Ranking potential AI workflows by commercial value, speed to benefit, data readiness, operational risk, implementation complexity and governance requirement.
Each AI opportunity is scored against six criteria on a scale of 1–5. The weighted total determines priority ranking and Phase 1 recommendation.
| Criterion | Weight | What it measures | Score interpretation |
|---|---|---|---|
| Commercial value | 25% | Direct financial impact through recovered margin, reduced cost or increased revenue capacity | 5 = £200k+ / yr 1 = <£25k / yr |
| Speed to benefit | 20% | How quickly measurable improvement can be demonstrated after implementation begins | 5 = <8 weeks 1 = 6+ months |
| Data readiness | 20% | Whether the input data, documents and systems are available, structured and exportable | 5 = Ready now 1 = Major gaps |
| Operational risk | 15% | Disruption risk to existing workflows, teams and client relationships during implementation | 5 = Minimal risk 1 = High disruption |
| Implementation complexity | 10% | Technical complexity, integration requirements, change management and training burden | 5 = Straightforward 1 = Complex |
| Governance requirement | 10% | Level of approval, audit trail, compliance and oversight required for safe deployment | 5 = Standard controls 1 = Heavy governance |
Six AI opportunities ranked by weighted score. The top two are recommended for Phase 1 implementation.
| # | Opportunity | Commercial 25% |
Speed 20% |
Data 20% |
Risk 15% |
Complex 10% |
Govern 10% |
Weighted total |
Phase |
|---|---|---|---|---|---|---|---|---|---|
1 |
Commercial capture & variation recovery Structured change event capture, notice automation, evidence indexing and variation narrative generation |
5 | 4 | 4 | 5 | 4 | 3 | 4.35 | Phase 1 |
2 |
Tender production & bid automation Scope extraction, assumptions library, compliance pre-fill, QA checklist generation and tender assembly |
4 | 5 | 5 | 5 | 4 | 4 | 4.55 | Phase 1 |
3 |
Margin reporting & WIP visibility Automated monthly margin snapshots, CNI tracking, invoice readiness dashboards and board narrative |
4 | 4 | 3 | 4 | 3 | 4 | 3.75 | Phase 2 |
4 |
RAMS & compliance automation Task-based RAMS generation, hazard/control mapping, COSHH indexing and permit checklists |
3 | 4 | 3 | 3 | 3 | 2 | 3.15 | Phase 2 |
5 |
Invoice readiness & application support Evidence assembly, PO validation, completion record linking and application pre-population |
3 | 3 | 2 | 3 | 2 | 3 | 2.70 | Phase 3 |
6 |
Service desk triage & SLA evidence Priority classification, dispatch-ready ticketing, SLA tracking and deduction evidence packs |
2 | 3 | 2 | 3 | 2 | 3 | 2.45 | Phase 3 |
Plotting each opportunity by commercial value (vertical) against implementation readiness (horizontal) to visualise priority positioning.
Detailed assessment of each opportunity, including the workflow it maps to, the leakage it addresses, scoring rationale and implementation considerations.
This opportunity directly addresses the largest single leakage category identified in Deliverable 02 — £620k in missed commercial recovery. The root causes are structural: late notice, fragmented evidence, manual narrative assembly and no pipeline visibility. All of these are addressable through AI-assisted workflow without changing the commercial team's role or requiring system integration.
The variation log and existing evidence (emails, photographs, site records) provide sufficient input data for an AI workflow to structure change event capture, draft contractual notices against template clauses, assemble evidence packs and generate cause-and-effect narratives for QS review. The commercial team retains full review and approval control.
Tender production is the highest-readiness opportunity in the portfolio. Data is immediately available — tender packs are self-contained document sets that require no system integration. The workflow operates alongside the existing estimating process rather than replacing it, making operational risk minimal.
The 40% admin burden identified in Deliverable 02 (38 hours per tender on repetitive work) is directly addressable through AI-assisted scope extraction, assumptions generation, compliance pre-population and QA checking. Reducing this burden by 30–40% would release 11–15 hours per tender — enough to increase capacity by 15–20 tenders per year without additional headcount.
Margin reporting targets the £410k aged WIP and CNI drag plus the £165k reporting overhead identified in Deliverable 02. The opportunity is strong but scores lower on data readiness because the Sage 200 CVR export requires manual reconciliation with contract-level data held in spreadsheets. Phase 1 implementation of commercial capture and tender workflows will improve data quality, making this workflow more effective when deployed in Phase 2.
RAMS automation has strong speed potential but carries a heavier governance burden than commercial or tender workflows. Safety documentation requires competent review, approval gates and clear audit trails. The lower governance score reflects the additional controls needed — not a barrier, but a factor that makes Phase 2 more appropriate after governance frameworks (Deliverable 05) are established.
Invoice readiness scores lower on data readiness because the current evidence (photographs, completion records, PO references) is fragmented across multiple systems and personal devices. The workflow requires upstream data improvements — particularly completion record discipline and PO management — before AI automation adds maximum value. Phase 1 and 2 workflows will improve the data environment for this opportunity.
Service desk triage has the smallest financial impact (£100k in SLA deductions) and the most complex data environment (Concept CAFM integration, real-time ticket routing, engineer attendance tracking). It is a genuine opportunity but delivers best value once the foundational data, governance and workflow patterns are established through earlier phases.
How the six opportunities sequence into a 12-month transformation roadmap, with each phase building on the data, governance and workflow patterns established in the previous phase.
How the roadmap was compiled and how it connects to the other audit deliverables.
| Input source | How it was used |
|---|---|
| Deliverable 01 — AI Maturity Scorecard | Domain maturity scores informed data readiness and governance requirement ratings for each opportunity |
| Deliverable 02 — Margin Leakage Analysis | Leakage values provided the commercial value scoring inputs. Recovery estimates set the financial targets per opportunity |
| Stakeholder interviews | Speed to benefit and operational risk scores were validated with the MD, Commercial Director and Estimating Lead |
| Document & data review | Implementation complexity scores were based on actual system exports, document quality and integration requirements assessed during the audit |
| Built AI workflow library | Each opportunity was mapped to the most appropriate Built AI workflow pack based on scope, data requirements and delivery model |