I reconstructed a clinic group's marketing year from the outside before day one: what an outside-in content inventory finds that a briefing doesn't

The landing page for a brand-new clinic opening said "new from June 2025". The clinic had opened in June 2026. Nobody inside the business had looked at that page since it was cloned from the one before it, and the one before that, and somewhere back down the chain a year had been left behind.

I found it in a screenshot, on day minus thirty, before I had a single login. That is the argument for this whole exercise in one detail.

When I take over marketing in a business I don't know yet, I do not start with the briefing. I start by reconstructing what the company actually published over the past year, entirely from the outside, and I do it before anyone tells me what the strategy was. What follows is the method, drawn from the most recent time I ran it: a multi-market clinic group in medical aesthetics, roughly nine months of output, about three hundred and fifty dated items across websites, social, paid media, video and press.

Why the briefing is the wrong starting point

A briefing tells you what people believe happened. It is assembled by the people who were there, filtered through what they were responsible for, and shaped by what they would like to be true going into a new quarter. None of that is dishonest. It is just a different artefact from the record.

The record is what shipped. Pages with published dates, ads with start dates, videos with upload months, posts with timestamps. It has no opinion about itself.

Ben Horowitz's line in What You Do Is Who You Are is about culture, but it transfers exactly: culture "is a consequence of actions rather than beliefs, it almost never ends up exactly as you intend it." Marketing strategy is the same. What a company shipped is its strategy, whatever the deck says.

In this case the briefing I had been given got three things wrong that mattered — two opening dates off by a year, and a set of patient stories described as current that turned out to be two years old. I did not catch those by being clever. I caught them because dated public records disagreed with a document, and a dated record wins.

What you can actually read from outside, in 2026

More than most people assume, and far more than five years ago. The EU's Digital Services Act obliges the large platforms to run public ad repositories, and the practical effect for anyone doing competitive or pre-mandate work is that a year of a company's paid creative is sitting there, dated, searchable and free. The transparency obligations have tightened further since. Google's Ads Transparency Center is the other half of the same picture. I wrote up the full outside-in stack when I built it for reading a competitor's channel mix — the same instrument works on a company you are about to join.

The website is the bigger seam and almost nobody mines it. A WordPress REST endpoint will hand you every page with its published and modified date, per language. Run it across every locale and you have the build order of an entire international rollout. The Wayback Machine fills the gaps and, more usefully, lets you diff the page's own tag stack between two dates — which is how I established that a chat widget had quietly disappeared from every market on a specific day in June, and that no replatform had happened despite a rumour that one was coming.

Three hundred and fifty rows came out of that. About a third were genuinely new pages; the rest were edits.

The findings are mostly absences

This is the part that surprised me the first time I did it, and it is now the thing I expect. The headline findings were not campaigns done badly. They were categories of work that had not happened at all.

No new blog content in three of the markets, all year. No long-form video since spring. Not a single piece of third-party press coverage on any of the new openings — for a business whose openings are its main growth event. No visible brand campaign anywhere, only offers. Ten social handles with duplicates among them, so the audience was being split across accounts nobody had consolidated.

On the other side, exactly one mature playbook existed and it was genuinely good: a repeatable pre-opening sequence — a teaser campaign about two months out, a city landing page, an opening-offers page, a countdown, a video introducing the new doctor, then an "it's open" wave. That sequence had been run four times in one country and it worked. It had never been ported anywhere else.

That single sentence — the one thing that works is not being copied — is worth more in a first week than any audit of what to improve. It is a decision you can take on day one, with no new budget, and no one has to be told their work was wrong.

Why this is a fractional problem specifically

A permanent hire gets to absorb a company over six months. That is a real luxury and it is not one I have. When you come in part-time and the clock is short, the first fortnight cannot be spent forming an impression; it has to produce a shared, dated picture that everyone can argue with.

Michael Watkins built an entire book on the principle that the early weeks are for diagnosis rather than action, and I agree with him — but diagnosis normally means interviews, and interviews give you beliefs. An outside-in inventory gives you the record first and lets the interviews be about the gap between the two. Those are much better conversations. Nobody is defending anything, because the thing on the table is a list of dates.

It also changes what I ask for internally. Because I already know what shipped publicly, my list of questions is short and specific: paid history for the two markets where I can see nothing, the ads change history, what the CRM actually sends. Not "tell me about your marketing".

The restraint matters as much as the inventory. Finding thirty problems in week one and naming all thirty is the fastest way to be ignored by week four — I've argued that at length in the art of restraint, and the inventory makes it easier to hold, because a dated list makes the ranking obvious.

The whole thing took a day and a half and cost nothing but tool time. I am turning it into a monthly routine for the engagement, because the same query that reconstructs a year reconstructs a month, and "what actually shipped" is a better status report than the one anyone writes by hand.

Read the record before you read the room. The record has no reason to flatter anyone.

Sources & further reading

External: European Commission — the Digital Services Act · Lewis Silkin — the EU raises the bar on ad transparency · Google Ads Transparency Center · Internet Archive Wayback Machine · Michael D. Watkins, The First 90 Days (HBR)

Related posts: How to read a competitor's channel mix from the outside · When clinic advertising is really advertising empty appointment slots · The brand book didn't exist · The art of restraint