Geo-targeting is a common tool for targeting information to a specific geographical area and is widely used in digital marketing, which uses geolocation, in combination with information gleaned from available user data about consumer preferences, to deliver location-specific content. It is a costeffective way to match information with consumers.
We set out to examine geo-targeting of information in the context of the 2019 Finnish parliamentary and Indonesian presidential elections. Finland and Indonesia are socioeconomically very different and their internet penetration and adaptation of technology also vary significantly. However, both faced an important election, thus, are good targets for analysis.
This report examines information related to elections presented in certain geographical locations across four platforms (Google, YouTube, Instagram and Facebook), and aims to unveil any targeted content. Further analysis of the information and sources presented is done using qualitative and quantitative means. The following section introduces the case studies of the 2019 Finnish parliamentary election and Indonesian presidential election, as well as the methodology used to collect data and obtain results. The third section describes the results we obtained for both countries on each platform. Conclusions and recommendations are presented in the final section.
Conclusions and Recommendations
Through our research in Indonesia and Finland, we aimed to answer two questions about the role of geo-targeting during election periods.
First, did election-related geo-targeting take place in Indonesia and Finland? And if so, did geo-targeting pose a threat to the democratic process in either country? In order to distinguish between normal and geo-targeted posts, we hypothesised that geo-targeted content had to fulfil two of three criteria. They had to be:
1) posts appearing in just one region and
2) appear at least twice in that region or
3) appear as a paid-for ad in that region.
Using aggregated Facebook, Instagram, YouTube and Google data from multiple regions in both countries, we applied these criteria and identified a number of geo-targeted posts on all social media platforms in our monitored regions of Indonesia and Finland.
In order to address the question of whether geo-targeting posed a threat to the elections in Indonesia and Finland, we qualitatively analysed the geo-targeted posts for the substance and purpose of their content.
In Indonesia, we concluded that most top trending content provided apolitical and educational information to their audiences.
In contrast, we noted that a high number of geo-targeted posts shared articles and videos relating to the overseas voting scandal in Malaysia, suggesting that there was potentially an attempt to undermine trust in democratic institutions. However, media coverage of the scandal was widespread among other regions as well, which received higher interaction rates than those articles targeted to specific regions. Ultimately, the heightened awareness of possible voter fraud in Malaysia appears to indicate a healthy media environment in Indonesia, even if the geo-targeted posts we identified are suspicious by virtue of the intent of targeting voters in a specific geography.
However, in light of the disinformation-fueled violence that broke out after the election, media coverage of the ballot scandal in Malaysia may have been a precursor to the subsequent claims of massive election fraud.
Given the small share of geo-targeted posts that contained politically charged content in Finland, did not appear to jeopardise the democratic election process. We found that Finnish social media users searching with generalised keywords would not have found much pertinent information relating to the 2019 parliamentary elections. We did, however, identify a number of ads on Google promoting the National Coalition Party both nationwide and specifically in Uusimaa. This indicates that the National Coalition Party may have attempted to capture the attention of Finns who sought broad knowledge about the 2019 elections.
Despite these findings, we acknowledge that our research focused on the search results of generalised election-related keywords and an - alysed a fairly small sample of posts.
Given the limited results of our research, we recommend that the topic of geo-targeting during potentially contentious elections be pursued further.
We aimed to understand what the average individual in Indonesia and Finland would see when typing in general election search terms. However, we would recommend applying this methodological framework in different cases, using issue- or party-centric keywords, to com - pare how findings would differ.
Further investigation into the role of geo-tar - geting is vital to both understanding how so - cial media platforms guide users to specific content and how their systems share content during election periods. Distorted or untruth - ful information, particularly when targeted at certain audience segments, can have a lasting impact on the world’s democracies.
Hence, it is important to consider the tailor - ing of information to specific locations when monitoring the information environment during crucial political periods.