Module 8 · 95 min
AI Strategy, Portfolio & Economics
Where to Place the Bets, and Why. Opportunity and use-case prioritisation, portfolio choices, buy/build/partner decisions, vendor strategy and due diligence, total cost of ownership, ROI and value realisation, funding and operating-model choices. The central question: what should we pursue, and why?
- Frame AI strategy as a set of choices under scarcity — capital, leadership attention, integration capacity, clinical change capacity and data access — rather than as a list of technologies to adopt.
- Turn an AI idea into a strategic opportunity by attaching it to a service problem, a value hypothesis and a named route to realising the value.
- Compare candidate use cases on value potential, strategic fit and reuse, feasibility, evidence confidence and delivery burden, without collapsing them into a single misleading score.
- Use 'high value but not ready' as a deliberate portfolio position with a stated readiness gap, instead of a rejection or an indefinite pilot.
- Build a portfolio that balances near-term productivity, clinical value bets, enabling capabilities and longer-horizon options, and identify the shared capabilities that unlock several use cases at once.
- Name the full cost stack of an AI deployment — acquisition, integration, data preparation, evidence, redesign, training, oversight, infrastructure, compliance, monitoring and exit — and separate fixed from variable and one-off from recurring.
- Distinguish cashable from non-cashable benefit, and trace a claimed benefit through operational mechanism, measurable KPI, financial or clinical consequence and named owner.
- Recognise misalignment between who pays and who benefits, and say what that implies for funding and sequencing.
- Express an economic case as scenarios with stated assumptions and uncertainty rather than as a single-point return figure.
- Choose between buy, build and partner on capability, differentiation and maintenance appetite, and run a vendor diligence conversation covering lock-in, data rights, version control and exit.
- Make and defend fund, fund-conditionally, hold and stop decisions across a constrained portfolio.
From time saved to realised value
'The tool saves each clinician twenty minutes a day' is the most common sentence in healthcare AI business cases, and on its own it states nothing about value to the organisation. Twenty minutes is an input. Whether it becomes value depends entirely on what happens in those twenty minutes afterwards — and that is a management decision, not a property of the software.
If the released time is absorbed into a longer lunch break, the organisation has bought staff wellbeing. That may be a legitimate objective, particularly where attrition is expensive, but it should be claimed as wellbeing and measured as wellbeing, not converted into a cash saving in a spreadsheet. If the released time lets a clinic see two more patients a session, that is capacity — and it is real only if the demand, the room and the supporting staff exist. If it lets the service stop paying for a locum session, that is a cashable saving and will show up in a budget.
The discipline is to say which of those is intended before deployment, and to arrange for it. Released capacity does not redeploy itself.
Seven categories of benefit
Clinical outcomes and quality
Earlier escalation, fewer missed findings, more consistent guideline-concordant care.
The hardest to attribute and often the reason the investment exists. Requires the evidence discipline of Module 6 rather than a benefits spreadsheet.
Capacity
More activity from the same establishment; backlog reduction; shorter waits.
Only real if demand exists to fill it and the surrounding constraints — rooms, staff, diagnostics — move too.
Direct cost reduction
Fewer locum or agency sessions; retired legacy licence; reduced outsourced transcription.
The only category that reliably appears in a budget line. Requires someone to actually remove the spend.
Avoided cost
Fewer readmissions, fewer complaints, less litigation exposure, avoided expansion.
Genuine but rarely cashable in the short term: the ward does not close because admissions fell slightly.
Revenue or access
Additional funded activity, improved coding accuracy, retained referrals — where the local funding model permits.
Entirely dependent on the payment system. In many settings this category is empty and should be shown as empty rather than assumed.
Workforce and patient experience
Lower documentation burden after hours, improved retention, better-understood discharge instructions.
Measurable with instruments and surveys; non-cashable in most cases, and undervalued where attrition costs are high.
Strategic and knowledge value
A reusable capability, a research-ready data asset, organisational skill in deploying and governing AI.
Real and easy to inflate. State it as a capability acquired, not as a monetary figure.
Cashable benefit releases money that can be spent on something else: a post not filled, a contract cancelled, sessions not purchased. Non-cashable benefit is real value that does not free a budget line — time returned in fragments across many people, improved experience, reduced risk. Both belong in the case; conflating them destroys its credibility with a finance director faster than any other error, because the promised saving never appears.
The value-realisation bridge
Between a claimed benefit and a realised one sit three things that are usually missing from the paper: the mechanism by which the change actually happens, the measure that would show it, and the person who has agreed to make it happen. Write every claimed benefit as a five-column row.
- 1. Claimed benefit: The value asserted, stated as a direction rather than a promised number.
- 2. Operational mechanism: The specific change in how work is done that produces it. If this is blank, the benefit is an aspiration.
- 3. Measurable KPI and baseline: What is measured, against what current value, at what cadence — plus a balancing measure.
- 4. Financial or clinical consequence: What follows if the KPI moves: cashable saving, capacity redeployed to a named purpose, or a stated non-cashable outcome.
- 5. Owner: The named person accountable for the mechanism happening — usually an operational leader, not the project manager and never the vendor.
Worked bridge: releasing secretarial capacity in a fictional outpatient service
- Claimed benefit
- Letters reach GPs materially sooner, and secretarial effort per letter falls.
- Operational mechanism
- The clinician signs a drafted letter in clinic, so the dictation-transcription-return loop is removed for the majority of straightforward letters. Secretarial staff move from typing to backlog clearance and complex correspondence.
- Measurable KPI and baseline
- Median and 90th-percentile days from clinic to letter received, against a 4-day median baseline; letters awaiting typing, against a baseline of 620. Balancing measure: sampled letter quality and GP-reported adequacy.
- Financial or clinical consequence
- Non-cashable in year one — the released secretarial hours are redeployed to backlog, not removed. Becomes cashable only if the service subsequently chooses not to replace a departing post, which is a separate decision with its own consequences.
- Owner
- The outpatient operations manager owns the redeployment and the KPI; the clinical lead owns the quality balancing measure.
Written this way, the case is harder to sell and much harder to be wrong about. Note what the fourth row does: it prevents an eight-minute-per-letter saving from being multiplied by an hourly rate and presented as cash that no one will ever find in a budget.
Who pays, who benefits
Costs and benefits routinely land in different places, and the misalignment is structural rather than a sign of bad faith. A discharge-summary tool may be paid for by the acute trust while much of the benefit accrues to primary care and to patients. A monitoring capability is paid for by digital and benefits every clinical service. A tool that reduces admissions may improve outcomes and reduce the paid activity of the organisation buying it, depending on the funding model.
The practical consequence is that a business case has to name who pays and who benefits, and if they differ, say how that is resolved: central funding, a shared arrangement with the beneficiary, or an explicit decision to accept the imbalance because it serves the mission. Leaving it unstated does not make it go away — it surfaces as an unexplained refusal to fund at the last approval gate.
Deep dive: scenarios instead of a single return figureOptional. Why a point estimate misleads, and how to choose where the first decision gate goes.
A single return figure carried to two decimal places is a claim about precision that no healthcare AI investment can support. The adoption rate, the effect size, the volume and the review burden are all uncertain, and multiplying uncertain quantities together produces a number whose confidence interval is wider than the answer.
Present three scenarios instead, with the assumptions visible: a cautious case in which adoption reaches perhaps a third of eligible clinicians and the effect is at the low end; a central case; and an optimistic case. Then state which assumption the answer is most sensitive to. In most documentation cases that single assumption is adoption; in most risk-model cases it is whether the response capacity exists to act on the alerts.
This is also what makes a staged commitment coherent. If the case turns on adoption, the first stage should be designed to measure adoption early and cheaply, and the gate should be set where that measurement arrives.
Exercise: cashable, conditional or not a claim?
Six claims from fictional business cases. Classify each. 'Cashable only if a decision is taken' is the category that matters most in practice — it is where genuine value is either realised by management action or quietly lost.
1. Locum sessions in the reporting rota
Worklist triage is expected to let the department clear its urgent reporting backlog without the four locum sessions per week currently purchased. The service has confirmed it will stop booking them once the backlog measure holds for two months.
2. Twelve minutes per clinician per day
A documentation assistant is expected to save around twelve minutes daily across 240 clinicians. At the £50 per clinician-hour used elsewhere in this module, twelve minutes across 240 clinicians is about 48 hours a day, roughly 10,800 hours over 225 working days — about £540,000 a year, presented as 'efficiency savings'.
3. Two additional patients per clinic session
Released clinician time is intended to increase throughput in a specialty with a substantial waiting list. Rooms are available. The service has not yet agreed to change templated clinic capacity, and doing so would require additional nursing and administrative support.
4. After-hours documentation time
Clinicians in the pilot report substantially less record work after their shift ends. The service considers this important for retention in a group where replacing a departing consultant is expensive and slow.
5. Retiring an outsourced transcription contract
The organisation currently pays an external transcription supplier under a contract with a six-month notice period. If ambient drafting covers the same correspondence, the contract can be allowed to lapse at renewal.
6. Fewer avoidable readmissions
A deterioration model is projected to reduce readmissions, and the case values each avoided readmission at the average cost of a bed day multiplied by the average length of stay.
Sources & evidence · 4 sources
This module cites public or consensus guidance, scholarly literature, technical documentation.
Content reviewed: September 2026. Publication dates of the individual sources are shown in each citation.
World Health Organization. Health technology assessment of medical devices, 2nd ed. Geneva: WHO; 30 May 2025. ISBN 9789240110878.
WHO's practical guidance on health technology assessment as a multidisciplinary evaluation of clinical, economic, organisational, ethical and social implications, intended to support efficient allocation of resources. It frames the discipline of assessing value before adoption; it is guidance on process, not evidence about any particular technology, and it does not prescribe the prioritisation categories used in this module.
Open sourceNational Institute for Health and Care Excellence. NICE HealthTech programme manual (PMG48): early-use HealthTech guidance assessments. Published 14 July 2025; updated 17 December 2025.
Describes how NICE assesses promising health technologies where evidence is still developing — weighing unmet need, potential value for money, the nature of remaining uncertainty and what further evidence generation is required, including workforce and system efficiency impacts. Directly relevant to staged commitment and to treating a weak evidence base as a reason to size a first investment as learning. It is a UK process manual, not a general economic method.
Open sourceNational Institute for Health and Care Excellence. Introducing CHEERS-AI: improving health economic evaluation reporting for AI technologies. NICE blog, 31 October 2024.
Introduces an AI-specific extension to health economic evaluation reporting, covering transparency about input data, model functionality, conflicts of interest and the assumptions behind claimed economic effects. A reporting standard: it improves how an economic case is described and audited, and following it does not make the underlying case correct.
Open sourceBinkley CE, Bouslov D, Zaidi A, et al. An early pipeline framework for assessing vendor AI solutions to support return on investment. npj Digit Med. 2025;8:368. doi:10.1038/s41746-025-01767-z
Proposes an early-stage framework for triaging vendor AI proposals around strategic alignment, shared accountability, measurable value and risk, with an impact/value case covering the problem, the solution landscape, quantifiable objectives and costs. Supports the case for structured early triage before procurement; it is a proposed framework rather than validated evidence that using it improves investment outcomes.
Open source
Capstone Board Pack
Add what you just learned to your own strategy document while it is fresh.