Reaching the unreachable: Using targeted online advertising to promote STI screening among sex workers in Israel

How to cite

Fritz, Manuela; Grimm, Michael; Yom-Tov, Elad. 2025. "Reaching the unreachable: Using targeted online advertising to promote STI screening among sex workers in Israel." AEA RCT Registry. AEARCTR-0016317. https://doi.org/10.1257/rct.16317-2.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Manuela Fritz* , Michael Grimm , Elad Yom-Tov

* Primary investigator

Registry ID
AEARCTR-0016317
DOI
https://doi.org/10.1257/rct.16317-2.0
First registered
Last updated
Status
In development
Trial dates
2025-07-01 – 2026-04-30
Countries
Israel ()
JEL codes
I12, I14, I18
Registry record
http://www.socialscienceregistry.org/trials/16317

Abstract

Sex workers represent one of the highest-risk population for sexually transmitted infections (STIs), yet they remain underserved by conventional health systems. While STI prevalence remains relatively low in the general populations of most high-income countries, sex workers face significantly higher exposure to both viral and bacterial infections. Despite the importance of early diagnosis and treatment, structural and social barriers, including stigma, legal precarity, and lack of health insurance, limit access to preventive care and routine screening. This study evaluates the effectiveness of a keyword-based online advertising campaign aimed at increasing engagement with free and anonymous STI screening services among sex workers in Israel. The campaign targets individuals actively searching for STI-related symptoms or sexual health information, using algorithm-driven ad delivery to guide them toward testing services within a local clinic. We assess the campaign’s effectiveness in reaching the target population and promoting actual service uptake, using metrics such as click-through rates, appointment scheduling, and confirmed clinic visits. Beyond measuring reach and appointment scheduling, the study tests which types of messaging, i.e. framing mechanisms such as appeals to present bias, social support, competitiveness, or alert, are most effective in encouraging testing behavior. We compare two methods of linking online ad exposure to subsequent health behavior: (1) conversions through online appointment scheduling, and (2) offline, “real-world” conversions, by linking actual clinic visits back to the corresponding ads. Finally, a cost-effectiveness analysis will estimate the cost per additional STI screening generated by the campaign. The study aims to generate evidence on how digital advertising tools, particularly those driven by AI and behavioral insights, can be leveraged to improve public health access for hard-to-reach and stigmatized populations.

Intervention

Individuals searching for specifc keywords linking to STIs on Google linking, will be shown one out of several ads that promote screening for STIs at a local clinic. The intervention consists of showing different ads that link to four different messaging channels: present-bias/impatience, social support, competitiveness, and shock/alert. For each channel, we design several different ads. Once an individual’s search query on Google matches any of the defined keywords, they will be randomly shown one out of the different ads (randomizing across the four messaging channels).

Experimental design

We design and evaluate a targeted online advertising campaign as a means to increase engagement with STI screening services among sex workers in Israel. The campaign utilizes keyword-based targeting to display health-related advertisements to individuals actively searching online for information about STI symptoms, testing options or sex-work related keywords. These ads guide users to accessible, anonymous screening services. We assess whether this digital outreach strategy is a cost-effective method to enhance the uptake of STI screening among the target population.

We develop and design the keyword-based advertising campaign organized around three thematic keyword families (medical care in relation to STIs, sex work, STI symptoms) and four messaging channels (present-bias/impatience, social support, competitiveness, and shock/alert). For each channel, we design several different ads. Once an individual’s search query on Google matches any of the defined keywords, they will be randomly shown one out of the different ads (randomizing across the four messaging channels). When clicking on these ads, individuals will be redirected to the clinic’s website, on which they can schedule an appointment for STI testing.

Design details

Randomization method
Randomization done by Google Ads A/B split test
Randomization unit
Individual
Clusters
There is no pre-specified number of clusters (individuals). We aim to reach as many individuals as possible with the campaign.
Observations
There is no pre-specified number of observations/individuals. We aim to reach as many individuals as possible with the campaign.
Arms
There is no pre-specified sample size. We aim to reach as many individuals as possible with the campaign.
Primary outcomes
To measure the success of the campaign, more specifically, of the different ads, we focus on the following outcomes: 1) The number of individuals who have been exposed to our ads making an appointment for a STI test or a visit to the STI clinic (measured via the designated clinic website) 2) The number of individuals that have been exposed to our ads undergoing a STI test (measured in the clinic and linking back individuals to the respective ad they saw) 3) Number of individuals who click on our ads (click throughs) We will compare the performance and cost of each ad variant to identify the most effective messaging strategy for driving cost-effective uptake of STI screening services.

Cited by

No citation data available for this registration yet.

Keywords