Spark
Book call
Methodology

How the survey was run.

Two surveys, one comparison sheet, all of it auditable.

Why we ran it

Jersey-specific data on AI adoption has been thin. Trade press covers the FTSE 100; the Big Four cover global enterprise. Nobody covers the 5,000 Jersey businesses in the middle, the firms that keep the island running. We wanted a number against which a Jersey business could say "that's us" or "that's not us", and act on the answer.

The result is a frozen baseline of 43 Jersey businesses and 200 UK comparators, plus 50 qualitative interviews, refreshed every time someone takes the survey on this site.

The numbers at a glance

N=43
Jersey quantitative
Spark survey, October 2025. Distributed via Notion forms.
N=50
Jersey qualitative
Interviews across financial, professional services, and operations.
N=200
UK comparator
Pollfish panel, two waves. Stratified by company size, weighted to UK SME distribution.

How we did it

01

Designed thirteen comparable questions.

We picked questions that would land cleanly in both Jersey context and a generic UK panel. Buckets were aligned where possible (company size, AI tool count) and translated where not (sector-specific phrasing). The full mapping is in the workbook, not on this page.
02

Ran the Jersey survey ourselves.

43 valid responses through a Notion form, distributed via Spark's network and Digital Jersey channels through October 2025. No incentives, no panel.
03

Ran the UK survey on Pollfish.

200 respondents in two waves: a first 100 across all sizes, then a sub-200-employee top-up to balance the sample against the UK SME population. Pollfish stratified by company size and applied weighting to match ONS small-business distribution.
04

Did 50 qualitative interviews.

In-person and remote, 30 to 60 minutes each, covering wealth management, fund admin, legal, accounting, retail, and operations. Interviews informed the report's narrative, not the quantitative percentages.
05

Built a live aggregate.

Every survey submission with dataset consent joins a live Jersey aggregate that is recomputed on every dashboard load. The aggregate is hidden until at least five consented responses are in, to protect early-adopter identifiability.

Sample sizes and limitations

Jersey N=43 is small. We are honest about that. Where a single answer option had fewer than five Jersey respondents we report the share but we do not draw conclusions from the gap. The UK comparator at N=200 is a Pollfish panel and not a perfect mirror of Jersey's industry mix, so the Jersey vs UK deltas should be read as directional rather than precise.

The qualitative interviews are not random and were not weighted. They give the report's narrative texture, not its statistics.

What the survey collects

The 13 questions, plus light optional metadata.

Required

  • Email address
  • Location: one of jersey, uk, other
  • Company size, single choice from 1-10, 11-50, 51-200, 200+
  • AI tool count: 0, 1-2, 3-5, 6-10, 10+
  • AI capabilities used, multi-select from 11 options including custom-built tools and an exclusive "none of the above"
  • Three 1-10 sliders: AI understanding, data confidence, likelihood to adopt more in 12 months
  • Hours of automatable work per week: Less than 5 hours, 5-10 hours, 11-20 hours, 21-40 hours, 40+ hours, Don't know
  • Biggest pain point, value of solving it, top three barriers
  • Realistic 12-month investment band
  • Jersey-only: does being on island make AI adoption easier or harder
  • Where you would go for help with AI
  • Explicit consent to add your anonymised answers to the dataset

Optional

  • Company name (used for the dashboard greeting)
  • Sector, dropdown of Banking & Capital Markets, Insurance, Trust & Fund Admin, Legal, Accounting, Other Professional Services, Other
  • Role, dropdown of Owner / MD, C-suite, Director / Partner, Manager, Operations, Other
  • Marketing consent (for occasional Spark research updates)

Full data-handling details are in the privacy policy.

Behind it

The research and the microsite are by Spark AI Agency, an arm of Spark Curiosity Ltd. Author: Seb Lawson. The qualitative fieldwork ran in partnership with Digital Jersey, whose distribution help on the quantitative survey is gratefully acknowledged.

How to cite

Cite the report as: Spark AI Agency, State of AI in Jersey 2026 (Jersey: Spark Curiosity Ltd, 2026).

Journalists are welcome to quote the headline findings with attribution. For the underlying dataset or methodology questions, email seb@sparkcuriosityhq.com.

Want in?

Take the survey. Add yourself to the next aggregate.

Take the 5-minute survey