Commentary on News

Qualitative analysis of disinformation in Kenya

Evidano4 min read

Kenya’s online campaigns in 2024–2025 illustrate how coordinated hashtags, automated accounts and AI-generated clips can rapidly reshape public narratives. This post shows researchers and UX/policy teams how to run a focused qualitative analysis of disinformation (identify themes, measure coordinated amplification, and map actor networks) and how to operationalize that workflow with www.evidano.com. You’ll get a short methods checklist, key signals to watch (dates and counts from the reporting), and a reproducible two-week pilot plan.

Fast take: Why this matters for qualitative researchers

A recent investigation documents a surge of coordinated online campaigns targeting Kenyan activists (report published 21 Aug 2025). Read the full article: www.ibtimes.com/weaponising-feed-inside-kenyas-online-war-against-activists-3781355.

  • Problem: activists (e.g., Rose Njeri, detained in May 2025) were smeared by coordinated hashtags and AI-generated clips.
  • Payoff: with a reproducible qualitative pipeline you can surface narratives, count coordination signals, and present evidence-backed visualizations for policy or UX decisions.

Findings snapshot (key numbers and dates)

MetricValueSource / DateImplication
Article published21 Aug 2025International Business TimesContext for timeline and quotes
High-profile protest date25 June 2024ReportingReference point for subsequent campaigns
Hashtag views: #BBCForChaos≈5.3 millionAFP analysis cited in IBTimesLarge-scale amplification signal
Hashtag views: #ToxicActivists≈365, 000AFP analysis cited in IBTimesSustained narrative framing activists as 'paid'
Hashtag reach: #AsanteSamia≈1.5 million viewsReportingPro-government praise narrative
Allocated spyware budget$1.15 millionGovernment budget item (reported)Increased surveillance capacity
Example account activity226, 140, 104 posts under single hashtagsCode for Africa / AFP citedAutomated or highly active accounts driving amplification
Individual mentionedRose Njeri, 35, jailed in May 2025ReportingReal-world harms from online campaigns

What happened (plain English)

Between April and August 2025, rights groups and journalists documented a pattern: coordinated hashtag campaigns, high-frequency posting from repeat accounts, and AI-generated audiovisual forgeries used to discredit activists tied to June 25, 2024 protests. Tactics included mass reposting within seconds, fake video clips mimicking broadcasters, false graphics and narratives casting activists as 'paid' or foreign-funded.

  • Coordination signals: multiple accounts posting identical hashtags within seconds and accounts with prior pro-government history seeding narratives.
  • Escalation: online narratives accompanied legal and surveillance moves (terror charges, proposed social media ID laws, and a $1.15M spyware allocation).
  • Effect: reputational harm, detention (e.g., Rose Njeri), and chilling of civic tech work.

Implications for researchers and analysts

For qualitative researchers

You need reproducible methods to show narrative origin, cadence, and cross-platform spread: not just anecdotes.

Quantify coordination (posting timestamps, volume by account), then triangulate with qualitative codebooks to label motives, frames, and attack types.

For UX and product teams

Design user-safety and verification flows that account for rapid, coordinated misinformation surges.

Prioritize detection signals (burst posting, new hashtags seeded by accounts with prior campaign history, deepfake indicators) when building moderation rules.

For policy and rights analysts

Pair qualitative theme maps with evidentiary visualizations (network graphs, sample quote lists) to brief decision-makers.

Document harms with timestamps and preserved artifact captures to support legal or advocacy actions.

Do this with Evidano: AI-enabled qualitative analysis mapped to the case

Ingest & preserve the evidence

Scrape social posts and media using Evidano’s website/social scraping to capture hashtags, timestamps, and media artifacts tied to campaigns.

Store encrypted archives; Evidano does not use your data to train third-party models.

Rapid thematic coding and cross-segment analysis

Upload scraped posts and interview transcripts; run thematic and frequency analysis to find recurring frames like 'paid activist' or 'foreign-funded'.

Use hierarchical codes → subcodes to separate tactic (e.g., 'deepfake video') from target (e.g., 'activist name').

Detect coordination signals

Use timestamp clustering and co-occurrence networks to surface accounts that post identical hashtags in tight windows (the article cites multiple accounts posting 100+ times).

Evidano visualizations (co-occurrence network; actor clusters) turn raw timestamps into reproducible evidence you can share.

Explainable outputs for stakeholders

Generate clickable quote banks and exportable visual reports for policy briefs or legal teams.

AI chat over your corpus surfaces exact source examples (with preserved links) so you can show, not just tell.

Secure collection and follow-up

Transcription and translation with custom dictionaries (useful for multilingual Kenyan content); PII redaction options for sensitive interviews.

If you need more qualitative data, Evidano supports AI-avatar interviews to collect structured follow-ups while preserving anonymity.

Two-week pilot: triangle signals → themes → brief

Run this reproducible pilot to convert reporting into research-ready evidence for advocacy or product decisions.

  • Day 0–2: Ingest sample, scrape 2–3 hashtags and related accounts around June 25, 2024 and May–Aug 2025; archive media artifacts.
  • Day 3–6: Auto-transcribe/translate any audio/video, run initial thematic extraction and frequency counts to surface top frames.
  • Day 7–10: Run co-occurrence and account-timestamp clustering to identify likely coordination; flag accounts with >100 posts in one day.
  • Day 11–12: Produce visualizations (network graph, hierarchical code map) and a one-page evidence brief with clickable quotes.
  • Day 13–14: Stakeholder review; export findings and prepare a targeted intervention or policy memo.

Conclusion, next steps

Kenya’s recent cases show how quickly coordinated disinformation can translate into offline harm. A rigorous qualitative pipeline that couples theme extraction with coordination metrics turns scattered signals into actionable evidence.

  • Ready to reproduce this analysis on your corpus? Start a pilot at www.evidano.com and map narratives, actors, and timed coordination in two weeks.
  • Ethics note: this guidance is research-focused and not clinical or legal advice; handle sensitive data with informed consent and appropriate safeguards.

Keep reading

Browse all articles