The municipal elections held on 15 and 22 March 2026 served as an unprecedented testing ground for measuring the gap between social media activity, search engine signals and final ballot results. Drawing on a corpus of over 24,000 comments analysed by Netino and consolidated search data from Trajaan, this article offers a rigorous comparative assessment: what the digital noise anticipated — and what French voters actually decided.
Sources: Netino thematic analysis report (24,558 comments posted under 561 articles published between 01/12/25 and 14/04/26), Trajaan search data report (data as of 30 March 2026), and official results from the Ministry of the Interior (votes of 15 and 22 March 2026).
1. Pre-Election Period: Digital Noise Reflecting a Local Vote Captured by the National Agenda
A debate 2.7 times more national than local
The analysis of the pre-electoral corpus conducted by Netino reveals a first structural paradox: municipal elections generate almost no municipal conversations. Substantive local topics (transport, housing, public services, education) account for a mere 4.5% of comments, compared with 12.5% for national themes (immigration, Islam, fascism). And 83.5% of comments address no specific substantive issue at all, reducing the election to a succession of knee-jerk reactions, partisan jabs and value judgements.
The national-to-local ratio stands at 2.7x before the vote. This disproportion does not improve after the results: the post-election report records a ratio of 3.2x, with two thirds of comments addressing no concrete topic. The nationalisation of the municipal debate is not a by-product of the campaign: it is a structural constant in the way French people engage with local elections on social media.
Source: Netino thematic analysis report.
The Parisian Obsession and the Dominance of Personalities
Paris accounts for 5% of all pre-electoral corpus mentions, far ahead of Nice (0.8%), Bordeaux (0.4%) and Marseille (0.3%). The race for the Hôtel de Ville absorbs the bulk of national conversational energy, to the point that local dynamics in other major cities — Lyon, Toulouse, Nantes — remain almost entirely invisible in the comments.
This hyper-Parisian focus is confirmed by pre-electoral search data (September 2025–March 2026): Trajaan records 5.53 million searches for “Sarah Knafo” over six months (up 45% year-on-year) and 2.76 million for “Rachida Dati” (+74% year-on-year). These two names alone generate 8.28 million searches — a volume with no equivalent in the most-searched municipal policy topic, “cleanliness”, which reaches only 54,500 queries. The conclusion is unambiguous: French people enter the municipal campaign through faces, not through manifestos.
Source: Trajaan report, search data as of 30 March 2026.
LFI as the Central Foil, the RN as a Diffuse Presence
In the pre-electoral corpus, LFI is the most mentioned party at 6.4% of comments — but almost exclusively as a target. Accusations focus on “Islamo-leftism”, alleged support for the Muslim Brotherhood and the strategy of capturing town halls as a “national springboard”. The RN (3.3%) and the PS (2.9%) lag far behind.
Identity-related themes dominate the top rankings: “fascism / far right” (5% of the corpus), “grand remplacement” (2.9%), “Islam / Muslim” (1.8%), “racism” (1.7%). The topic “debt / public finances” (1.4%) is the only local governance entry in the top 10, but it functions more as generic criticism of “left-wing cities” than as any analysis of municipal budgets.
The phrase “grand remplacement” appears 166 times in the corpus — often ironically (“we were told it was a fantasy”). “Machine à perdre” [losing machine] (22 occurrences) crystallises the anxieties of a Parisian right whose forces are fragmented across multiple candidacies — Pierre-Yves Bournazel and Sarah Knafo chief among them — rather than rallying behind Rachida Dati.
Source: Netino thematic analysis report.
A Combative, Ironic Tone — But Not Predominantly Violent
The dominant tone of the pre-electoral corpus is irony and sarcasm (2.5%), ahead of fear (2.2%), calls to vote (2.1%) and verbal aggression (1.7%). Blunt insults account for just 1.3%. This corpus points to an angry but structured political conversation, where invective often yields to detached commentary or sharp one-liners.
Source: Netino thematic analysis report.
2. Post-Election: Surprising Results in Light of the Digital Noise
Turnout: higher than 2020, but persistently elevated abstention
The votes of 15 and 22 March 2026 continue the pattern of persistently high abstention in French local elections. First-round turnout was estimated at 56% (Ipsos-BVA for France Télévisions), and the second round recorded a definitive turnout of 57.03% according to the Ministry of the Interior — historically low outside the exceptional Covid circumstances of 2020, but up significantly from the 2020 second-round figure of 41.6%.
Abstention remains socially differentiated: 25-to-34-year-olds constitute the most abstentionist age group in the first round. The abundant and sometimes virulent digital noise did not translate into exceptional mobilisation, raising the question of how representative online conversations are of the actual electorate.
Sources: Ministry of the Interior, official data 15 and 22 March 2026; Ipsos-BVA-Cesi for France Télévisions.
Paris: a clear Grégoire victory — the “losing machine” proved right
Paris was, before the vote, the centre of gravity of digital conversations (5% of pre-electoral corpus mentions). It also delivered the most widely discussed result of election night. In the first round, Emmanuel Grégoire (PS-Écologistes-PCF) came a clear first with 37.98% of the vote, ahead of Rachida Dati (LR-MoDem) at 25.46%, Sophia Chikirou (LFI) at 11.72%, Pierre-Yves Bournazel (Horizons-Renaissance) at 11.34% and Sarah Knafo (Reconquête) at 10.4%.
Between the two rounds, Knafo stood down and called for a vote against the left. Bournazel merged his list with Dati’s. Grégoire refused any alliance with LFI. Sophia Chikirou remained in the race. In the second round, Grégoire won the mayoral race with 50.52% of the vote, against 41.52% for Dati and 7.96% for Chikirou.
The socialist’s victory is both clear-cut and symbolically weighty: it validates the strategy of left-wing union excluding LFI, defies predictions of a “tight” contest after the right-wing list merger, and consecrates the continuity of socialist-green governance of the capital after twenty-five years of PS mandates. The phrase “machine à perdre” (22 occurrences) crystallised fears that the Parisian right’s fragmented candidacies — Bournazel and Knafo chief among them — would prevent a rally behind Dati.
Sources: Paris City Hall, definitive results 22 March 2026; France 24; France 3 Île-de-France.
Lyon: Doucet’s Comeback Against Aulas — a Story Almost Absent from Online Conversations
Lyon represented the second major political storyline of the vote, yet it was virtually absent from the pre-electoral digital corpus (0.2% of mentions). It was nevertheless the scene of one of the most intense confrontations of the election: incumbent Greens mayor Grégory Doucet, widely considered the underdog against Jean-Michel Aulas backed by the centre-right, ultimately prevailed by the narrowest of margins with 50.67% of the vote (104,702 ballots) against 49.33% for the former OL president (101,940 votes) — a gap of fewer than 3,000 votes.
Doucet’s re-election is partly explained by the alliance forged with LFI between the rounds, which provided a decisive additional vote share, at the cost of a strategy regarded as risky by part of the left. In contrast, Bruno Bernard (Greens) lost the presidency of the Greater Lyon authority to LR candidate Véronique Sarselli, allied with Aulas.
The emergence of the Aulas storyline in the post-electoral corpus (nearly 2% of total comments) confirms that online debate reacts to events but does not anticipate them.
Sources: France 24, official results 22 March 2026; Netino report.
Marseille and Nice: two diverging trajectories
In Marseille, Benoît Payan (left, excluding LFI) was comfortably re-elected with 54.34% of the vote, relegating RN candidate Franck Allisio to 40.3% and Martine Vassal (centre-right) to 5.36%. This result runs counter to the pre-15 March fears expressed online about a possible RN victory in France’s second city — fears amplified by close first-round scores (Payan at 36.7%, Allisio at 35.02%).
Nice tells the opposite story: Éric Ciotti (UDR, allied with the RN) won the mayoral race with 48.54% of the vote, ahead of incumbent Christian Estrosi (Horizons) at 37.20% and the PS-PCF-Écologistes candidate at 14.26%. Nice thus becomes one of the rare large cities to shift towards the right-far right bloc, confirming the strength of this alliance in the Mediterranean south-east.
Sources: France 24, official results 22 March 2026.
The national picture: broad-bloc stability, a Green retreat, RN advances in mid-sized cities
At the national level, results suggest relative stability in the major cities but significant shifts across intermediate territories. Among the 42 cities with more than 100,000 inhabitants, franceinfo counts 12 right-wing cities (same as in 2020), 6 centre (+1), 22 left (-2) and 2 far-right (+1).
The most clear-cut lesson is the Greens’ retreat: from 16 to 6 town halls won among the large cities. The defeats of Pierre Hurmic in Bordeaux (to Macronist Thomas Cazenave with 50.95%) and of several green mayors elected during the 2020 “green wave” illustrate the exhaustion of a specific electoral cycle, even as environmental issues have not disappeared from local agendas. The PS, by contrast, consolidates its strongholds and secures symbolic victories.
The RN’s progress is real but geographically concentrated. The far-right bloc wins around 61 municipalities — a record high — but this advance remains limited to the party’s historical strongholds and peripheral zones. In major university and economic cities, the sociological glass ceiling persists. According to consolidated franceinfo data, left-wing independent municipalities fall from 196 to 148, and socialist communes from 141 to 129, while the right gains ground in mid-sized cities, reaching 464 communes compared to 455 in 2020.
Sources: franceinfo, figures published 23 March 2026; Ministry of the Interior; Sud Radio analysis, 26 March 2026
3. Comparative assessment: what digital signals got right — and what they missed
What Digital Signals Correctly Anticipated
Several dynamics foreshadowed by digital noise did materialise in the results. The conversational dominance of Paris (5% of mentions, 8.28 million searches for Dati and Knafo) correctly prefigured the symbolic and political importance of the capital’s battle. The alliance storyline (86.7% of the between-round corpus, 16–22 March 2026, concerned this topic exclusively) reflected an effective reality: it was indeed in alliance configurations that several close contests were decided.
The centrality of the LFI debate in comments (9.6% of the post-electoral corpus) found an echo in reality: the question of allying with La France Insoumise structured strategic choices across the left in many cities, with contrasting outcomes — decisive in Lyon (Doucet), counterproductive in Limoges, Bordeaux and Toulouse. The search dynamic around the Nice municipal race (+3,538% annual growth according to Trajaan) correctly signalled a high-stakes contest, confirmed by Ciotti’s historic victory.
What digital noise missed or over-amplified
The identity-related themes (“grand remplacement”, Islam, immigration) that formed one of the densest conversational clusters in the pre-electoral corpus (January–14 March 2026) collapsed in post-electoral discussion after 22 March (0.9% of the corpus). This deactivation suggests these themes functioned as political identity markers rather than as actual voting drivers. Ballot boxes spoke of alliances, local governance and personalities — not of “grand remplacement”.
The hyper-concentration on Sarah Knafo (5.53 million searches, dominant presence in the corpus) did not translate into decisive electoral weight: qualifying with only 10.4% in the first round in Paris, she stood down before the second. Her actual vote share remained in the corpus’s lower margins. Digital visibility is not a reliable indicator of electoral weight, particularly since a large proportion of searches is driven by curiosity, hostility or media coverage rather than voting intent.
The phrase “machine à perdre” (22 occurrences, the lexical signature of the Parisian right in the pre-electoral corpus, before 15 March) pointed in comments to Bournazel as the dividing force responsible for Dati’s forthcoming defeat. Reality confirmed Dati’s failure (41.52% against 50.52% for Grégoire), but the causal analysis is more complex: even with the Bournazel-Dati merger and Knafo’s withdrawal, the right failed to close a 12-point gap established as early as the first round.
Finally, Lyon — virtually absent from the pre-electoral corpus — delivered the most fiercely contested result of the election. This digital blind spot is telling: online conversations organise around nationally prominent faces, not around real local dynamics.
Abstention: A Weak Digital Signal, a Heavy Electoral Reality
The pre-electoral corpus (through 14 March 2026) mentions abstention in 4.9% of comments (“democracy / voting” theme) alongside signals of disengagement (“what’s the point”, “they’re all the same”). The post-electoral corpus (from 23 March 2026 onward) confirms this representational crisis: a notable fraction of post-result comments expresses disillusionment, democratic fatigue and a shared sense of betrayal across all camps.
Yet actual turnout (57% in the second round) was higher than in 2020. This paradox is explained by a compositional bias in the digital corpus: the most active users on political social media are not representative of the median voter. Disengagement expressed online coexists with real mobilisation — but that mobilisation benefited different age groups and territories differently.
Sources: France Bleu, definitive turnout 22 March 2026; Netino report.
A search ecosystem driven by name recognition, not by policy substance
The Trajaan report (consolidated data as of 30 March 2026) confirms that the electoral search market is structured around names, not policies. The “cleanliness” segment (54,500 searches) represents less than 1% of the combined search volume for “Sarah Knafo” and “Rachida Dati”. Local concerns (transport, debt, housing, security) attract attention in comments when they worsen, but remain under-represented in search queries.
This finding has concrete implications for political information actors: capturing an audience requires going through personalities before redirecting attention to the issues. An editorial strategy focused from the outset on local manifestos remains structurally disconnected from the actual search behaviour of French internet users.
Source: Trajaan report, strategic recommendations, March 2026.
Conclusion: digital noise as a mood barometer, not an electoral oracle
One month after the 2026 municipal elections, the comparative analysis of online conversations and ballot results yields one key lesson: digital noise is an excellent gauge of the prevailing political climate and the identity obsessions of the moment, but a signal to challenge when it comes to actual electoral power dynamics.
The nationalisation of the debate (2.7x to 3.2x ratio) is the most thoroughly documented systemic distortion: social media transforms every local election into a proxy for the national political battle, erasing governance specificities, less media-visible personalities and territorial dynamics. The Lyon result — near-invisibility before the vote, a shockwave after — is the most eloquent demonstration.
The dominance of names over programmes in search data and comments confirms that digital attention follows the logic of political spectacle, not that of democratic deliberation. Sarah Knafo crystallises 5.53 million searches for a 10.4% vote share in Paris.
For those who use these signals (political campaigns, media outlets, opinion monitoring teams), methodological caution is essential: the digital corpus is a biased sample (over-representation of extremes, the most politically engaged, active commenters), temporally distorted (immediate reactions dominate) and thematically skewed (concrete local issues are chronically under-represented).
The results of these municipal elections illustrate a reality that any organisation grappling with online opinion measurement eventually encounters: the volume of available data has never been greater, but its interpretation remains a discipline in its own right. Capturing 24,000 comments is a technical feat; distinguishing within that mass what constitutes background noise, a weak signal or an emerging dynamic — and translating it into actionable recommendations — is an entirely different matter.
Netino: turning content moderation and social listening into strategic intelligence
Netino has established itself as a benchmark actor in digital content analysis and moderation in France. Its value does not lie solely in its capacity to process massive data volumes — 17,447 comments analysed in this single post-election corpus — but in the methodological rigour with which raw material is structured, segmented, and interpreted.
For a public institution (ministry, local authority, government agency…), this analytical capability enables real-time measurement of how public policies are perceived, early identification of friction points before they escalate into crises, and evaluation of the effectiveness of territorial communication campaigns. In an electoral context, it makes it possible to distinguish diffuse opinion from what is actually shaping voting behaviour.
For an association or NGO, the benefits are different but equally concrete: mapping the communities engaged with a given issue, identifying both potential allies and detractors, and measuring the impact of a public statement or citizen mobilisation campaign.
For a company — whether operating in services, industry, or consumer goods — analysis of online conversations provides a competitive intelligence tool, a reputation management resource, and a means of understanding customer expectations that standard automated tools cannot deliver with the same depth. The ability to segment a corpus by tone, topic, or geography, and to extract insights directly actionable by communications, marketing, or executive teams, represents a genuine competitive advantage.
What the reports analysed in this article demonstrate compellingly is that the quality of insight depends directly on the quality of the method: an analysis limited to counting mentions will miss the 2.7x national-to-local ratio, the post-election deactivation of identity themes, or the Lyon blind spot. It is precisely this level of reading — analytical, critical, and contextualised — that Netino makes available to its clients.
Trajaan: search intelligence as a compass for real intent
Netino works in partnership with Trajaan, a company specialising in online search intelligence, according to the specific needs and objectives of each client engagement. This partnership illustrates a valuable methodological complementarity: where comment and social conversation analysis reveals what people say publicly, Trajaan’s search data reveals what they seek in silence.
These two signals are often divergent. In the context of the 2026 municipal elections, comment feeds were saturated with identity themes — but search queries focused overwhelmingly on candidates (8.28 million combined queries for Knafo and Dati) and on practical electoral information (3.59 million for “municipal elections 2026”). Trajaan’s search data made it possible to qualify the intent behind the attention: French internet users wanted to understand who these candidates were — not necessarily to vote for them.
For any organisation seeking to better calibrate its online presence, content, or campaigns, this distinction is fundamental. A public institution seeking to communicate about a reform, a company launching a product, or an association advancing a cause all stand to gain from knowing not only what people say about their subject, but what they are trying to find out — and therefore what their content must genuinely deliver in order to capture and convert an audience.
The combination of qualitative and tonal conversation analysis (Netino) with the mapping of search intent (Trajaan) constitutes one of the most comprehensive approaches currently available for understanding digital opinion in its dual dimension: what is said and what is sought. In an increasingly saturated information environment, it is this analytical depth — not data volume alone — that makes the difference between mere monitoring and genuine strategic intelligence.
This article draws on three primary sources: the Netino thematic analysis report (pre-electoral corpus, 7,111 comments), the Netino post-election analysis report (17,447 comments divided into four segments: results, between-rounds, 2027 projections, and generic conversation), and the Trajaan search intelligence report (data as of 30 March 2026). Electoral results cited are drawn from official data published by the French Ministry of the Interior (votes of 15 and 22 March 2026) and from reporting by France 24, franceinfo and Public Sénat.
