In December 2022, Google updated its Search Quality Rater Guidelines to add a second E to E-A-T. Experience joined Expertise, Authoritativeness, and Trustworthiness, and the accompanying guidance was specific about what it meant: whether the content creator has firsthand, life experience with the topic. Not credentials. Not citations. Whether the person writing has done the thing.
That change formalized something that was already true of readers and has since become true of answer engines. The premium is on the person who has been in the room, and the discount on everyone summarizing what the people in the room said.
Which puts most companies in an awkward position. The expertise exists inside the building, held by three or four people who are billing, shipping, or selling, and none of whom will ever write a blog post. The content team, meanwhile, has neither the knowledge nor the access, so they produce competent summaries of public information and wonder why nothing lands.
The gap is not a writing problem. It is an extraction problem, and it has a method.
The bottleneck is extraction, not writing
Every failed attempt to turn expertise into content that I have seen made the same structural error. Someone asked the expert to write.
This fails for reasons that have nothing to do with willingness. Writing is a separate skill from knowing, and asking a principal engineer or a surgeon or a partner at a firm to produce a publishable draft is asking them to do an unfamiliar job badly, on top of a job they do well. Most decline. The ones who agree produce something six weeks late that reads like documentation, and the experience sours everyone on the idea permanently.
The correct model is that the expert supplies raw material and someone else builds the artifact. This is how business publishing has worked for a century. The expert talks. A writer shapes. The expert reviews and approves. The byline goes on the expert because the ideas are theirs, and the writer gets paid rather than credited.
What changes when you adopt this model is the ask. You are no longer requesting a day of writing from someone whose day costs the company a great deal. You are requesting forty-five minutes of talking about something they find interesting. That request gets accepted, and it gets accepted repeatedly, which is what turns a one-off into a program.
The skill you need to develop, then, is not writing. It is extraction: the ability to sit with someone who knows something and get the knowledge out in a form that can be built on. That skill is teachable and most content teams have never been taught it.
Run the extraction interview
Here is the protocol. Call it the Extraction Interview. It runs forty-five minutes, it is recorded, and it moves through five stages in order.
Stage one is the concrete case. Do not open with the topic. Open by asking the expert to walk you through a specific instance: the last time they dealt with this, with the actual client, the actual system, the actual decision. Names removed, details intact. Abstract questions produce abstract answers, and abstraction is where content goes to die. A specific case produces specific language, and specific language is the whole point.
Stage two is the decision points. Inside the case, find the forks. Where did they choose one path over another? What were the options? What made them pick? This is where the expertise lives, because a decision under uncertainty is exactly the thing a novice cannot make and an expert makes without noticing. Push here. “How did you know?” is the most productive question in the entire interview, and you should ask it four or five times.
Stage three is the wrong answer. Ask what most people do instead, and why it fails. Experts are far more articulate about error than about correctness, because errors are memorable and correctness is habitual. This stage produces the argument spine of the eventual piece, and it produces it almost verbatim.
Stage four is the counterexample. Ask when their approach does not work. Every real expert has boundary conditions and every fake expert claims universality. The boundary conditions are what make a published piece credible, because a reader who has hit that boundary will trust everything else you said.
Stage five is the number. Ask for anything quantified: how long it takes, how much it costs, how often the failure occurs, what the range looks like. Experts carry these figures and rarely volunteer them, because to them they are unremarkable. To a reader they are the most valuable content in the piece.
Forty-five minutes across those five stages produces a transcript with more usable material than a week of desk research. The writer’s job afterward is selection and shape, not invention.
The curse of knowledge is the real enemy
Economists Colin Camerer, George Loewenstein, and Martin Weber published a paper in 1989 in the Journal of Political Economy titled “The Curse of Knowledge in Economic Settings.” The finding was that people who possess information cannot accurately model the reasoning of people who lack it. Chip and Dan Heath later carried the term into popular business writing with Made to Stick, and it has been the best available explanation for bad expert content ever since.
The curse shows up in a predictable way. The expert skips the step that took them three years to internalize, because to them it is no longer a step. They use a term of art without defining it, because in their world the term is ordinary. They answer a different question than the one asked, because the asked question contains a false premise they corrected silently in their head.
This is why the interviewer matters more than most teams assume, and why a writer without domain background is often the better choice. A knowledgeable interviewer nods at the exact moments a reader would be lost, so the explanation never happens. An outsider is forced to say “wait, what does that mean,” and the resulting explanation is the thing readers came for.
Two habits break the curse. The first is asking the expert to explain it as they would to a smart new hire on their first week, which is a specific enough audience to calibrate against without triggering the condescension that “explain it simply” produces. The second is reading the draft back and marking every place where a reader would need to already know something. Those marks are your revision list.
One warning. Breaking the curse of knowledge does not mean removing the difficulty. It means removing the unexplained difficulty. A piece that keeps the hard idea and explains the terms is far more valuable than one that removes the hard idea to stay accessible. When people say expert content got dumbed down, this is the failure they are describing.
The distinction matters commercially as well as editorially. Simplified content competes with every other simplified explanation of the same subject, which is an infinite and worthless field. Content that carries the hard idea intact, with the vocabulary explained as it appears, competes with almost nothing, because very few people can produce it and even fewer are willing to. The reader who needs the simple version was never going to buy from you anyway. The reader who needs the hard version is your entire market.
Mine the artifacts you already produce
Interviews are the primary source. They are not the only one, and the secondary sources cost nothing because they already exist.
Sales call recordings are the richest. Your best rep answers the same fifteen objections every week, in language that has been tested against real skepticism hundreds of times. That language is superior to anything a marketer would draft, and it maps directly to what buyers search for and ask AI assistants.
Support tickets and internal Slack threads are the second seam. The question that gets asked every month by a different customer is a piece of content with proven demand. The internal thread where two engineers argue about the right approach is a byline waiting to be transcribed.
Client deliverables are the third. Strategy documents, audits, technical assessments, and post-mortems contain original analysis that was written once and read by four people. Genericized, with client detail removed, these become the most differentiated pieces on your site because no competitor can produce their equivalent.
Internal training material is the fourth and the most underused. Whatever you teach new hires is, by construction, the compressed version of your firm’s expertise. It was written to be understood by someone who does not yet know, which means the curse of knowledge has already been fought and won.
The work here is a standing habit rather than a project: someone on the content team reviews these sources every week, flags what has publishing potential, and adds it to a queue. That queue will outrun your publishing capacity within a month, which is the correct problem to have.
Two guardrails on mining. First, confidentiality is not negotiable and the fix is structural rather than editorial: strip identifying detail at the point of capture, before the material enters the queue, so no one is relying on a final read to catch it. Second, get consent from the person whose thinking you are publishing. An engineer who finds their Slack argument quoted on the company blog without warning will stop arguing in Slack, and you will have traded one article for the source that produces them.
Keep the nuance, cut the hedging
There is a difference between nuance and hedging, and experts confuse them constantly.
Nuance is conditional precision. “This works for teams over about fifty people, and below that the coordination cost is higher than the benefit.” That is a real qualification that makes the advice usable.
Hedging is confidence laundering. “Results may vary depending on your specific situation and objectives.” That sentence says nothing, protects nobody, and signals that the author is unwilling to be held to anything.
When you turn expertise into content, protect the first and delete the second without mercy. Every conditional in the draft should be interrogated: does this qualifier tell the reader when the advice applies, or does it exist to make the author feel safe? The first stays. The second goes.
Expect resistance on this in the review pass. Experts hedge because they are calibrated to the exceptions they have personally seen, and because being publicly wrong in their field carries real professional cost. The negotiation that works is offering to state the exception explicitly rather than diluting the main claim. “Say the rule, then name the case where it breaks” gives the expert their accuracy back without stripping the piece of a position.
The result reads as authority rather than caution. And it happens to be what answer engines extract cleanly, because a direct claim with stated conditions is parseable in a way that a fog of qualifiers is not. If you are unsure whether your material is reaching them, running a free AI Citation Checker against your category questions will show you whether your expertise is being quoted or ignored.
Build a cadence the expert will tolerate
The programs that turn expertise into content year after year share one design feature: they cost the expert less over time rather than more.
A workable rhythm is one 45-minute interview per month per expert, producing one substantial piece plus derivative assets. Batch the interviews so a single afternoon covers a quarter. Send questions in advance, not because experts prepare (most will not) but because it lets them decline the topics they have nothing to say about, which improves the ones they keep.
Review is where programs die. If you send a 2,000-word draft to a busy expert with no instruction, it will sit for three weeks and come back with a comment about a comma. Send it with a specific ask: check every factual claim, mark anything you would not defend in front of a client, ignore the prose. That framing converts an open-ended editing task into a fifteen-minute verification task, and it gets done.
Then close the loop visibly. Send the expert the outcome: the piece was cited by a publication, a prospect referenced it on a call, it started ranking for the question they answered. Experts participate in content programs for status and impact, not for marketing metrics. Show them the impact and the next interview gets scheduled without a chase.
The firms that build this into how they operate, rather than treating it as a campaign, end up with an archive no competitor can replicate, and that gap widens every quarter it runs.