Satellite campaigns, organisations that campaign for a party without being part of it, became a visible part of British elections in 2017. The best known was Momentum, formed to support Jeremy Corbyn's leadership of the Labour Party. I ask what Momentum's Facebook page did that Labour's own pages did not. I compare all 153 retrievable posts from Momentum's page in the month before the 2017 general election with 278 posts from the Labour Party page and 144 from Corbyn's page, each hand coded for 154 features. The party page ran a campaign for votes: 85% of its posts asked for a vote, 66% made a pledge and 80% used the party slogan. Momentum asked for a vote in 12% of posts. Instead about half its posts asked people to help on the doorstep, volunteer or sign up to campaign, against 3% to 7% on Labour's pages, and it used humour in a quarter of its posts. Corbyn's page was close to the party page overall, but on all eleven reliably coded features where it differed significantly from the party page, it differed in Momentum's direction: less policy, fewer calls to vote, more news clips and more attacks on the opposition leader. On the Labour side of the 2017 campaign, the leader's page sat between a formal party campaign and a satellite that asked its followers to campaign. This preprint replaces my 2021 working paper on the same data (https://doi.org/10.31235/osf.io/hdqrg). The deposit includes the coded counts by page and code, with each code's reliability. The underlying posts are not shared. Provenance and AI use. The data were collected with Netvizz for my doctoral research, before Netvizz closed in 2019. This paper builds on my doctoral thesis (Hotham 2021, University of Bath). An earlier version was circulated as a SocArXiv working paper in 2021 (https://doi.org/10.31235/osf.io/hdqrg) and presented at the Political Studies Association conference in 2021, and the analysis develops Chapter 7 of the thesis. The analysis and writing were completed in 2026. The analysis code and drafts of the text were prepared with the assistance of an AI system (Claude, by Anthropic), which was also used to review drafts. I checked the analysis and the text and take responsibility for both. This version adds this statement to the paper; the findings are unchanged.
No takes yet. Share an insight, caveat, or question.
Tristan Hotham (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: