Questioning requirements, and facilitating the conversation.
In-house at a group that gives away around £25m a year. I ran the first user journey workshop the team had ever had, turned the room into the reference plan the work now runs from, and built the content system that keeps it fed. This page is also about the last step of that system, which I am not allowed to build.
A workshop is the richest context you will ever hand a model.
Every workshop I have been in produces four photos of a whiteboard and a shared drive nobody opens again. What changed my mind about that is realising what those photos actually are: a full record of a team's collective process, insights, ideas and feelings, captured in one afternoon, that no amount of prompting could invent.
Photograph the boards and the handwritten sheets. Transcribe the lot, verbatim, including the ideas that are obviously never happening.
Feed it alongside brand guidelines and existing research, so synthesis argues against real context rather than inventing its own.
Then the unglamorous bit. Every claim traced back to something someone actually said or wrote in the room.
The output was a single reference document, structured across the journey stages, that the team keeps returning to. The part I would point at is not the recommendations. It is that every one of the forty-odd ideas from the room is accounted for by name. Twenty-two are traced into the plan, with the section they landed in. Eighteen are listed as not yet incorporated, each with a reason. Nobody's idea quietly disappeared, and that is the difference between a summary and a synthesis. The model will hand you the first one dressed as the second.
The last step is the one I am not allowed to build.
The end-to-end version is not speculative. Once a webinar finishes it trims the ends, transcribes with a list of names, organisations and keywords, chapters it, uploads it, drafts the articles, cuts the clips per speaker with copy attached, then listens for how each piece performs and curates that back into itself. Every component of that exists today.
It does not run, because the loop cannot see the analytics. There are legitimate data protection questions and a wider organisational caution about AI tooling, and until those are resolved the system improves from my feedback alone rather than from the evidence of what actually worked. So it is a good loop with the sensor disconnected.
Wrote the proposal, and led with the boundary rather than the upside. A one-pager putting Claude across the web redesign, with the governance stated on the page: every output reviewed before use, no customer records, no financial information, no personal details. Integration across the platform team and agencies was named as a later step, not a day-one ask.
A cautious organisation does not need to be sold the capability. It needs to be shown where the edges are, by someone who has already thought about them.
Twelve people. Eight ideas each. Eight minutes.
The document above came out of a user journey workshop on the charity support experience, run for a team that had never had one. It was the fastest way to see the objectives, the gaps, and the places the service actually breaks.
Awareness, consideration, process, relationship, advocacy. Colour-coded so touchpoints, what the user does, what the team does behind the scenes and where it breaks could never get mixed up on the same surface.
I end a workshop this way because the last few boxes are where people stop giving you their safe answers. Ninety-six boxes filled, and more than forty distinct, comprehensible ideas once the duplicates came out.
Map the ideas onto the axes and the group reaches its own consensus about the most important problems and the easiest wins. Nobody has to win an argument.
The skill comes directly from my RCA experience, and applying it at Benefact Trust is where the understanding of charities and fundraisers came from in the first place.
The stated aim was the journey. The real aim was alignment. Once you start clustering issues from across the functions you form patterns, and patterns stop blockers: the same obstacle turns up in three lanes from three teams who had each assumed it was only their problem. That session kicked off ongoing work on web redesign, on compliance process, and on brand development, none of which were on the agenda when we walked in.
It also surfaced tension, and a lot of what came up in that room is still unresolved. That is the honest half of facilitation. I came in with radical ideas and a fast timeline. The lesson was that organisations move in small steps: unpick assumptions with data, challenge the status quo where you can prove it, and pick your battles.
AI generated content is all about context.
I own charity support communications, which in practice means a standing content problem and no standing content team. Podcasts and webinars turned out to be the answer, because a transcript is the long-form source everything else can be cut from.
Tent poling is my own name for it: stretch one pole and the whole tent stands up. It is not a clever prompt. It is a decision about where the reusable material already sits, plus SEO research so the headings point at demand that already exists. The group content strategy came out of the same approach. I defined it with eight colleagues across internal and external comms, PR and HR, in a session run as a room and a call at once, with senior people in both.
A licensed stock hedgehog, and the burrow built around it.
Movement for Good runs a rolling programme of special draws, each needing a themed asset set across web, social, screensavers and business-unit variants. Four steps, and they are the whole method.
The hedgehog is an Adobe Stock illustration, licensed. What generation did was build the burrow around it: the straw, the branches, the moss, the leaf caught on a twig. That is the useful distinction. The alternative was hunting for a stock scene that happens to contain the right animal in the right mood and settling when it does not, or commissioning an illustration there is no budget for. This is a third option, and it keeps the whole chain inside one licence, which for a financial services group is not a footnote. It is the reason the method is allowed at all.
And only then, video
Four seconds, portrait, silent. It is a Story: fun, loud, on screen for three seconds and gone, which is exactly why the artwork can be as dense as it is. Something with this much detail does not reward being looked at for a minute, so it is built for the one format where nobody will. Motion comes last on purpose, because animating an art direction nobody has signed off is the most expensive way to discover it was the wrong one.
Simplifying the asset requirements themselves. A lot of what arrives as a request for eleven sizes is really a request for three, and the fastest way to raise the quality of a campaign is to make fewer things properly. Questioning the requirement comes before improving the output.
What a team says about you six months later.
From a team exercise on sticky notes. Two of them describe a change in how someone works rather than a compliment, which is the only kind worth keeping.
Expanding what a group thinks it is capable of is the actual output of facilitation. It is slower and less legible than a deliverable, and it outlasts one.
That is most of what this year has taught me, and it applies just as much to the systems as to the people. The tools are ready before the organisations are, and getting an organisation ready is design work too. The constraint is not capability. It has not been capability for a while.
Related · service design at Benefact Trust →