool with Latino men who have sex with men (LMSM). We deployed four paid ad campaigns July–September 2022 in English and Spanish on Meta and Grindr featuring happy or risqué images of LMSM, documenting engagement and cost metrics. The four campaigns generated a total of 1,893,738 impressions and 1078 clicks (0.057 click-through rate) with a total cost of $7,989.39. Of the 58 people who accessed the study screener, 31 completed it (53.4%), 13 were eligible (22.4%), but none enrolled. Comparing platforms, Meta had higher engagement metrics than Grindr, while Grindr had higher proportions of those who completed the screener (57.9%) and were eligible (26.3%) than Meta (52.6% and 21.0%, respectively). Challenges to using paid ads as an LMSM recruitment tool included intersecting pandemics (Mpox, COVID-19), and limited connection between platforms and staff for study enrollment.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11524-024-00848-w.

Introduction

Enrolling LMSM in HIV prevention research is essential for understanding this priority population’s specific needs. Scholars have labeled LMSM as “hard-to-reach” [1] and in-person recruitment strategies have been the primary method for enrolling LMSM in health research [2–4]. However, pandemics severely impact such in-person strategies [5]. Technology-based strategies such as advertisements (“ads”) on social media platforms (“platforms”) [2, 6–8] may increase LMSM recruitment. A scoping review found social media was the most effective recruitment method for “hard-to-reach” populations [1]. Grov et al. [7] compared ad images to recruit MSM, transmen, and transwomen and found that an ad with a body (torso) and face was the most cost-effective in enrolling participants and engaged the greatest proportion of people of color, compared to images with a body and no face, or text only. However, there was no comparison by platform. Martinez et al. [8] did not purchase ads but compared the recruitment reach of various online/app-based platforms by posting and direct messaging study flyers on their study-specific profiles to recruit LMSM couples. Evaluating the reach of paid ads across platforms may help optimize the recruitment of LMSM in research.

The current study assesses the effectiveness of recruiting LMSM living in San Diego County into a longitudinal HIV prevention social network study using paid ads with diverse images (i.e., happy vs risqué bodies with faces) across two platforms, Meta and Grindr. Specifically, we hypothesized that there would be no difference in engagement by Spanish vs. English language, but the risqué images would generate greater interaction compared to happy images across both platforms.

Methods

The parent study began enrolling LMSM on 05/24/2021 leveraging the reference model from Francis et al., [9] using traditional in-reach (e.g., recruiting LMSM seeking services through community partners) and out-reach recruitment strategies (e.g., tabling at local gay/Latino events, flyers). Outreach strategies were limited during state-mandated COVID-19 “shelter-in-place” policies. The parent study’s eligibility criteria include (1) 18 + years of age, (2) cisgender man, (3) Latino/Hispanic ethnicity, (4) sexual activity with another man in the past 6 months, (5) HIV negative (confirmed through HIV test), (6) resides in San Diego County with no plans to move during the study period.

To supplement the parent study’s traditional recruitment strategies, the current study deployed paid ads on Meta (Facebook and Instagram) and Grindr. In preparing ads, the research team, inclusive of LMSM, selected images of Latino men as previous research found that people better respond to ads featuring people who look like them, [10] and that images outperformed text alone [7]. Meta ads were 536 × 892 pixels and placed using advantage + placements which leverages Meta’s algorithm to place ads in locations that maximize ad cost effectiveness. Grindr ads were 300 × 250 pixels and placed using the Medium Banner option. Using the geo-fencing feature on both platforms, we specified that ads only be shown to users within San Diego County. On Meta, we further targeted the ads to only be shown to people who identify as men. We deployed a Happy ad image on 07/22/2022 (Grindr, 28 days) and 07/23/2022 (Meta, 27 days) until 08/17/2022 and a Risqué ad image on 08/29/2022 (Meta, 27 days) and 09/05/2022 (Grindr, 21 days) until 09/26/2022 in both English and Spanish (see Supplemental Fig. 1). Upon clicking on the ads, participants were forwarded to an online screener via Qualtrics. Hyperlinks were embedded with source information to determine which participants were recruited through ads and from which platform specifically. After screening eligible for the study, they had the option of providing their contact information so study staff could reach out to them or given staff’s contact information to initiate contact with the study. All study materials were developed in English, translated to Spanish, and back-translated to English using an iterative review process across bilingual-bicultural members of the study team, community partners, and community advisory board (CAB) [11].

We budgeted $4000 per platform ($8000 total). For each campaign, we evaluated impressions (number of people who saw the ad), days run, number of clicks, amount spent, number of participants who began the eligibility screener, and the number who completed it. From these data, we calculated the click-through rate (CTR) as the number of clicks per 100 impressions, cost per 1000 impressions (CPM), and cost per eligible screened person (CPE). We used a member-checking approach to gather insight from the study’s LMSM CAB and staff [12]. During a quarterly CAB meeting, the CAB was presented with the study methods, ads, and results and was asked to question, critique, and give feedback, creating a collaborative environment to contextualize the interpretation of the results. This method offers a check on the conceptualization and possible gaps in cultural understanding of the researchers so that the findings integrate the perspectives of potential study participants [12]. The University of California San Diego Institutional Review Board approved this study.

Results

The four campaigns generated a total of 1,893,738 impressions and 1078 clicks (0.0569 CTR) with a total cost of $7,989.39. Fifty-eight people accessed the eligibility screener and 31 people completed it (53.4%). Of the 31 people who completed the eligibility survey, 100% were assigned male sex at birth, 83.9% identified as cisgender (16.1% identified as non-binary), 76.7% identified as Latino, 16% identified their race as “other,” 63.3% were proficient in both English and Spanish, and 38.7% were aware of PrEP but not taking it. Thirteen LMSM screened eligible, a cost of $614.57 per eligible person screened. None of the thirteen eligible LMSM reached through ads were enrolled in the parent study. Nuances in the ad performance across different ad characteristics are presented below.

Platforms

We received almost four times as many impressions and almost double the CTR with Meta than through Grindr. However, a higher percentage of those who accessed their eligibility through Grindr successfully completed the screener (57.9%) and were eligible (26.3%) in comparison to Meta (52.6% and 21.0%, respectively). See Table 1 for full stratified results.

Table 1.

Relative effectiveness of ad campaigns

	Impressions	Days run	Clicks	CTRa	CPMb	Total spent	Accessed eligibility	Completed eligibility	Eligible	CPEe
Grindr	398,939	42	144	0.0361	$10.00	$3989	19	11	5	$797.80
Happy	264,485	28	113	0.0427	$10.00	$2645	16	10	5	$529.00
English	132,505	c	63	0.0475	$10.00	$1325	d	d	d
Spanish	131,980	c	50	0.0379	$10.00	$1320	d	d	d
Risqué	134,454	14	31	0.0231	$10.00	$1344	3	1	0	-
English	66,961	c	17	0.0254	$10.00	$669	d	d	d
Spanish	67,493	c	14	0.0207	$10.00	$675	d	d	d
Meta	1,494,799	48	934	0.0625	$2.69	$4000	38	20	8	$500.00
Happy	785,869	27	428	0.0545	$2.41	$1894	10	7	2	$947.00
English	384,947	c	210	0.0546	$2.44	$940	d	d	d
Spanish	400,922	c	218	0.0544	$2.38	$954	d	d	d
Risqué	708,930	21	506	0.0714	$2.97	$2106	28	13	6	$351.00
English	350,351	c	248	0.0708	$3.02	$1059	d	d	d
Spanish	358,579	c	258	0.0720	$2.92	$1047	d	d	d
Total	1,893,738		1078	0.0569		$7989	57	31	13	$614.54

^aClick through rate (clicks per 100 impressions)

^bCost per thousand impressions

^cAd campaigns were run in both English and Spanish simultaneously

^dData was not collected to assess

^eCost per eligible participant screened

Language

We received a total of 934,764 impressions and 538 clicks on the English ads and 958,974 impressions and 540 clicks on the Spanish ads. On Grindr, impressions between English and Spanish were very similar, but English outperformed Spanish on clicks. However, on Meta, Spanish received 24,203 more impressions and 18 more clicks than English.

Image

Over the combined 55 (Meta = 27; Grindr = 28) day run of the Happy ad we received 1,050,354 impressions (19,097/day) and 541 clicks (10/day). Over the combined 35 (Meta = 21; Grindr = 14) day run of the Risqué ad, we received 843,384 impressions (24,097/day) and 537 clicks (15/day). The Risqué ad on Meta yielded the highest CTR among all ads (0.0714); however, it yielded the lowest CTR among all ads on Grindr (0.0231).

Member Checking

Consultations with the study staff and CAB suggested that engagement on Grindr was lower because people were intentionally reducing their number of sexual encounters in response to the emergence of Mpox. While our study deployed our first ad campaign a week after San Diego Pride, the CAB suggested that ads may have been more successful if deployed before and during Pride as more people may have been using platforms to make event plans and meet new people.

Discussion

Despite previous research successfully using ads to recruit “hard-to-reach” populations, including LMSM [7, 13], our study did not enroll any LMSM using this method. While Meta had more impressions and higher ad interactions, among those who accessed the eligibility screener, Grindr had better screener completion and a higher proportion of those who were eligible for the study. Future studies with LMSM should consider this finding when setting their ad recruitment goals (e.g., reach vs sensitivity).

Surprisingly, the Risqué ad had the highest CTR and number of eligible potential participants on Meta, but the lowest CTR and number of eligible potential participants on Grindr. We initially hypothesized that the Risqué ad would have received greater interaction across both platforms. However, our Risqué ad on Grindr and Happy ad on Meta may have been unable to break through the noise of similar advertising content on the platforms yielding lower-than-expected engagement [14]. Similar to results related to program-advertisement congruency that found sexual program content hindered sexual advertisement recall [15], ads that are unexpected but appealing may prove as a useful tactic for engaging LMSM in different virtual contexts. Compared to previous studies [7, 13], social media ads did not lead to study enrollment; future research on how to optimize the follow-through rate with LMSM is needed. To note, the Risqué and Happy ad did have minor differences (e.g., background color and instructions on the Risqué ad indicating to click to check eligibility), which also could have contributed to different interaction patterns.

Lessons Learned

Intersecting Pandemics

During ad deployment, governmental restrictions related to COVID-19 continued to wane but community Mpox concerns were mounting. This could suggest that platforms which facilitate in-person interactions may be less effective for recruitment during times of elevated person-to-person transmission risk.

Event Planned Advertising

The research team chose to launch the ad campaign after San Diego Pride 2022 to minimize the number of impressions from ineligible MSM visiting for Pride. Future campaigns targeting LMSM should test the effectiveness of ads deployed surrounding large events that engage the MSM community.

Breaking the Virtual Barrier

Previous research recruiting MSM via ads was effective, where all study activities are typically conducted completely online with no in-person communication between participants and the study team [7, 13, 16]. Components of our paid ads study design may have served as a barrier for people to enroll in the study by not being able to speak or interact with study staff at the time they saw the ad on social media. Before deploying social media ad strategies, burdens introduced to participants by breaking the virtual barrier should be considered.

Current Recruitment Strategies

To date, the parent study has successfully recruited 400 participants in large part due to leveraging in-reach strategies with trusted community partners and interactive outreach strategies via the study’s Instagram account. In combination, these strategies help to generate trust and rapport with study staff, provide a timely response to potential participants’ questions, and grow our social media following. This strategy is similar to other successful recruitment strategies employed to recruit LMSM couples in other contexts [8].

Conclusion

Recruitment of LMSM through social media ads may present limitations for research studies. Strategies that leverage platforms to facilitate interactions with the community while recruiting participants may present a better strategy for reaching LMSM.

Supplementary Information

Below is the link to the electronic supplementary material.

Data Availability

The data underlying this study can be requested by emailing the corresponding author.