Pilots that convert

Why pilotitis happens

Module 07 named the disease; here is the pathology, because the causes are structural rather than personal:

• A pilot needs no recurrent budget. Non-recurrent money — year-end underspend, a transformation pot, national programme funding — can fund one without troubling the investment committee, while scaling needs recurrent revenue, which does. The pilot is easy precisely because it dodges the hard decision.
• Nobody owns scaling. The clinician owns the trial, the transformation team owns the project, adoption belongs to no one. Evaluation capacity is scarce, so pilots end with anecdotes. And testing is rewarded while adopting is not: innovation activity is visible and reportable; quiet scale-up is neither.
• People move, so over twelve months expect one of your champion, sponsor or programme manager to change role. And a programme-funded pilot hits the national funding cliff when that money ends, with the local recurrent conversation started too late.

Voice from the field

At NHS ConfedExpo in June 2026, NHS England's national clinical lead for AI, Dr Shankar Sridharan, called the continued local piloting of ambient voice technology "nuts", arguing that with the evidence available "the time has come to move together" — a rare instance of the national centre telling providers to stop testing and start adopting.
— reported by Digital Health, June 2026

Voice from the field

"I see so many pilots, so many pilots." … "At what point does a pilot become a test of change?"
— Helen Balsdon, national Chief Nursing Information Officer, NHS England, speaking at NHS ConfedExpo, June 2026 (reported by Digital Health)

The pilot contract terms that protect you

Never run a pilot on an email exchange. A short written agreement — two to four pages — should cover:

1. Scope and sites: which service, how many users, which patients, what is out of scope.
2. Duration and a hard stop date, with a decision point before the end, not on it.
3. The named executive sponsor and the named decision-making forum.
4. Success criteria in writing, with baseline measurement, data source, collector and dates. If the baseline is not measured before go-live, the pilot cannot prove anything.
5. Evaluation ownership: who analyses, who writes it up, whether you see and comment on results, who may publish.
6. Data terms: controller/processor roles, what data you may use for service improvement or model training (state it explicitly — silence becomes a fight later), retention, return or deletion on exit.
7. IP, price at scale fixed now (agreed during the pilot, not afterwards from a weak position), and the conversion mechanism — if criteria are met, a named route and an indicative timetable.
8. No exclusivity without payment, no unlimited free extensions, and clear termination and exit terms including safe withdrawal from a live service.

Success criteria that map to the business case

The test of a good pilot metric: would this number appear in the trust's own reporting? Metrics drawn from board reports, national targets, the CIP or the risk register are credible to the committee that decides; metrics you invented are not. Good criteria are specific, baselined, attributable and few — typically one outcome, one operational, one safety and one adoption metric — and include a counterfactual. Agree in advance what would count as a failure, because criteria that cannot fail persuade nobody.

Who signs off conversion

Larger commitments go through a business case, often in stages — strategic outline, outline, then full — reviewed by a finance or investment committee against delegated approval limits, and by the board above them; centrally funded schemes need central approval regardless of value. Understand which committee and when it meets: they sit monthly or bi-monthly with paper deadlines a fortnight before, so a missed deadline is a six-week delay and three of those is your quarter. And your champion will not write the case alone — offer to draft the benefits section, evidence appendix and risk table, unbranded.

How to price a pilot so it doesn't cannibalise the deal

• Free is not neutral. It signals your product is worth nothing, attracts organisations with no budget, and gives nobody internal a reason to fight for money.
• Charge something, even if modest. A paid pilot needs a purchase order, which needs a budget-holder, which surfaces the funding question in month one rather than month eleven.
• Price as a fraction of year one, with an explicit credit against the full contract if it converts within a defined window: the buyer gets a discount for deciding quickly, you get a decision. Charge separately for evaluation support if you are supplying analyst time.
• Keep it short. Three to six months. Twelve-month pilots outlive their sponsors.
• When they ask for a free extension: decline warmly and offer a paid extension with written conversion criteria and a scheduled decision. If there is no sponsor and no budget, qualify out. A pilot without an owner is cost, not pipeline.

7. The evidence ladder

Every NHS buyer asks some version of "does it work, does it work here, and can we afford it?" The ladder is how you answer, at increasing cost.

Rung 0 — the mechanism. A clear logic model: what the product does, what changes in the pathway, what outcome follows, what must be true for that chain to hold. Costs nothing, and a supplier who cannot articulate it will not survive a CCIO's questions.

Rung 1 — case studies and reference sites. Named organisations, named contacts, specific numbers, honest about context. Weak evidence, but enough for a first meeting — and a reference call with a peer at a comparable trust remains one of the most persuasive things in the process.

Rung 2 — service evaluation. The rung small suppliers most often skip and should not. Learn the distinction the NHS draws between audit (practice against an existing standard), service evaluation (a service as currently delivered) and research (new generalisable knowledge): research requires ethics approval and a formal governance route; the other two generally do not. A well-designed service evaluation with a proper baseline, pre-registered criteria and an NHS co-author is cheap, fast, and answers "does it work here?" — the question the buyer actually asked.

Rung 3 — health economics. A budget impact model answers "what does this do to my budget over three years?" — the question a trust finance director and an ICB actually ask — and should separate cash-releasing from non-cash-releasing benefit and let the buyer change the assumptions. A cost-effectiveness analysis answers "is this good value per unit of health gained?", typically cost per QALY: the language of NICE and national decision-making, not a trust's monthly finance meeting. Small vendors overinvest in the second and underinvest in the first. Build the budget impact model early; it is what unblocks local deals.

Rung 4 — peer-reviewed publication. Slow and expensive, but it converts sceptical clinicians and satisfies national evidence frameworks as nothing else does. Co-author with the NHS site that generated the data.

Rung 5 — national evaluation and NICE routes. As of 2026:

• NICE's HealthTech programme consolidated the previously separate medical technologies evaluation, diagnostics assessment and interventional procedures programmes into one lifecycle programme, with a manual published in July 2025, assessing technologies at early use, routine use and existing use — and the "early use" methods replace the earlier interim Early Value Assessment approach. If you learned "EVA" as a name, the concept persists inside the early-use route; check what NICE currently calls it before saying it out loud.
• The National HealthTech Access Programme, announced in February 2026 by NICE, DHSC, NHS England, the MHRA and the Office for Life Sciences, aims at faster national access routes (check current status — this is new).
• The Evidence Standards Framework (ESF) remains NICE's statement of what evidence a digital technology of a given risk tier should hold, and is being adapted for adult social care. The MedTech Funding Mandate still obliges commissioners and providers to fund selected NICE-recommended cost-saving technologies, with no new products added for 2026/27 pending review.

What each stakeholder actually wants

Clinician: does it work in a population like mine, is it safe, does it cost me time — peer-reviewed evidence or a credible local evaluation, plus a peer to vouch. CCIO/CNIO: workflow fit, clicks, alert burden, adoption data. CIO: integration evidence, support model, security posture, references from similar estates. IG/DPO: not evidence of effect at all, but of lawful, minimised, controlled processing. Finance: a budget impact model with transparent assumptions. Operational manager: throughput and capacity effects, and what it costs their staff to run. Board: alignment with strategy and the risk register, deliverability, reputational safety.

How a small vendor builds evidence cheaply

• Design evaluation into the first deployments. Baseline before go-live; agree metrics in the pilot agreement. Retrofitting a baseline is impossible.
• Use routinely collected data the trust already reports, not bespoke collection nobody has time for.
• Partner locally. A Health Innovation Network, university department or interested NHS analyst supplies methodological credibility you cannot buy — and their name on the report is worth more than yours.
• Pre-register your criteria so results cannot be dismissed as chosen afterwards, and publish service evaluations in accessible venues; not everything needs a high-impact journal.
• Chase non-dilutive funding: SBRI Healthcare competitions exist precisely to fund evidence generation for small companies.
• Be honest about limitations. A candid evaluation with stated caveats fares far better with NHS reviewers — usually clinicians trained to find the caveats you omitted.

The most common evidence mistake in health tech is generating the wrong evidence expensively: eighteen months and a large sum spent on a cost-per-QALY analysis, and the finance director asks what it does to her pay bill next April. Match evidence to decision-maker — national access routes want cost-effectiveness; local buyers want budget impact, safety and a peer who will take their call.