Innovation Theatre: Why the Lab’s Ideas Never Reached the Core
An idea without a funded path to production has not been approved; it has merely been admired.
The morning after the demo day
There is a particular quiet that settles over an innovation lab the week after its launch. The photographs have been taken — the exposed brick, the writable walls, an executive cutting a ribbon in front of a wall of coloured sticky notes. The demo day has happened: a dozen prototypes, each shown for four minutes to a room of senior sponsors who applauded and meant it. And then, on the Monday, the people who will actually have to turn any of this into something a customer can use go back to their desks two floors down, or in another building, or in another city entirely — and hear nothing further about it for months.
That gap, between the applause and the silence, is the subject of this piece. Over the past two years the corporate innovation lab has become the standard organisational response to digital disruption. Every board has seen the same slides; nearly every large organisation now has, or is building, its own version — the lab, the garage, the digital studio, the in-house accelerator. The prevailing story is that these places exist because the core cannot innovate: it is too slow, too risk-averse, too encrusted with process, and so the new must be grown somewhere protected from the old.
I want to argue something less comfortable, and to argue it from the delivery side — from the vantage of the teams who were meant to receive what the lab produced. The lab was rarely short of ideas. What it lacked was any working connection to the organisation that would have to absorb them. The failure now becoming visible across the sector is not a failure of creativity. It is a failure of wiring. Innovation theatre, the phrase that has begun to circulate, is in one sense too kind: it implies the only problem is that the lab is for show. The harder truth is that even when the lab is sincere — even when the ideas are genuinely good — the path from prototype to production was never built.
The country with no roads out
The lab is usually set up as a separate country. That is done deliberately, and for defensible reasons. Put the new venture inside the core and the core’s immune system attacks it: the funding gets raided to cover an overspend elsewhere, the best engineers get pulled onto a regulatory deadline, the governance forum that meets to approve a customer trial spends its time debating a change-control form. So the lab is given its own space, its own budget line, its own lighter-touch process, and often its own reporting line straight to a newly appointed digital chief. Separation is presented not as a side effect but as the whole point.
The strongest version of this argument deserves to be taken seriously, because it is largely correct about the disease. Core organisations do smother fragile ideas. The antibodies are real, and they are not stupid — they exist because the core carries obligations the lab does not: uptime, auditability, the promise made to millions of existing customers that the thing they rely on will still work tomorrow. A team optimising for those obligations will, quite rationally, treat an untested prototype as a threat. So yes: give the new some protection from the old. On that much, the orthodoxy is right.
Where it goes wrong is in mistaking protection for severance. Protecting a young idea from premature governance is not the same as building it in a place with no roads leading back. And that is what we built. The lab was given every input it needed — talent, money, autonomy, a mandate — and almost no defined output path. There was a lavish on-ramp and no off-ramp. Seen from delivery, this was obvious from the first handover, and it was structural, not attitudinal. Nobody was refusing to cooperate. There was simply no mechanism by which cooperation was supposed to happen.
The productionisation tax
Consider what actually reaches the delivery organisation when a prototype “graduates”. It arrives as a working demonstration built for a stage, not for a customer: it runs on a laptop or a single cloud instance, against a copied slice of data, with the authentication stubbed out, the error handling absent, the integration to the systems of record faked with a spreadsheet behind the curtain. This is not a criticism of the lab — a prototype is supposed to cut those corners; that is what makes it fast. The problem is what the graduation ceremony implies: that the thing is now nearly done, when in fact the hardest, least glamorous nine-tenths of the work has not started.
I have watched this arithmetic play out closely enough to put numbers on it. In one composite that is faithful to several I have seen, a lab produced fourteen prototypes over eighteen months. Two were formally endorsed for “productionisation”. Of those two, both were rebuilt essentially from scratch by the core delivery team, because none of the prototype code could survive contact with the organisation’s identity platform, its data-protection controls, its release process, or its operational support model. The lab reported fourteen innovations delivered. The core absorbed the cost of two rebuilds and quietly wrote off the other twelve. Nobody, anywhere on a slide, recorded a productionisation tax — the multiple of the original build cost required to make a demo into a service the organisation can actually run and stand behind.
That tax is where the disconnection does its damage. It is paid entirely by the delivery side, out of a budget that was never expanded to cover it, on a timeline nobody planned for. And because it is invisible in the lab’s own metrics, it is never designed down. The lab is measured on throughput — ideas generated, prototypes shipped, demo days held — while the organisation is served only by adoption, which no one is measured on at all.
| What the lab optimised for | What the core actually needed |
|---|---|
| Ideas generated and prototypes demonstrated | Services adopted and running in production |
| Speed to a working demo | Survivability after handover |
| Autonomy from the core’s process | A designed path back into the core’s process |
| A protected budget for building | A funded budget for the last, unglamorous mile |
| Novelty as the measure of success | Absorption as the measure of success |
What the textbooks got wrong
The literature of the last few years told us to worry about the wrong risks. It warned, at length, that the core would strangle the new, and it prescribed autonomy as the cure. It said far less about the mirror-image failure that has turned out to be at least as common: the new, granted its autonomy, drifting so far from the core that nothing it makes can ever come home. We were handed a great deal of advice on how to start an innovation function and almost none on how to land its output.
The lean-startup thinking that shaped these labs is not wrong — building a minimum viable product, testing it against real users, and iterating is sound. But a discipline built for founders who are the whole company was transplanted into organisations where the team that validates an idea and the team that must operate it for a decade are entirely different people, in different reporting lines, with different incentives. In a start-up, the person who runs the experiment inherits its consequences. In our labs, the person who ran the experiment moved on to the next one, and the consequences were inherited by a platform team that had never been in the room.
The orthodoxy asked, how do we protect the new idea from the old organisation? The question that actually decided outcomes was the one almost nobody asked: how does the new idea get back into the old organisation without being rebuilt from nothing?
There is also a quieter cultural cost, and delivery teams felt it first. When a lab is celebrated for novelty while the people who make things reliably work are treated as the drag on progress, you teach your best engineers that the interesting, well-resourced, visible work happens somewhere they are not. The core does not just miss out on the prototypes. It slowly loses the belief that it is where real work is done — and that belief, once gone, is expensive to rebuild.
Wiring the lab to the core
If the failure is one of wiring, then the remedy is not to abolish the lab — the separation solved a real problem — but to build the roads out that we never built in. In practice that has meant a small number of unglamorous things.
- Measure the lab on adoption, not output. The single most powerful change is to stop counting prototypes and start counting services that are live and in use twelve months later. A lab scored on adoption cannot treat the handover as someone else’s problem, because the handover is now the only thing that moves its own numbers.
- Put delivery in the room from the start. An enterprise architect or a platform engineer embedded in the lab from the first sprint is not the antibody arriving to kill the idea; they are the person who ensures the idea is buildable against the identity platform, the data controls, and the release process while it is still cheap to change. Their early “no” is worth ten late ones.
- Fund the last mile explicitly. The productionisation tax should be named, estimated, and budgeted before a prototype graduates — not discovered afterwards by the team left holding it. An idea without a funded path to production has not been approved; it has merely been admired.
- Make graduation a joint act, not a throw over the wall. The lab team should stay attached to a prototype through its first period in production, sharing the pain of the rebuild it caused. Nothing improves the realism of a prototype faster than knowing you will personally help operate it.
None of this is exotic. It is, mostly, the ordinary discipline of transformation applied to a function that was allowed to imagine it was exempt. The lab was exempted from the core’s process on the theory that the process was the enemy. Some of it was. But the parts that make a service survivable — the identity, the controls, the operability, the support model — were never the enemy. They were the destination.
The measure that would have changed everything
If I could return to the launch day of any of these labs, ribbon and sticky notes and all, I would change one thing before the first prototype was ever built. I would define, in advance, what it means for the lab to have succeeded, and I would make that definition adoption by the core rather than activity within the lab. Everything else — the embedded architects, the funded last mile, the shared pain of the rebuild — follows almost automatically once the lab’s success is tied to the core’s absorption instead of divorced from it.
The corporate innovation lab was never a bad idea. Seen from the delivery side, its tragedy is narrower and more fixable than the innovation theatre verdict suggests. We did not fail to have ideas. We built a place to have them and forgot to build the road home. The organisations that will get value from the next wave of this will not be the ones with the best-photographed labs. They will be the ones that treated the handover, not the demo day, as the moment that mattered.