What This Research Will Yield

An outcomes map for The Applied Layer — how the aims, objectives, and two-tier questionnaire translate into evidence, findings, and decisions

This sheet answers a single question: when the fieldwork is complete, what will the programme actually have? It traces the chain from each objective to the concrete evidence the interviews will produce, then up to a consolidated finding per pillar and a synthesis across the whole programme. Every outcome links back to an objective ID, so nothing the instrument collects is unaccounted for, and nothing the programme promises is unmeasured.

The evidence chain. Aim → objectives (P1-O1 … P5-O4) → two-tier questions → interview evidence → a finding per pillar → a programme-wide synthesis and forward agenda. The outcomes below are organised along this chain: programme-level first, then pillar-level, then the objective-by-objective detail.

1 · Programme-Level Outcomes

The headline yields — what the study produces as a whole:

• A maturity baseline. A defensible picture of where organisations actually sit between experimentation and AI-native, with the pilot-to-production gap quantified rather than asserted.
• A test of the ‘applied layer’ thesis. Evidence for or against the programme’s central claim — that architecture, operating model, economics, and governance, not model choice, decide enterprise AI outcomes.
• A cross-pillar pattern library. What separates organisations getting real value from those treating AI as procurement — the recurring conditions, failure modes, and decisions.
• A benchmark dataset. Structured, objective-tagged interview evidence across all five pillars, comparable across organisations and reusable for longitudinal follow-up.
• Publishable findings + a forward agenda. Evidence-based articles for the series, a confidential per-participant benchmark, and the next research questions worth pursuing (P1-O4).

2 · Pillar-Level Outcomes

What each research stream establishes, the artefact it produces, and who it helps.

Pillar	What the research establishes	Deliverable produced	Who it helps
P1 · Beyond the Model	A maturity baseline and an evidence-based reality check on adoption versus impact.	Maturity-ladder placement + a validated failure-mode taxonomy.	CEOs & boards — set strategy on evidence, not hype.
P2 · Production AI Architecture	Which architectures and agent tiers are genuinely in production, and which work.	A production-readiness map by workload + the ‘architecture over model’ evidence.	CTOs & architects — invest in the retrieval/architecture layer first.
P3 · Operating Models	Which operating-model archetype each organisation runs, and the conditions that separate success from failure.	Archetype distribution + the six success conditions, validated in the field.	COOs & transformation leads — choose or fix the operating model.
P4 · Cost & Platform Landscape	The true, fully-loaded cost shape and the real drivers of platform choice.	A cost-stack model + a platform-selection logic (the ‘gravity’ factors).	CFOs & CIOs — build defensible business cases and platform decisions.
P5 · Trust, Evaluation & Governance	The distance between governance on paper and governance in practice, with a maturity read.	Evaluation–governance maturity placement + a regulatory-readiness picture.	GC, CRO & boards — turn obligations into operational practice.

3 · Objective-Level Outcomes — the Detailed Yield

For each objective: the specific evidence the interviews will produce, and the tier that draws it out. This is the granular promise of the instrument — twenty objectives, twenty yields.

PILLAR 1 · Beyond the Model
Obj.	Outcome — what the evidence will yield	Drawn from
P1-O1	A measured read on the adoption-vs-impact gap — what share of initiatives actually reach production and show P&L impact, tested against the ~95%-no-measured-impact baseline.	Holistic
P1-O2	A field-validated map of the structural failure modes — buying tools vs building capability; central AI labs vs line-manager ownership — and which dominate in practice.	Holistic
P1-O3	Each organisation’s position on the maturity ladder (experiment → pilot → scaled → embedded → AI-native), and where the cohort clusters.	Both tiers
P1-O4	A grounded forward research agenda — the unanswered questions leaders most want resolved over the next 12–24 months.	Granular

PILLAR 2 · Production AI Architecture
Obj.	Outcome — what the evidence will yield	Drawn from
P2-O1	Which retrieval patterns are actually deployed (naive → hybrid → reranking → graph) and which correlate with reliable, faithful answers.	Granular
P2-O2	Where systems sit on the five-tier agentic taxonomy — how autonomous they really are, and how failure is handled.	Granular
P2-O3	Evidence on which tiers and patterns are production-ready for which workloads.	Holistic
P2-O4	The real integration and infrastructure constraints (legacy systems, technical debt) that gate production — the inputs to the decision framework.	Granular

PILLAR 3 · Operating Models
Obj.	Outcome — what the evidence will yield	Drawn from
P3-O1	The distribution of operating-model archetypes across the cohort (centralised vs federated × platform vs delivery).	Holistic
P3-O2	Which of the six success conditions are present in scaling programmes versus stalled ones.	Granular
P3-O3	Field analogues to the case studies (JPMorgan, Sanofi, Walmart, Klarna, McDonald’s) — which patterns repeat.	Both tiers
P3-O4	Which of the five operating-model components are weakest, and how talent and funding flows shape outcomes.	Granular

PILLAR 4 · Cost & Platform Landscape
Obj.	Outcome — what the evidence will yield	Drawn from
P4-O1	A fully-loaded cost picture — how organisations budget and meter the eight cost categories, and how visible the non-inference 60–80% really is.	Both tiers
P4-O2	Where budgets systematically break down by workload archetype and scale tier — the overrun pattern.	Holistic
P4-O3	How platform shortlists are actually formed and which evidence buyers trust, including exposure to global and lower-cost models.	Granular
P4-O4	The real decision drivers — data residency, accreditation, and existing stack (‘gravity’) versus raw model capability.	Granular

PILLAR 5 · Trust, Evaluation & Governance
Obj.	Outcome — what the evidence will yield	Drawn from
P5-O1	What organisations actually measure (the seven dimensions) and which evaluation methods they use in production.	Granular
P5-O2	The gap between documented governance and implemented governance — the ‘compliance theatre’ signal.	Both tiers
P5-O3	How regulatory obligations (EU AI Act, NIST AI RMF, ISO/IEC 42001) are translated into delivery, and where teams get stuck.	Holistic
P5-O4	Each organisation’s evaluation–governance maturity placement on the four-level ladder, and the blockers to the next level.	Granular

4 · Cross-Cutting Outcomes

Six cross-cutting questions (three per tier) capture themes that sit between pillars and let findings be triangulated rather than read in isolation:

• People, skills & adoption — How the workforce is responding and whether capability is being built — the human signal behind P1-O2 and P3-O4.
• Value & outcomes — Whether value is measured in business terms or mere activity — reconciling P1-O1, P4-O2, and P5-O1.
• Failure, incidents & recovery — How systems fail and recover in production — a stress-test of P2-O2 and P5-O2.
• Build vs buy & technical debt — The make-or-buy logic and lock-in risk shaping the stack — P2-O4 and P4-O4.
• Measuring what matters — How close current metrics sit to real outcomes — P1-O1 and P5-O1.

5 · How the Outcomes Will Be Used

Audience	What they get from it
The publication	Evidence-based articles for The Applied Layer series, the cross-pillar synthesis, and a defensible forward research agenda.
Participating organisations	A confidential benchmark — where they sit versus peers on each pillar, and their largest, most addressable gaps.
The wider field	A corrective to hype — what is actually working in production, grounded in primary evidence rather than vendor claims.
The research itself	A reusable, objective-tagged instrument and dataset — a baseline that can be re-run to measure change over time.

6 · Why the Yield Is Complete by Design

Coverage. Every interview question maps to one or more objectives, and the coverage matrix in the questionnaire confirms all twenty objectives (P1-O1 … P5-O4) are asked. No objective is left unmeasured.Breadth and depth. The holistic tier sweeps all five pillars for breadth; the granular tier goes deep on the respondent’s home pillar. Across the respondent pool every pillar is probed in depth, so the dataset is complete by construction, not by chance.Traceability. Because each response carries its objective ID, the analysis can be coded straight back to the aims — turning interview transcripts into structured, comparable evidence.