Innovation with a P&L attached
Every large hospital group I know has had an innovation team, and most of them have had two, because the first one was quietly dissolved. The pattern is consistent. A board wants the group to be seen as forward-looking. A team is formed, a floor or a lab is fitted out, a startup programme is announced, and within eighteen months the team’s principal output is a tour. Visitors see a robot, a wearable, a screen with a dashboard on it. The finance pack does not change.
That is the difference between a programme and a hobby. A programme has a P&L line it expects to move and a date by which it expects to move it. A hobby has a demo. Both can be called innovation, and only one survives a cost review.
I have built the cases for innovation spend in a group that measures everything in occupancy, sat at the gates where projects were funded or stopped, and kept one alive longer than I should have because a board member liked it. This is what keeps an innovation programme honest, and what it costs to do that.
Why innovation teams drift into showcases
Nobody sets out to build a showroom. The drift happens because the team is measured on activity — pilots launched, partnerships signed, awards, mentions in the annual report — rather than on a number anyone else in the group is measured on. Activity is easy to produce and easy to present. The innovation team’s year-end review is a list of things started, and no one asks what finished.
It is fed by three sources of projects that arrive without a P&L. Vendor showcases, which come with a free pilot and an agenda. Doctors’ pet projects, which come with a champion nobody wants to disappoint. And board or promoter interest, which comes with a photograph. All three can be good projects. None of them arrives with the question the programme should be asking first: which line in the management pack changes if this works?
And it is protected by how it is funded: as a cost centre with no expected return, so it is never held to one. A team that is not expected to move a number will not, and will be closed in the first cost programme by a CFO who cannot say what it was for.
Gate zero: a problem with a number
The discipline starts before a project exists. No idea enters the programme without a named P&L line — occupancy at a unit, average length of stay, revenue per occupied bed, cost per discharge, receivable days, contact-centre cost per appointment, nurse hours per patient-day — a stated direction, and a unit head or function head who says that line matters to them and would notice if it moved.
This kills most showcase projects at the door, which is the point. A wearable that produces data nobody is waiting for does not have a line. An AI tool that saves nurse minutes has a line only if someone will take those minutes out of the roster, and it is the roster owner, not the vendor, who has to say so.
Gate zero also fixes the kill criteria. What result, by what date, means this stops? Writing that down before anyone is invested is the only time it can be written honestly. Six months in, with a vendor relationship, a doctor champion and a slide in the board deck, nobody can.
Growth’s own lines — enquiry to appointment, cost per appointment, repeat visits — are not exempt. A pilot that claims to have improved conversion using its own dashboard has not moved a line until the CRM and the finance pack agree it has. The innovation team gets the same reconciliation the executive committee demands of everyone else’s number, and it is usually the innovation team that objects.
The gates after that
Gate one is a pilot in one unit, with the unit head as owner, a defined period, and the innovation team in support rather than in charge. A pilot the unit does not own is a pilot the unit will not adopt, however well it goes.
Gate two is evidence that the line moved. Not satisfaction scores, not usage, not a testimonial from the doctor champion. The line, in the unit’s own management pack, against the period before, with the unit head and the CFO’s analyst agreeing on the reading. If the line did not move, the pilot ends here. Most should.
Gate three is scale — and the rule that matters most is that scale is funded from the operating budgets of the units that adopt it, not from the innovation budget. The moment a unit head has to find the money in their own P&L, you learn whether the pilot’s result was real. A project that can only be scaled with innovation money has not proved anything.
Gate four is the innovation team leaving. The project belongs to operations, IT, growth or clinical governance now. An innovation team that keeps its successes is a team that has stopped innovating and started operating, and it will be measured on that operating soon enough.
Who sits at the gate
Not the innovation team alone, ever. The gate is the CFO or a delegate, the medical director or a delegate, one unit head, and growth. Each has a reason to be there. Finance because the line is theirs. Clinical governance because the risk is theirs, and because a pilot that touches a patient is a clinical matter before it is anything else. The unit head because the adoption is theirs. Growth because a large share of hospital innovation is really about demand or the patient journey, and because somebody has to say whether the number is a number.
The innovation team presents. It does not vote. I have seen programmes where the head of innovation chaired their own gate, and every project passed.
What a real innovation budget looks like
Not a percentage of revenue. That is a number chosen to sound serious and it produces a budget that must be spent. A real budget is small, ring-fenced from corporate allocation so the programme is not carrying a share of the office, and released by gate rather than by year. Gate zero costs nothing but time. Gate one is funded per pilot, with a cap. Gate three is funded by the adopting units. What the innovation budget actually pays for is the team, a handful of pilots at any one time, and the evaluation work that most groups skip because it is unglamorous.
Two things must be in the budget paper for a group that measures everything in occupancy. The first is the expectation of failure, in writing: that most pilots will stop at gate two, that this is the programme working, and that the CFO agrees not to treat a killed pilot as wasted spend. The second is the line between innovation opex and the capital list. A pilot that needs a capital item goes through the capital queue like a cath lab does, with the same scrutiny. Innovation is not a side door for capex, and the first time it is used as one the CFO will close the programme.
The programme’s own number is the share of pilots that reach gate three and the movement in the lines they touched. A programme that has never killed a pilot is not running gates. A programme that has never passed one is not choosing problems.
Where projects hide their P&L
Cost avoidance is the favourite. “This will save the equivalent of four nurses.” It will not, unless four nurses are not hired, and the roster owner has to say that at gate zero. “Patient experience” is the second: real, but it needs a line — repeat visits, reviews, complaint volume, contact-centre load — or it is an adjective. The third is the pilot whose gain sits in a different department from its cost. The AI that saves radiologist time costs the IT team an integration and the unit a workflow change; the gate must see both sides of the ledger, and usually only one side is presented.
The most respectable hiding place is the future. “This does not pay back now but positions the group for what comes next.” Sometimes true. It still needs a gate two, even if gate two is two years out, and it needs the CFO to agree the date before the money moves.
How to kill a project without killing the appetite
The fear that stops most groups from running gates properly is that killing projects will make people stop bringing them. The opposite is true, if the kill is done well. What makes people stop is watching a project die in a corridor, six months after everyone knew, with the champion blamed.
Kill at the gate, against the criteria written at gate zero, on the date agreed. Not before, not after. Say the criteria were not met, not that the idea was wrong. Write a one-page note of what was learnt — about the problem, the unit, the vendor, the data — and circulate it, because the learning is the return on the pilot. Say publicly that the programme stopped it and why. Move the people to the next pilot the same week. And take particular care with the doctor champion: a consultant who backed a pilot that stopped must be seen to have done the right thing by backing it, or no consultant will back the next one.
The one I kept alive too long taught me the cost of not doing this. A board member had seen the demo. The gate-two evidence was thin and I let it through on the argument that the line would move once adoption improved. Adoption did not improve. The project consumed a year of a unit’s goodwill and the programme’s credibility with the CFO, and when it finally stopped, the champion left. The board member did not remember the demo. I would kill it at the gate now, and tell the board member first.
The kill rate is worth reporting to the executive committee every quarter, alongside the pilots that passed. A committee that sees pilots stopped on schedule, with a one-page note attached to each, stops asking whether the programme is a hobby.
If you are setting this up next quarter
- Write the gate criteria on one page and get the CFO and medical director to sign them before the first project is chosen.
- Take the existing portfolio through gate zero retrospectively. Anything without a line and an owner stops now.
- Rebuild the budget as gate-released, ring-fenced and small, with the failure expectation in the paper.
- Name the gate committee and put the innovation head in the presenting seat, not the chair.
- Choose the first three pilots for their lines, not their demos. One should be unglamorous — length of stay, discharge, receivables.
- Schedule the first kill review before the first pilot starts, so it is a date on the calendar and not a confrontation.
A group can afford innovation that fails. It cannot afford innovation that never finds out.
Questions people ask
Because they are measured on activity — pilots launched, partnerships signed, awards, annual-report mentions — rather than on a number anyone else in the group is measured on. Projects arrive from vendor showcases, doctors’ pet projects and board interest, none of which come with a P&L line. And the team is funded as a cost centre with no expected return, so it is never held to one and gets closed in the first cost review.
It is the discipline that no idea enters the programme without a named P&L line — occupancy, length of stay, revenue per occupied bed, cost per discharge, receivable days, contact-centre cost per appointment — a stated direction, and a unit or function head who says that line matters to them. Gate zero also fixes the kill criteria in writing before anyone is invested, because six months in, with a vendor and a doctor champion, nobody can write them honestly.
Gate one is a pilot in one unit, owned by the unit head, with the innovation team in support. Gate two is evidence that the line moved in the unit’s own management pack, with the unit head and the CFO’s analyst agreeing on the reading — not satisfaction scores or usage. Gate three is scale, funded from the adopting units’ operating budgets. Gate four is the innovation team leaving and the project belonging to operations, IT, growth or clinical governance.
The CFO or a delegate, because the line is theirs. The medical director or a delegate, because clinical risk is theirs and anything touching a patient is a clinical matter first. One unit head, because adoption is theirs. And growth, because much hospital innovation is really about demand or the patient journey. The innovation team presents and does not vote. Where the head of innovation chaired their own gate, every project passed.
Not a percentage of revenue — that number is chosen to sound serious and produces a budget that must be spent. A real budget is small, ring-fenced from corporate allocation and released by gate rather than by year. Gate zero costs only time. Gate one is funded per pilot with a cap. Gate three is funded by adopting units. The innovation budget pays for the team, a handful of pilots at a time, and the evaluation work most groups skip.
As the programme working. The budget paper should state in writing that most pilots will stop at gate two, and the CFO should agree not to treat a killed pilot as wasted spend. What the CFO should refuse is innovation used as a side door for capital: a pilot needing a capital item goes through the capital queue with the same scrutiny as a cath lab. The first time innovation bypasses that, the programme deserves to be closed.
Kill at the gate, against the criteria written at gate zero, on the agreed date — not before, not after. Say the criteria were not met, not that the idea was wrong. Write a one-page note of what was learnt and circulate it. Say publicly the programme stopped it and why. Move the people to the next pilot the same week. And protect the doctor champion, who must be seen to have done the right thing by backing it.
A consultant who backs a pilot lends it credibility with colleagues and takes a personal risk if it stops. That is why the kill has to be handled carefully: if the champion is blamed when a pilot ends at gate two, no consultant backs the next one. The programme should say the clinician did the right thing, publish the learning, and move on. Clinical governance also sits at the gate because any pilot touching a patient is a clinical matter first.
Cost avoidance is the favourite — a tool will save four nurses, but only if four nurses are not hired and the roster owner says so at gate zero. Patient experience is the second; it needs a line like repeat visits or complaint volume or it is an adjective. The third is a gain in one department with the cost in another, such as radiologist time saved at the price of an IT integration. The most respectable is the future, which still needs a dated gate two.
The share of pilots reaching gate three, alongside the movement in the P&L lines they touched. A programme that has never killed a pilot is not running gates. One that has never passed a pilot is not choosing problems. Report the kill rate to the executive committee every quarter with a one-page note per stopped pilot. A committee that sees pilots stopped on schedule stops asking whether the programme is a hobby.
A single hospital can run it, and arguably more easily, because the unit head, the CFO and the medical director are the same three people at every gate and the management pack is one document. The risk in a single hospital is that the gate committee is too close to the champion to kill anything. Write the criteria down before the first project, put the kill review on the calendar, and treat the first stopped pilot as proof the system works.
I kept a pilot alive too long because a board member had seen the demo. The gate-two evidence was thin and I let it through on the argument that the line would move once adoption improved. Adoption did not improve. The project consumed a year of a unit’s goodwill and the programme’s credibility with the CFO, and when it stopped, the champion left. The board member did not remember the demo. I would kill it at the gate now, and tell the board member first.
No free pilots that enter without a P&L line and an internal owner. A defined pilot period in one unit, with the unit head in charge rather than the vendor or the innovation team. Evidence judged from the hospital’s own management pack, not the vendor’s dashboard. A written kill date. And scale funded by adopting units, which means the vendor’s commercial case has to survive a unit head finding the money in their own budget.
Write the gate criteria on one page and have the CFO and medical director sign them before choosing a project. Take the existing portfolio through gate zero retrospectively; anything without a line and an owner stops. Rebuild the budget as gate-released, ring-fenced and small, with the failure expectation in the paper. Name the gate committee with the innovation head presenting, not chairing. Choose three pilots for their lines, one unglamorous. Schedule the first kill review before the first pilot starts.
