Most of what is written about AI in B2B marketing is about the marketer: tools that write the posts, score the leads, build the ads. The larger change is on the other side of the table. The buyer, the operations director at a 200-person company who needs to replace a system, now asks an assistant first. By the time any channel sees that person, the assistant has answered their first ten questions, built their shortlist and told them what to be wary of, from whatever it has read.
That journey is invisible to the operator's reports, and the invisibility is already changing budgets in the wrong direction.
Two years ago the buyer's research was a sequence of searches, each one landing on a page, each page a session with a source. The channels saw the whole journey because the journey ran through them. Marketing could count it, attribute it and fund it.
Now the buyer opens an assistant and asks: what should a company our size use for this; what does an implementation cost; what goes wrong; who does this well. The assistant answers from the pages it has read, including the vendors' and the consultancies', and the buyer refines the question five times without leaving the conversation. When they arrive at a site, if they arrive, they type the address or search the name. The analytics property records a direct visit, or a branded search, with no campaign, no referrer and no path.
The journey still happened. It happened somewhere the channels cannot see, and it arrived looking like it came from nowhere.
Three things, and all three point the budget the wrong way.
The direct and unassigned share grows. Every analytics property we look at shows it: more sessions with no source, more leads whose first touch is the homepage. The reflex reading is that tracking is broken. The accurate reading is that the first touch happened in an assistant.
Attributed content performance falls. The guide that the assistant cites when it answers the buyer's third question produces no click, because the buyer did not need to click; they got the answer. The content budget, judged on attributed leads, looks like it stopped working at the moment it started doing the first half of the sales job.
And the leads that do arrive are further along and fewer. Sales notices that the first call is shorter and the buyer already knows the price range and the risks. Marketing's report shows fewer leads and concludes the top of the funnel is shrinking, when the top of the funnel has moved out of view.
We described the table a CFO would sign in the companion piece on agency reporting; this is the reason the table's first-touch column is starting to read direct for the best opportunities.
Take a mid-market operations director deciding whether to leave QuickBooks for an ERP. Two years ago that was a dozen searches over a month: outgrowing QuickBooks, ERP for a company our size, ERP implementation cost, NetSuite versus the alternatives, and a dozen pages read on each, every one a session with a source. Now it is one conversation with an assistant that runs over a lunch hour: what are the signs we have outgrown it; what would an ERP cost at a hundred and fifty people; what goes wrong; which implementers do this for companies like ours; what should I ask them.
The assistant answers each from pages it has read. If one of those pages is yours, you are in the shortlist before the buyer has been to your site. If none of them is, you are not, and no amount of paid search will put you back, because the buyer is not searching. They are asking, and then they are typing a name into the address bar. Your analytics will show that visit as direct. Your sales team will take a first call from someone who already knows your price range. Your marketing report will show the content that put you on the shortlist producing no attributed leads this quarter.
Every one of those three readings is accurate about the data and wrong about the market, and the budget decision that follows from them is the one to avoid.
The journey can be seen, just not with clicks.
Ask the assistants. Put the buyer's ten questions to the major assistants, monthly, and record who they name, what they say about you, and which of your pages they cite. That is the share of voice that matters now, and it is a half-day's work a month that almost nobody does.
Count branded and direct demand as the outcome of content, not as a separate channel. A rise in people typing your name is the measurable trace of the invisible journey, and it should be put against the content spend rather than against nothing.
Ask the buyer. One question on the form, how did you first hear about us, with an assistant as an option. The answers are rough and they are the only direct evidence of the first half of the journey. Carry the answer into the CRM with the lead, so it survives to the opportunity, which is the join we keep coming back to.
And read the first sales call. If buyers arrive knowing the price range and the three risks, something told them, and the thing that told them read your content or a competitor's.
The assistant cites what is specific, verifiable and positioned. It does not cite the page that says a company is a trusted partner delivering value, because that page says nothing it can repeat. It cites the page that says what an implementation costs and which half of the cost is not software, the page that names the four things that go wrong and the mechanism of each, the page that takes a position a competitor would not.
That is a change in what marketing produces, from volume to statements, and it is the change we made in our own marketing when we started running it as a system rather than a schedule. The decisions involved, which pages to write, what position to take, what to measure, are the ones we set out in the five decisions AI can make in your marketing. None of those decisions is helped by an agency optimising for clicks on a journey that no longer produces them, which is the argument of an AI digital marketing agency versus a marketing system.
The practical question for the operator is what to do with the content budget when the attributed numbers fall. Cutting it is the natural reading of the report and the wrong reading of the market. The content is doing more and showing less. The right move is to change what is measured, to the assistants' answers, the branded demand and the buyer's own account of the journey, and to change what is produced, to the specific and positioned pages the assistants repeat.
We run our own marketing this way and report it from the CFO's side as a managed subscription for companies that want the same: the measures above, the one table, and the content built to be cited. Discovery is no-risk: we put the ten questions to the assistants for your category and show you what they say today, and you pay only if you go ahead.
Watch the 6-minute guided demo, then book a walkthrough: we read your CRM, ad accounts and content and show what a governed system would do differently. You pay only if you proceed. Or see how we approach it.
More on the same problem:
The Five Decisions AI Can Make in Your Marketing Today, and the One It Should Never Make Alone. (article)
Zoho Marketing Automation Is a Tool, Not a Marketing System (video)
Zoho Marketing Automation vs Zoho Campaigns: Which One a Mid-Market Team Actually Needs. (article)
Watch the 6-minute guided demo, then book a walkthrough: we read your CRM, ad accounts and content and show what a governed system would do differently. You pay only if you proceed. Or see how we approach it.
More on the same problem:
"Something isn't working." "We're outgrowing our tools." "We're afraid to make the wrong move." No-Risk Discovery is a short, practical conversation that gets you clarity before you commit to anything big. We'll tell you if we're a fit. If we're not, we'll tell you that too.
Say who you are and ask. A person replies within one business day.
