Proof
What I've worked on, and what I won't claim
Most consultancy sites attribute company results to their own work. I can't do that honestly, and I'd rather say so than imply it. This page sets out what the work actually produced, and where the line is.
The evidence at a glance
Hidden Hours: evidence into an enduring position
My contribution: narrative interpretation and continuity across the programme. What it shows: the argument could absorb new practitioner and survey evidence. Limit: no isolated commercial attribution.
Delivered editorial work; system under testCustomer stories: making repeated work teach the next brief
My contribution: writing customer narratives across a 15-story programme. What it shows: enough repeated practice to identify common briefing and review problems. Limit: the proposed AI workflow is not yet independently validated.
Professional editorial practice · outcome not isolatedGEO: using field evidence to improve editorial questions
My contribution: connecting AI-discovery questions with practitioner evidence. Limit: no isolated visibility or revenue attribution.
Self-built working implementationEditorial Intelligence: an AI-enabled publishing knowledge system
My contribution: independent architecture, editorial rules and day-to-day operation. What it shows: structured sources and reviews can support ongoing publishing. Limit: it is not proof of client-team adoption.
Editorial Intelligence came out of research-led programmes developed with Sage. They are public, dated and still running, which makes them checkable — a stronger form of evidence than a testimonial, and a weaker one than a controlled result. Both of those things are true and I've tried to write the page accordingly. The site itself is part of the same argument.
Hidden Hours
What I did. I led the narrative programme. The research itself was driven by others — my part was translating the findings into a position: deciding what the evidence meant, what argument it could support, and how that argument held together across everything published afterwards.
The survey is recent. The argument it supports is not. It came out of narrative work already underway with accountants — Making Tax Digital, pricing, the shape of practice change — built on direct conversations with practitioners rather than on survey data. When the new findings arrived, the work was translating one onto the other.
That sequence is worth being precise about, because it can be told two ways. A position formed first and evidence found afterwards is motivated reasoning. A position formed from practitioner evidence and then tested against a thousand-respondent survey that could have contradicted it is something else. This was the second, and the survey sharpened the argument rather than simply confirming it.
That division of labour is the argument of this site in miniature. Research produces findings. Something else has to decide what they mean, and that is the part organisations most often leave to whoever is writing at the time.
What the programme generated. A survey of 1,000 UK accountants and bookkeepers, alongside practitioner interviews, event conversations and customer stories gathered over the life of the programme.
What it found. Accountants spend 44% of the working week on core accounting and compliance, down from 50% the year before. 81% of firms regularly perform work outside their agreed scope. 70% say their fees don't reflect the range of support they provide, and only 19% charge for most or all of it. 62% are spending more time on out-of-scope work than a year ago; 74% more than three years ago. Underneath it, 53% say their services aren't always clearly defined for clients — undefined scope being the precondition for unbillable work.
What it produced. A narrative position rather than a set of findings: the profession is absorbing structural change in ways that are becoming unsustainable, and the evidence for it sits in the granular detail of how time is actually spent. Later work extended that position rather than replacing it, which is the behaviour the whole methodology is built to produce.
Winning in Small
What I did. I worked on the Winning in Small narrative programme, which began in 2025. Its online articulation has since been developed alongside newer research and practitioner evidence; I am not claiming that the 2026 Hidden Hours survey informed the original 2025 launch.
What it demonstrates. The current Winning in Small argument uses later Hidden Hours findings — including the 44% compliance figure, the 81% working outside scope and the change in beyond-the-brief work — to examine the wider small-business problem. That is an example of later evidence extending an earlier narrative, not evidence of a 2025 survey-led programme.
The useful distinction is chronological: the original narrative came first; later evidence sharpened its current explanation. Evidence can compound, but it cannot travel backwards in time.
Sage Future customer stories
What I did. I wrote the published stories across a 15-testimonial programme captured at Sage Future, working from interviews and source material into customer-ready narratives as part of a wider cross-functional process.
What it demonstrates. Repetition creates a different kind of editorial evidence. Once the same workflow has been used across enough real stories, recurring problems become visible: weak story spines, duplicated themes, unsupported claims, quote handling, review patterns and structural rewrites. That is the practical basis for the customer-story system work now being tested — not an abstract prompt, but a way of preserving lessons from one story and applying them earlier to the next.
I have also been testing a structured customer-story evidence register to retain story spines, proof points, gaps, questions we wish we had asked and editorial lessons for subsequent briefs.
I am not claiming the AI-enabled workflow has proved itself yet. The evidence here is narrower: there is now a substantial body of real editorial work against which briefing, drafting and QA improvements can be tested.
GEO-informed editorial decisions
The problem. Making Tax Digital for Income Tax content needed to answer more than basic readiness questions. AI-discovery prompts were a useful starting point, but not evidence of what practitioners were experiencing.
What I contributed. I connected questions around quarterly updates with evidence from Accountex North sessions and an SMB advisory board. That brought practical issues into focus: who is responsible for submissions, what confirmation looks like, and how accountants and clients divide the work. The point was to make the editorial questions and briefs more specific, not simply add keywords.
What this demonstrates. GEO can improve the decisions made before content is written: what to investigate, what can be supported, and where first-hand evidence is missing. This work took place within a wider Sage team. I cannot attribute AI-answer visibility, traffic or revenue outcomes to my contribution.
Read the related working note →
Lessons Behind the Badge
What I did. I helped shape the editorial proposition, structure, wireframe direction and copy for a customer-story experience built around the businesses behind EFL clubs.
What it demonstrates. The upstream side of editorial systems work. Before there are individual stories to draft, somebody has to decide what the experience is for, what kind of evidence belongs inside it, how the stories should be structured and how the parts connect. That design work and the Sage Future delivery work sit at opposite ends of the same discipline: shaping the system, then learning from the stories that run through it.
The Expanding CFO
What I did. Report writer and narrative translator. The findings were produced elsewhere and handed over as themes; the work was linking them to a story — deciding what the data was actually describing and turning it into an argument somebody could act on.
Why this is the clearest example on the page. The narrative says it outright, in a line that has been sitting on the site since long before there was a methodology to justify it:
The themes were given to me as data. The argument is mine.
What the argument was. Not that the CFO role is expanding — the research said that, and 89% of CFOs reporting a changed role says it more precisely than any narrative could. The position was that most CFOs are absorbing the expansion reactively, and the ones succeeding are shaping it deliberately.
That distinction is not in the data. It is a reading of the data, and it is the kind of thing that either survives contact with the people it describes or doesn't. This one did.
Trusted AI
Included because it is unfinished. It is published at version 0.1 and marked as a position in development, on the basis that a narrative should be visible while it is still gathering evidence rather than only once it looks complete.
Editorial Intelligence publishing and knowledge system
What I built. I maintain Editorial Intelligence as a structured publication connected to a separate body of working knowledge and methodology. The public pages have explicit relationships, editorial statuses and revision histories, while the working system sets rules for source retrieval, AI-assisted drafting, review and publication.
How it works. Working knowledge and source material sit apart from the public website. Editorial statuses, revision decisions and source checks can be revisited rather than reconstructed from a chat. AI assists with retrieval and drafting; decisions about evidence and publication remain editorial.
What it demonstrates. This is an actual independent implementation of an evidence-aware editorial workflow. It can show the structure and decisions behind the publishing process, not just the published output. The implementation guide explains the architecture →
What it does not establish. That the same approach will be adopted successfully by a multi-author team, reduce cost, improve accuracy or create commercial return. Those are the questions a bounded external pilot would need to test.
See the proposed founding pilot →
How the work travelled
Reach and revenue are the wrong measures for this kind of work, and they're also the ones I have least right to. What tells you whether a narrative worked is what happened to it afterwards.
The position outlived the campaign that carried it. The argument was not created by any single piece of research. It came out of accumulated work with accountants, absorbed a major survey, and has since taken in regulatory change, product feedback and event conversations with practitioners who were never part of the research. Each addition strengthened it or refined it. None forced a retreat.
Later evidence strengthened an existing argument. Winning in Small began in 2025; the 2026 Hidden Hours research subsequently gave its online narrative a more specific account of the pressure carried by accountants and small businesses. That is a useful example of reinterpretation over time, not a claim that the later survey was available at the original launch.
It ended up as a framework rather than a campaign. The narrative has a structural home, which means it stays available to teams, product work and customer conversations instead of finishing when the content did.
Article-level performance figures exist and I'm glad to go through them in a conversation. They're not on this page because they measure the least interesting thing — whether individual pieces were read — rather than whether the argument held.
What I won't claim
I won't tell you this work produced a commercial outcome, because I can't separate it from everything else that was happening — product, pricing, partnerships, regulation, and the efforts of a great many other people. Any consultant claiming otherwise about a company of that size is guessing, or hoping you won't check.
It would also contradict the argument this entire site makes. Productivity measurement in this kind of work has broken, and Hidden Hours is itself a study of valuable work that doesn't show up in the visible numbers. I'm not going to make an exception for my own.
What I'll stand behind is narrower and more useful: the findings were turned into a position, that position held up as new evidence arrived, and the work compounded instead of resetting. If that's the capability you need, the record above is the thing to judge.
Evidence I'm generating independently
The programmes above were built inside one organisation, in B2B technology, for one market. That's a real limit on how far the methodology has been tested, and the reason for the practitioner research: conversations with people working on the same problems in contexts that aren't mine, designed to find where Editorial Intelligence doesn't hold rather than to confirm it.
Work with me →