They do not.

55.0% of directors and board members told us their content was very effective. Among managers, 37.4%. On whether content is deliberately built so AI and search tools can read and cite it, the numbers are 88.0% against 58.3%. On readiness for AI-mediated buying, 56.0% against 29.6%. On every judgement we asked for, directors and board members are twenty to thirty points more positive.

Then we asked both groups what actually happens week to week. Whether ideas get rebuilt for each channel. Whether named people carry them. Whether anyone can point to a business result. On all of it, the two groups gave the same answers, not merely similar ones.

So the difference is not in what these organisations do. It is in what people believe about it.

Teams agree what is being done. They don’t agree on whether it is working.

The people closest to the work are the least impressed by it, and they are the ones who would know. A content programme gets quietly marked as solved. That is the real cost of this finding: a content programme gets quietly marked as solved several rungs above the people who could tell you it is not, and nobody escalates a problem the boss has already declared fixed.

These are different people at different organisations, not bosses and their own teams, so read it as a pattern rather than a diagnosis of any one company. But it is an easy cheap pattern to test in your own business.

This is the finding that interested us most, and the one that changed what we thought we were testing.

We push clients to put named ambassadors behind their content, because experience shows how well it works. It is a hassle to organise and it takes effort to measure, and both of those are why it often does not happen. But the research says it is worth the trouble.

162 organisations said trusted individuals shape their thinking and carry it to market on an ongoing basis. The other 53 said they activate people occasionally, or rely on the brand account alone. Of the first group, 58.0% could point to both a conversation their content started and a business outcome. Of the second, only 18.9%.

That is three times as many. Getting your own people to carry a message to market makes an organisation three times as likely to have something to show for it, which justifies a great deal of effort and a good deal of spend. It is an association rather than a cause, and organisations already getting results may find it easier to keep a programme of named voices going. It is still the widest gap in the study, by a distance.

The reason is not mysterious. What we counted was conversations, replies and introductions, and business that followed, and those things happen between people. An organisation can announce something, but somebody has to be standing there when the reader decides they want to talk. An introduction cannot be made to a logo.

A brand account can publish. It cannot converse.

Throughout the research we wanted to test the gap between what marketers believe and what is actually happening. Nowhere is it starker than here.

75.3% said trusted individuals shape their thinking and carry it to market on an ongoing basis. Not occasionally and not for a single campaign, but as a matter of routine. But of those people, 37.7% could not name a single person who had carried anything in the past year.

These are not people who said the idea sounded good. They described a continuing programme, and then nearly four in ten of them could not produce one piece of evidence that it had happened.

There is a reason this layer is the hardest to credit. When a named person carries an idea, the proof of it lands in their feed, their inbox and their conversations, where the marketing team cannot see it without effort. So the thing that works best is also the thing that is hardest to evidence, but the data shows it is worth the investment.

The human layer is something most plan for, and few execute consistently.

Almost everyone we spoke to is firmly aware of how AI is changing search.

Some clients think it important enough that they are discussing shadow sites built for no purpose except to be citable by assistants. So it was no surprise that 92.6% said their organisation was fairly or very prepared for a shift in AI-mediated buying, or that 91.6% said being found and cited by AI mattered.

However, despite the importance of AI in search, 27.4% said their content was structured for it inconsistently or not at all. So nearly everyone believes AI matters to how buyers will find them, but more than a quarter are not geared up for it.

We also tested whether structuring content for AI had any bearing on results today. It had none. But it is important to not that every evidence question in this research asked what happened over the past year. As AI assistants and their impact on search is evolving so rapidly, it may be that the impact of structuring content for AI citability is not year demonstrable.

That said, there is a truth any marketer knows. Nothing gets quoted that cannot first be read. When a buyer asks an AI assistant which firms to consider, the answer is assembled out of what the assistant could actually reach. A report behind a form is not readable. A summary that holds the substance back lacks value. And the PDF you are proud of is a layout rather than a document: the headings, the hierarchy and the tables that make an argument quotable mostly do not survive the AI trip.

Nine in ten said they were ready. But more than a quarter had structured their content to make it citable.

Planning distribution at the start, before the content is made, is the textbook answer. It is what we have advised clients to do, and 65.6% of the organisations we surveyed said they do just that.

Interestingly, they did no better than everyone else. 49.6% of the early planners could point to both a conversation and a piece of business, against 45.9% of those who plan distribution later or not at all. On a sample of this size that is not a difference.

What did separate them was whether they actually rebuilt each idea for the channel it appeared on: here 54.9% could point to both a conversation and a piece of business, against 37.8% who did not go to the effort.

But it is The rebuilding is what shows up in the results. A plan made at the start and not acted on is worth nothing at all, and a rebuild that happens late is worth a great deal.

The plan is not the point. The rebuild is.

We expected all of them to matter in roughly equal measure. A system means parts that each contribute and none decides on its own. The research did not agree.

43.3% of organisations did all three things we tested. Two thirds of that group could point to both a conversation and a piece of business. Among everyone else, 34.4%. On the face of it, that is a system working.

There was one area of the data set that we found particularly unexpected. Doing none of the three layers: 18.8% could still show a result. Doing one, 43.9%. Doing two, 32.3%. Actioning two layers appears to yield less results than just one. This could be a statistical anomaly, but looking beyond that nuance, the bigger point is that simple logic suggests that running two out of three should get you two thirds of the way. Instead it barely moves the dial.

So we split the sample into all eight possible combinations. There is a line across the results, and it does not fall where we expected. Every combination that includes a named human being sits above 33%. Every combination without one sits below 28%. How much of the rest an organisation does makes no difference to which side of the line it lands on.

Which produces the most uncomfortable number in the study. Organisations with content deliberately built to be found and cited, rebuilt properly for every channel it appeared on, and nobody’s name on it, could show a result 16.7% of the time. Organisations doing none of the three managed 18.8%. In other words, do the two hard technical things, but skip the human one, and you land where you would have landed by doing nothing.

Authority and distribution are the volume. Named people are the switch.

The organisations doing nothing but named voices, with no structured content and no native rebuilding, could show a result 63.6% of the time.

That group is worth a caveat, because it is easy to misread. They were not organisations with nothing to say. Everyone in this survey had content. What they lacked was content deliberately structured to be found and quoted, which is a different thing and a more recent requirement.