Global Talent: How To — Impact & Methodology
We publish the service metrics, methods and evidence behind the Global Talent pro bono clinic.
Reporting period: Nov 2024–Sep 2025 Last updated: 30 September 2025, UK time Public channels only
Answered ≤7 days
96–98%
each month
Resolved ≤7 days
76-81%
each month
Median time‑to‑first‑answer
3.3–3.6 h
(P90 ≈ 11–12 h)
1‑to‑1 under adviser oversight
826
Volunteer hours
8,713 h
Supported endorsements
274
Outcomes confirmed with a 90‑day lag; staff override removes AI within 12h; peer answers counted only when staff‑confirmed.
Highlights
Fast & consistent:
SLA stays high and steady across months; distribution tails (P75/P90) are disclosed.
Fair & supervised:
every consultation is adviser‑reviewed; peer support is staff‑confirmed before it counts.
Scalable triage:
AI helps with first answers; a 12‑hour staff override keeps quality and safety.
Low single‑digit share of the route:
outcomes represent a small but meaningful share UK‑wide.
A. Service performance (SLA)
Answer rate ≤7 days
96–98% each month
Resolved rate ≤7 days
76–81% each month
Median time‑to‑first‑answer
3.3–3.6 h (P90 ≈ 11–12 h)
B. Delivery & capacity
Supported endorsements
274
1‑to‑1 under adviser oversight
826
C. Triage & community
How we count
Scope & unit of analysis
Public channels only; private/DMs excluded
Questions are deduplicated with a 10‑minute window
Time stamps in UK time
Counting rules
First‑answer flags:
if staff intervene within 12 hours, the AI flag is removed; peer answers count only when staff‑confirmed.
AI‑only (resolved):
AI‑first cases that closed without staff intervention; peer confirmation allowed.
TTF:
time from first question to the first substantive answer. We publish P50 and reference P75/P90 in notes.
Outcomes & lags
Resolved ≤7 days:
share of cases fully resolved within 7 days (case‑ID based).
Endorsements:
confirmed Global Talent outcomes; 90‑day confirmation lag applies.
How we count
Scope & unit of analysis
Adviser oversight for 1‑to‑1; peer support ≠ legal advice.
Public‑only scope can undercount private interactions; late confirmations may restate monthly outcomes.
Seasonal variation (volunteer holidays) mildly affects throughput, not quality.
Data dictionary
Quality, verification & data access
Sanity checks we publish
AI + Staff + Peer = 100% each month.
AI‑only ≤ AI‑first.
Answered ≤7 days ≥ Resolved ≤7 days.
AI‑first ≈ round(Unique questions × AI% from First‑answer mix).
Volunteer hours = Σ categories (1‑to‑1, moderation, content/wiki, training, admin).
Σ monthly = total for 1‑to‑1 and volunteer hours (allowing for rounding).
Independent verification & data access
For datasets and methods, email verification@fairgo.org.uk (independently‑monitored inbox, UK time).
We aim to respond within 3 UK business days with an Evidence Pack (aggregates only; no PII).
What we share on request
CSV (aggregates)
the exact monthly datasets behind each chart (1‑to‑1 with the visuals).
Methodology (PDF)
data dictionary, counting rules, lags and QA.
Sanity‑check results
calculations for the checks above.
Oversight evidence
proof of licensed immigration adviser and the “second‑pair‑of‑eyes” SOP (summary)
Version log
release date and field changes
How to request (suggested email)
Subject:
Evidence Pack request — GTV Clinic
Body:
Name & organisation
Purpose (award panel / journalist / academic / regulator)
Charts/period needed
Deadline/format notes (CSV/XLSX/PDF)
Guardrails & privacy
Aggregates only; no PII or raw messages.
Scope: public channels only, 10‑minute dedup; 12‑hour staff override for AI; outcomes use a 90‑day confirmation lag; all times UK time.
References available on request.
Scale context
The Global Talent route processes endorsements UK‑wide in the thousands per year; our confirmed outcomes are a low single‑digit share of the route.
Changelog
Version 1.0.0
Introduced a new interface for data visualization.
Optimized loading times for enhanced performance.
Fixed bugs affecting chart displays.
Version 0.9.0
Added new filters for more precise data analysis.
Improved navigation within the control panel.
Included helpful tips for users.
Privacy & Ethics
No PII published; only aggregates.
Adviser boundary: peer support ≠ legal advice; all consultations are adviser‑reviewed.
AI guardrails: supervised use; staff override within 12 h.
Corrections & complaints: [link to policy].
UK time; English (en‑GB).