The candidate with the most Facebook likes in the 2013 Bangkok governor election finished third, and every campaign team agreed that likes could not predict votes

The March 2013 Bangkok governor election was the first in which every leading candidate campaigned seriously on social media, yet the candidates with more followers lost. This study asks which channels the top four candidates used and how, which ones their campaign teams believed worked, and whether social media actually moved votes.

A qualitative study: document analysis of every campaign item collected directly from the four leading candidates' channels, traditional and digital, from March 2013; in-depth interviews with four people representing the candidates and the election sector, analysed thematically; and a comparison of follower and engagement counts on Facebook, YouTube, Twitter, Instagram, Google+ and campaign websites against the actual vote.

• The winner did not lead on followers in most channels. The third-placed candidate had the most Facebook likes (238,138); the fourth-placed candidate had the most YouTube views (2.57 million, on a channel open since 2007) and the most Instagram followers. The winner led only on Facebook's “talking about this” count (146,294) and on Twitter followers (115,328).
• The actual result: the winner took 1,256,349 votes (47.75% of votes cast), the runner-up 1,077,899 (40.97%), third place 166,582 and fourth 78,825; 2,715,640 of 4,244,465 registered voters turned out.
• Everyone used Facebook as the main channel, posting text, photos and video; everyone had a YouTube channel and Twitter and Instagram accounts. Only the winner built a mobile app and a Google+ page; only the third-placed candidate used augmented reality. A social-listening service showed the candidate with the highest share of positive online comments (82%) was the one who finished fourth.
• Campaign teams preferred Twitter for its speed and room for dialogue, with Facebook as the gateway for distributing information. All agreed social media was new compared with the election four years earlier, and although the budget was small, almost every candidate still hired a team to run the online channels.
• Most interviewees insisted social media could not decide the vote. Likes and followers did not track votes because “anyone from anywhere can click likes”: followers might be under voting age, not registered in Bangkok, or living abroad. Voters still decided through face-to-face contact such as roadshows and community visits.
• The authors conclude that social media remains necessary even though it cannot predict votes, because it builds a modern image for the candidate and opens direct conversation with voters.

The lesson still holds for campaigns of every kind: online engagement and real-world action are different counts. Likes measure interested people anywhere; votes measure eligible people who reach the booth. Before using social media as a performance indicator, ask whether the followers belong to the group that can act, and use the channel for image and dialogue rather than for headcounts.