A structural analysis of AI disruption across the UK agency economy. 133 agency types measured for vulnerability. Revised September 2026 with 38 re-scored categories and three new ones.
By Matt Parry · The Future Collective
Every UK agency type scored, ranked, and analysed for AI disruption risk.
0 agencies · All Sectors
Bubble size = confidence level. Click any bubble for details.
Understand AI impact patterns across 18 industry sectors with detailed analytics.
Click any sector for detailed analysis
Average disruption scores by sector
Input your agency's characteristics to see where you'd land.
How much work can AI automate today?
Can AI replace strategic decisions?
How quickly will clients demand AI pricing?
How easily can clients bring this in-house?
Higher = more protected by proprietary methods
How hard is it to adapt your model?
Indicative score. Individual positioning may vary.
Strategic recommendations by resilience band.
Score 70+ · Urgent
Score 50-69 · 12-24mo
Score 30-49 · Strengthen
Score <30 · Capitalise
Seven months of evidence, not seven months of vibes. Thirty-eight categories moved, three were added, and — importantly — nine moved down. Every movement below carries a dated source you can check. Open any row in the index to see the reasoning behind its score.
The most useful finding in this revision is how little the headline moved. Mean disruption score went from 60.1 to 60.3, and exactly one category changed resilience band. The disruption is intensifying inside the bands — production and execution work getting more exposed, live and regulated work getting less — rather than reclassifying the industry. Anyone selling you a step-change this year is selling you something.
Where the evidence got worse between February and September.
Categories that look safer than they did in February. The index has to be able to move both ways to be worth anything.
Three disciplines that barely existed as billable service lines in February.
Thirty-eight categories re-scored against dated evidence published between February and September 2026. Three categories added: AI visibility and GEO, agent experience and agentic commerce, and AI governance, assurance and rights clearance. Every changed score now carries its reasoning and a source. Weightings, bands and the six-factor model are unchanged, so v2.0 and v3.0 scores remain directly comparable.
127 agency types scored across six weighted factors, with confidence bands, sector analysis and the self-assessment calculator.
A note on what this revision deliberately did not do: model capability rose sharply this year, but controlled studies still find no reliable real-world productivity gain for experienced developers, and Gartner expects more than 40% of agentic AI projects to be cancelled by 2027. Capability is running ahead of deployment. Where the only evidence was a benchmark, scores were left alone.
Transparent, evidence-led methodology.
Each agency type scored 1-5 across six structural factors determining vulnerability to AI-driven change.
The Future Collective works with agency leaders to develop AI-ready strategies and build defensible positions.
The index is revised when the evidence moves, not on a schedule. Leave an email and you'll get the next revision — what moved, and why — before it goes public. No newsletter, no sequence.
Your address is used for index revisions only. Unsubscribe any time.