AI answers the question before anyone clicks. What is content marketing for now?

Why is publishing more the wrong response?

Because volume was only ever a proxy for coverage, and coverage of general questions is now free. Producing forty explainers a month competes directly with a system that produces a tailored one in four seconds, for nothing, inside the interface where the question was asked.

Most B2B content programs were built on an assumption that has quietly stopped holding. The assumption was that if you answered enough questions well, some of that traffic would become pipeline. It worked because answering questions well used to be expensive and slow.

It is neither now. Which means the part of your content library that only restates general knowledge has lost its economic basis, however well written it is. That is not a prediction. Look at what happens on a query like what is answer engine optimization and count how many clicks the third-best explainer is realistically getting.

The uncomfortable implication is that a lot of content calendars are producing material that was already commoditized before it was commissioned. Writing it faster with a machine does not rescue it. It just lowers the cost of producing something with no remaining advantage.

What kind of content still cannot be generated?

Anything grounded in something only you have. Your own numbers. The specifics of work you actually did, with the detail left in. An opinion you are willing to defend when a competitor disagrees, and the messy particulars of how a thing goes wrong in practice rather than in theory.

The right column is harder to produce because it requires having done something. That difficulty is the entire point. Anything a competitor can commission from a freelance writer in an afternoon is worth roughly what it cost them.

It also happens to be the material that gets quoted. Systems that summarize the web need something concrete to cite, and a specific claim with a source attached is far more quotable than a general one.

Does traffic still matter if you get cited instead?

It matters less than it did and it has not stopped mattering. A citation places your name in front of a buyer at the moment they are forming a shortlist, which is worth more than a session that bounces, and worth less than a session that converts.

The honest position is that nobody has clean measurement for this yet. Assistant referrals show up inconsistently in analytics, some arrive with no referrer at all, and the number of people who saw your name in an answer and never clicked is not observable by anyone.

So the sensible things to track are indirect. Branded search volume. Direct traffic from people who typed the name. What buyers say when you ask how they found you. These are old fashioned measures and right now they are more reliable than the dashboard.

Treat any vendor's confident number about AI referral share as directional. The methods vary widely and the incentive to inflate is obvious. The trend on your own branded search is a better instrument than someone else's benchmark.

How should the writing itself change?

Answer first, then argue. Put the direct answer inside the first third of the page, make every passage able to stand alone when lifted out of context, and stop writing the slow build toward a conclusion.

None of this requires special markup or a hidden file for machines. It is the structure a busy reader wanted anyway, which is why it works on both audiences at once.

The one habit worth breaking deliberately is the long windup. Introductions that restate the question for three paragraphs were tolerable when the reader had already committed. They are now the reason a page loses to a system that opens with the answer.

• Lead each section with the answer in one or two sentences, then support it
• Write self-contained passages, since a quoted paragraph loses everything around it
• Use headings that match how a person would ask, not clever internal titles
• Attach numbers to a source and a date, because unsourced figures do not get quoted
• Say what you are not claiming, which is rarer and more credible than it sounds
• Keep the specifics that a summary would strip out, since those are the reason to click

So how much should a B2B company actually publish?

Less than most calendars assume, and with far more behind each piece. A useful floor is one genuinely original piece a month that nobody else could have written, plus maintenance on what already earns attention.

The trap is treating cadence as the goal. Eight posts a month of restated general knowledge is a cost center with a publishing schedule. One piece a month built on your own evidence, distributed properly and updated as things change, does more for a brand than the other thirty.

Maintenance is the underrated half. Content decays. Figures go stale, a platform changes a rule, a recommendation stops being true. Updating a page that already has some standing is usually a better use of an afternoon than starting a new one, and almost nobody schedules it.

If a team is going to publish heavily, the volume should sit in distribution rather than production. One substantial piece can carry a month of posts, an email, a talk, and three sales conversations. Most companies do the reverse and wonder why nothing lands.

Where does AI belong in the process?

In the parts that are not the value. Assembling research, drafting an outline, reformatting, spinning out distribution variants, a second editing pass: all fair game. What it cannot supply is the claim, the evidence under it, or the judgment holding those two together. Those are the only reasons the piece exists.

Google's stated position is about the quality and usefulness of the content rather than how it was produced, which is often quoted as permission. Read the second half of the sentence. Content produced primarily to game rankings is the thing being penalized, and mass-generated general explainers are exactly that, whatever tool made them.

There is a practical failure mode beyond ranking. Generated prose about a subject you have not done tends to be plausible and slightly wrong in ways a real practitioner notices immediately. Publishing it in front of the buyers you most want is an expensive way to demonstrate that you do not do the work.

The version that holds up is a person with something to say, using the tools to say it faster. That is a genuinely large productivity gain and it is not the same thing as generating a library.

How do you know if any of it is working?

By whether it shows up in sales conversations. Pipeline influenced, questions buyers stop asking because a page already answered them, and inbound that mentions something specific you published.

Sessions and rankings are inputs. They were always inputs, and treating them as the outcome was survivable while search sent generous traffic. It is not survivable now, because a program can lose a third of its sessions and be worth more than it was.

Ask the sales team a single question each month: which piece came up. If the answer is none for three months running, the program is producing coverage rather than credibility, and the fix is fewer pieces with more substance rather than a faster calendar.

That is the whole reframe. Content used to be a way to be found. It is now mostly a way to be chosen once you are found, and being chosen has never rewarded volume.