AI Quick Wins Planner

1. Source candidate opportunities

Start with what you already know. Members who have completed the Shadow AI Audit should treat its MONITOR and FORMALISE rows as their first candidate pool -  these are workflows where AI is already in use, ready to be made formal.

Members who have completed the Capability & Readiness Audit will find some quick wins suggested by capability strengths the team has but isn't yet exploiting. If you haven't done either yet, source from: front-line teams (who already pastes into ChatGPT?), back-office (repetitive copy/paste, manual reconciliation), board-pack production (manual collation, executive summaries), recent vendor demos. Aim for 15–25 candidates before you start filtering. The "From Prior Tools" sheet in the workbook walks through this in detail.

Watch for: do not start with vendor demos. Start with business friction.

2. Screen for governance fit

Run each candidate through the six-point screen: personal data, special-category or high-sensitivity data, EU AI Act classification, AI Use Policy Builder fit, vendor due diligence, reversibility. Anything marked Refer moves out of the quick-win pool. If you don't have an Acceptable Use Policy yet, complete the AI Use Policy Builder before scoring any candidate.

Watch for: a risky idea may still be valuable. It just belongs on the roadmap, not in this 90-day planner.

3. Score with two plain-English questions

We ask two questions per candidate - will this move a number leadership watches? (Yes / Partly / No) and can we deliver it with what we already have? (Yes / Mostly / No). The Excel turns those answers into a priority score automatically (Yes = 5, Partly/Mostly = 3, No = 1; Priority = Value × Feasibility, max 25). No 1–5 spreadsheet maths required.

Watch for: avoid false precision. This is a prioritisation tool, not a business-case model.

4. Sponsor your top three

Complete a one-page sponsor brief for each of the top three. Each needs a named business owner, an executive sponsor, an estimated annual value, a single KPI, and a go-live date inside 90 days.

Watch for: committees do not own quick wins. People do.

5. Lock measurement before launch

Capture a 4-week baseline, name a measurement owner, schedule day 30, 60, 90 reviews, and document the roll-back plan. Without this, the quick win is anecdote, not evidence.

Watch for: without a baseline, the story becomes anecdotal - and anecdote will not survive a board challenge.

Governance screen -  at a glance

Run this screen at Step 2 (Screen for governance fit), before you score any candidate. Anything in the Refer column is parked or moved to the AI roadmap.

How to read the score

Because Value and Feasibility each score 1, 3 or 5, the only possible priority scores are 1, 3, 5, 9, 15 or 25. The bands below cluster those into action tiers.

When you complete the tool you will have:

• A long-list of 15-25 candidate AI opportunities, scored objectively.
• A shortlist of 3 sponsored quick wins, each with a named owner, sponsor, KPI and go-live date.
• A governance audit trail showing each candidate has been screened against your AI Use Policy.
• A measurement plan ready to run from day one.
• A leadership-team artefact you can put in front of the governing body for sign-off.

What this is NOT

This is not your full AI strategy. Quick wins are the operating-model proof point - evidence that you can move from idea to value safely with governance intact. Longer-cycle initiatives belong in your broader AI roadmap, alongside the Capability & Readiness Audit.

Three regulatory threads matter for quick wins. The governance screen at Step 2 (Screen for governance fit), under the How to Use It dropdown, checks all three below.

• GDPR: Most quick wins involve some personal data, even if only staff names and email. The screen flags whether the use case is covered by an existing Article 30 record and a lawful basis. Special-category or high-sensitivity data (health, ethnicity, biometrics, financial hardship or vulnerable-customer indicators) is rarely compatible with a quick-win timeline - those use cases belong in the broader AI roadmap with a full Data Protection Impact Assessment (DPIA).
• EU AI Act: Quick wins are almost always minimal-risk under the EU AI Act. The screen makes you confirm this explicitly. If a candidate is high-risk (e.g. recruitment screening, credit scoring, biometric categorisation), it is out of scope here - handle it through your EU AI Act readiness assessment instead.
• Director duties - Companies Act 2014: Section 228(1)(g) of the Companies Act 2014 requires a director to exercise the care, skill and diligence which would be exercised in the same circumstances by a reasonable person. Quick wins meet that test where they are reversible, observable, and covered by policy. Sponsorship at director level - not just budget approval - is what discharges the governance obligation.

Track these reporting metrics quarterly:

• Long-list volume:  Number of candidate opportunities sourced per quarter - target 15–25.
• Screen pass rate: Percentage of long-list that passes the six-point governance screen - typical range 40-60%.
• Sponsorship throughput: Number of quick wins sponsored per quarter - target 3.
• Time to first value: Days from sponsorship to first measured KPI movement - target ≤ 60 days.
• Day-90 outcome split: Percentage of sponsored quick wins classified Scale / Sustain / Retire at day 90
• Annualised value realised: Cost of measured value across live quick wins - net of build cost.

The planner is a quarterly habit, not a one-off exercise. Embed it through these six disciplines:

1. Cadence: Run the planner every quarter. Treat the long-list as a rolling backlog - items not picked this quarter are first in line next.
2. Sponsorship discipline: Refuse to sponsor a quick win without a named business owner and a single measurable KPI. Committees do not own quick wins.
3. Stop criteria: Every sponsored quick win has a written stop threshold. If it isn't met by day 60, you retire, not extend.
4. Governance gate: The AI Use Policy Builder is the gate. If a quick win needs a policy exception, it is no longer a quick win.
5. Visible scoreboard: Publish the live shortlist and outcomes to the leadership team monthly. Quick wins die in the dark.
6. 90-day discipline: If a candidate can't credibly go live inside 90 days, it isn't a quick win - sponsor it on your AI roadmap instead. The planner stays sharp because the calendar is non-negotiable.