Co-creation, in practice, tends to mean getting enough people to nod that nobody can later say they weren’t asked. The workshop ends, the objections are noted, the design moves forward largely as planned, and everyone involved has technically been consulted. It looks participatory. It isn’t.
The trouble with treating consensus as the goal is that consensus is a low bar dressed up as a high one. Getting a room of people to agree, or at least not object, is achievable by watering the design down until nobody minds it. It rewards the safest version of every decision, and it quietly punishes the person willing to name the uncomfortable thing, because their objection now stands between the group and the appearance of alignment. Chase surface consensus and you will get it. What you’ll have designed is something nobody hates, which is a different thing from something that works.
There’s a second version of this mistake, and it looks like the opposite instinct while landing in the same place. This is the leader who takes co-creation seriously enough to wait for everyone to actually agree, properly, not just stop objecting. That patience sounds like rigour. In practice it produces one of two outcomes. Either the process runs until every stakeholder’s preference has been folded in somewhere, and the design grows a feature or a safeguard for each of them until it’s carrying more structure than the problem needed, or the group converges by trading away whatever the loudest disagreement was attached to, because that’s the fastest route to genuine agreement. Both outcomes have the same root: real agreement, chased as the goal itself, gets reached by addition or by subtraction, and neither one is design.
What I’ve seen hold up is a different aim: getting to the best answer, with everyone genuinely heard. Those are not the same thing. Hearing someone and agreeing with them are different acts, and conflating the two is what produces both failures above. Treat hearing as agreeing and you end up chasing agreement itself, either the fast, hollow version or the slow, over-negotiated one. Separate the two, and involvement stops being a vote. It becomes a way of getting at what the room actually knows, which the leader then has to weigh and decide on, agreement or not.
Genuine co-creation asks more of the person leading it, not less. It means going into the room without the answer already decided, and being willing to leave with a different one than you walked in with. It means treating disagreement as information rather than as a problem to be managed down to zero. And it means being honest, afterward, about which input actually changed the design and which didn’t, rather than letting everyone assume their voice shaped the outcome equally.
There’s a practical payoff to this, separate from whether it produces a better design. People who have been genuinely heard, and who can see their point reflected somewhere in the reasoning even when the final decision goes against them, are far more likely to get behind the outcome. The objection that actually stops people committing is rarely the decision itself. It’s the suspicion that no one understood their piece of it before deciding. Genuine hearing closes that off. Someone can still think the decision is wrong. What they lose is the ability to say it was made in ignorance of what they raised, and that difference is what turns disagreement into something people will live with rather than work around or actively resist.
This matters more than it used to now that we are talking about automating parts of the business or handing them over to AI agents. The people closest to the work are holding parts of the operating model that were never written down anywhere: the workaround that quietly keeps a broken handoff functioning, the judgement call that one person makes without anyone else knowing it’s being made, the reason a process that looks inefficient on paper is actually protecting something real. None of that shows up in a process map. Most of it doesn’t surface in a process done to people rather than with them.
Consensus-seeking and top-down design fail the same way from opposite directions. One collects the invisible detail and then sands it down until it disappears into agreement. The other never collects it at all. Either way, you end up redesigning something that won’t actually work, and finding out what you missed later, at a worse time, in a more expensive form.
There’s a simple test for whether a co-creation process did its job. It isn’t whether everyone left the room happy. It’s whether someone left having said the thing they were nervous to say, and watched it actually land somewhere in the design. That’s a much higher bar than consensus, and a much better one.