Rebuilding a delivery function in three months.
A global travel loyalty program, delivered through a London agency partner
A global travel loyalty program with a delivery team of more than twenty people. The work was arriving. The team was capable. Delivery was not holding.
I was contracted by the agency to produce a document. A PDF setting out a new project management process, written and handed over.
I told them it could not be done. You cannot change a process you have no visibility into. Writing a process document for a function nobody has observed produces a document, not a change.
I asked for three days on site with the client instead, and a first week spent on interviews, group and individual.
The engagement ran three and a half months.
The diagnostic order does not vary. People first, then culture, then process. Running it in that order is what produces a picture that holds.
People
The first week was interviews. Group first, then individuals.
My opening question in the first group session was simple. How do you deliver projects here?
One person half raised a finger, then lowered it. Nobody in the room could answer.
That is the finding. Not that the process was wrong, that there was no shared account of it. Twenty capable people, each carrying a version of the work in their own head, none of it written down, none of it visible to anyone else.
The individual interviews gave the rest of the picture. What the group session could not surface, the private conversations did.
Influence and expertise were in different places. The UX and engineering teams held the knowledge about what should be built and why. They did not hold the floor. Direction arrived from senior management as ideas, not as briefs backed by research or data, and it arrived without a route to challenge it.
The people best placed to say "the data does not support this" were the people least able to say it.
That is where the leadership misalignment was landing. Not as a strategic disagreement at the top, but as a team being asked repeatedly to build things they had good reason to believe were wrong, with no mechanism for saying so.
Culture
The team did not huddle. There was no regular cadence of meetings about the work. Meetings happened when management had an idea, or when a build or a fix demanded one.
When the team was in a room together, the loudest voices took it. I watched this in their own meetings. The people who spoke were frequently not the people who knew, and the people who knew said very little.
Both of those are cultural facts rather than process ones. A calendar cannot fix a room where speaking is a contest.
Process
There was no delivery system. Not a broken one, none. Work moved through the function on individual relationships and personal memory rather than through anything documented, repeatable, or visible.
That is a more common finding than it sounds, and it is rarely the first thing anyone reports. A team without a system does not describe itself that way. It describes itself as busy.
With the full picture from the interviews, the actual process could be mapped. Not the process as anyone would have described it in a document, the one the work was really moving through.
The PDF I had originally been asked to write would have described a process that did not exist.
Quick wins are not optional and they are not decoration. A team that has been let down before gives one window of willingness, and it stays open only as long as they can see something changing that benefits them. A quick win that serves leadership optics and not the team is worse than none at all.
The first changes were cultural, and they were for the team.
- Morning scrums. Fifteen minutes, timeboxed, the whole team standing around a screen. Every day.
- A football. Nobody spoke unless they were holding it. That single rule ended the contest for the floor. The people with the expertise got to finish a sentence, which many of them had not been doing in meetings for a long time.
- A physical Kanban wall in the office. Not a tool, a wall. The work became visible to everyone in the room, including the people upstairs who walked past it.
- A weekly planning cadence, so the team knew when decisions would be made rather than waiting to be interrupted by them.
- A product manager, introduced to coordinate these sessions and make sure the right people were talking at the right point.
None of this benefited management. It cost them the ability to interrupt at will. It gave the team a structure where the person who knew the answer could say it out loud.
That is what opened the window.
A bespoke delivery system, built from nothing, in three and a half months. Not adopted from a framework and imposed, built around how this team actually worked.
Four things came out of the engagement:
Simple enough to be used, structured enough to hold. Adopted across the full team of twenty plus.
A way for the function to try something, measure it, and keep or discard it without escalating every decision upward.
This is the change that mattered most to the team, and it is worth being precise about why. Before it existed, work arrived as an idea from senior management and there was no legitimate way to question it. After it existed, proposals were tested and decisions were made against what the data showed. The same senior managers were still in the building. What changed is that "I think" stopped being sufficient grounds to commit engineering time.
The frustration in the UX and engineering teams did not reduce because people became more reasonable. It reduced because there was finally a mechanism through which being right could matter.
A documented map of the capability the function had against the capability the work required.
The gap analysis turned into an evidence-backed argument that could survive an executive forum, rather than an opinion that could be dismissed as one.
The backlog was cut by fifty percent inside the same period.
Staff were trained at every level. Not a handover document, actual training, so the system had people who could operate it after I left.
By the end of the engagement the senior leadership team had enough confidence in the function to commission more than thirty thousand pounds of additional agency resource.
That number is worth reading correctly. It is not a delivery metric. It is a trust metric. Leadership does not increase investment in a function it has stopped believing in. The additional resource was the measurable form of a judgement that had changed.
Once I was confident the system worked, because I had been delivering inside it myself, I began transitioning it to the person taking over my position.
Not a handover document. Coaching. How to run it as it stood, and how to develop it, because it had been built to scale and she needed to know that she was allowed to build on it.
What she inherited was a capability, not an instruction manual. Something simple, repeatable and scalable that could grow with the function rather than constrain it.
The goal of every engagement is a self-managing function. If an organization still needs me when I leave, the engagement has failed.
This engagement is where Change From The Inside™ was first formally defined. The sequence had been used before. This is where it was written down, tested against a function that had no system at all, and proved out.
