Twenty-two percent of the traffic was Mexican. Thirteen percent was British.

We were being paid to rebuild that site for the British market.

The number had been sitting in the analytics the entire time. It was not buried, not hard to reach, not the product of some clever piece of analysis we performed. It was one click from a dashboard several people had open most days. It had just never been anybody’s question.

The mandate was more honest than most:

To create an operational prototype for a new digital experience platform for bp’s UK market, alongside workable designs for the supporting governance and operating models.

That second clause is the whole engagement, and it is rarer than it sounds. Most platform briefs ask for a site and treat the question of who will run the site as an implementation detail for somebody else, later. This one named the operating model on day one. Which meant discovery could not stop at the website. It had to reach the machine producing the website.

I led it as Strategy Director. Eight workstreams running in parallel, twenty-four stakeholder interviews across the communications organisation, an ecosystem map, a content audit, a performance benchmark, and a technical assessment of what the existing stack could actually carry.

The organisation was not short of prior work. There were troves of it: audience maps, third-party research, executive summaries, all of it competent, most of it unread. There was also genuine appetite for change. What sat alongside the appetite, in the phrasing that survived into the readout, was the inertia of entrenched practices.

Appetite and inertia in the same building is the normal condition of any large organisation. It is also why discovery has to be evidentiary rather than persuasive. Nobody there needed convincing that something was wrong. What did not exist was a shared, specific account of what.

The map surprised the people who owned it

The UK site did not sit alone. It sat inside a constellation: global corporate properties, consumer brands, fleet portals, loyalty apps, charging networks, careers sites, pension services, retail partnerships. Each with its own audience, its own owner, its own entirely defensible reason to exist.

Plotted on one wall against ownership, it stopped being a list and became an argument. Organisations are routinely surprised by their own map, and the surprise is the deliverable. No memo does that work. A memo can be disagreed with by someone who does not want to look.

Underneath the sprawl was a pattern I have now seen in warehouses and building portfolios and municipal agencies. Brand and marketing teams had defected to outside creative agencies for campaign work, pushed there by restrictive budgets and policies. Adoption of the internal asset manager, the CMS, the collaboration tooling, was low. When a central system will not give a team what it needs, the team routes around it. Every route-around then becomes a permanent piece of the estate, maintained by nobody, owned by nobody, and impossible to remove because something now depends on it.

Thousands of pages accumulate that way. Not through negligence. Through a long series of individually reasonable local decisions.

The second number

Among the UK visitors the site did have, 54% came exactly once. 56% of visits lasted under ten seconds.

Ten-second sessions and a single-visit majority are the measurable downstream signature of content produced without prioritisation and published against no particular definition of success. You cannot lay that out of existence. The technical benchmark agreed from a different direction: core web vitals mostly passing, visual stability poor, largest-contentful-paint failing. The fingerprints of a page assembled by many hands to no shared standard.

The slide that decided the recommendation

We mapped the content workflow as it was supposed to work. A service blueprint, a major campaign, an interactive feature, drawn as the ideal path.

Then we looked at it.

Even in the best case, content moved through many disjointed groups, handing responsibility back and forth iteratively, with no single point of accountability anywhere along the chain. In reality no execution is ever that clean. And when the best-case version of a process is already that fragile, what you are looking at are not exceptions or bad actors or people who need more training. You are looking at the process working exactly as designed.

That reframed the deliverable. The recommendation ran governance first, content model and workflow second, technical roadmap third, sequenced so the workflow changes landed before migration rather than after it.

That order is the most contested call in engagements like this and the one worth defending hardest. Migrate first and you get a clean platform that immediately begins re-accumulating identical debt, because the machine that produced the debt is untouched and still running. Fix governance first and progress is slower, much harder to show an executive audience, and it holds.

The window

The thing that made any of it adoptable had nothing to do with the analysis.

The work coincided with a formal restructuring of bp’s communications departments. The org chart was already in motion. Reporting lines that would normally take a year of change management to shift were, that quarter, genuinely movable, because somebody was going to move them anyway.

So we scoped the workflow recommendations to fit inside a reorganisation that was happening regardless. Not because it was the most elegant version of the recommendation, but because it was the version that could be executed by people who already had the authority to execute it.

Strategy that arrives during a reorganisation gets called a plan. The same strategy six months later gets called a memo.

Trevor MacDermid was Strategy Director on the bp engagement, leading discovery and the resulting recommendations on organisational workflow, content, and technology deployment.