Social marketing (SM) campaigns promoting testing for sexually transmitted and blood-borne infections (STBBIs) can effectively increase the uptake of testing [1,2]. However, SM testing campaigns are rarely evaluated rigorously owing to pressures of real-world implementation (eg, evaluation budget and difficulty determining campaign-specific effects in an exposed population) [1,3]. Web-based elements of SM campaigns are often evaluated through monitoring the number of views (impressions) of Web-based ads and comparing the proportions of individuals clicking through to visit (click-through rate) and use a website or service (conversion rate) [4,5]; this information is used to identify promotional venues with higher yield, allowing redirection of efforts to optimize campaign reach and inform future campaigns [1].

More robust evaluations of SM campaigns are possible for campaigns promoting internet-based health services, where users are tracked through service progression. If designed appropriately, campaign evaluations can follow individuals from their initial campaign ad view through to their program outcomes, permitting an assessment of yield of different venues based on actual health outcomes. This paper aims to describe the results of using such a design to evaluate the impact of an SM campaign on increasing the uptake of GetCheckedOnline.com (GCO) [6], an internet-based STBBI testing service in British Columbia (BC), Canada.

Overall, 177 individuals created a GCO account because of the campaign, and 22.0% (39/177) of these completed testing; all results were negative (Table 1). The highest number of accounts was from individuals viewing campaign images on geosocial apps (83/177, 46.9%) followed by individuals requesting an invitation from the campaign website (52/177, 29.4%), location ads (21/177, 11.9%), and a news website (20/177, 11.3%). The completion of testing showed little variation across venues, except the news website (1/20, 5%). We spent Can $20,801 on promotion; the average cost was Can $118 per account created and Can $533 per test (Table 2). The costliest approach per account created was geosocial apps (Can $211), followed by Web-based news (Can $105) and location ads (Can $53). Web-based news had the highest cost per test (Can $2104). Over 19 million impressions of the JMS campaign occurred through geosocial apps, with the highest click-through rate on Grindr (0.7%). The highest numbers of accounts were created from ads on Manhunt and Squirt, resulting in low costs per account created (Can $83 and Can $213, respectively).

While commonly applied in e-commerce (eg, linking ad exposures to Web-based purchases), this study demonstrated the value of using this evaluation method to understand the effects of campaign ads (in a range of promotional venues) on internet-based health services. For example, just under half of all GCO accounts were created as a result of ads on geosocial apps, where we spent the bulk of our promotional budget. While Grindr had the highest click-through rate and would, therefore, typically be considered a successful promotional venue, Grindr had the lowest proportion of GCO accounts created and a higher cost per account and per test. Conversely, Manhunt and Squirt had lower click-through rates but had higher yield in terms of GCO program outcomes. In addition, we observed the highest account creation rate among individuals exposed to our Web-based news advertisement, although a much lower proportion proceeded to test. These differences in outcomes might be explained by several factors, including the characteristics of GBMSM on these different websites and apps, such as differences in response to the JMS campaign (influencing click-through rates); demographic factors (eg, age and ethnicity); and behavioral risk or testing barriers (influencing account creation and testing rates)—all aspects worthy of further study [12,13]. Furthermore, our findings demonstrate that the promotion in physical venues is important and cost-effective, as location ads had the lowest cost per account of all venues. However, we were unable to account for view through conversion, where GBMSM seeing campaign ads may have later requested a GCO invitation on the campaign website (29% of all accounts created).

We did not observe a large uptake in testing as a result of the JMS campaign. The 39 individuals testing through GCO may be “early adopters” of this intervention with the ongoing diffusion of this innovation through GBMSM networks [14]. A shift in the message may also be needed. The feedback from GBMSM and providers following the campaign suggested convenience may not be the best selling point, given the relative availability of STBBI testing services for GBMSM in the Vancouver area (Edwards J, personal communication, November 2016); this may explain why only 1 in 5 men creating accounts tested through GCO, a measure associated with motivation to get tested in our prior evaluations [15].

In conclusion, this study demonstrates the value of developing mechanisms for tracking individuals from their Web-based exposure to SM campaign ads about an internet-based health service to their program outcomes. In addition, this study reveals that Web-based venues with high click-through rates may not always have a high conversion to service use, which is ultimately the desired outcome of SM campaigns. We are continuing to use venue-specific access codes to evaluate promotional efforts as GCO expands to other communities across BC.