Every few months I end up in some version of the same meeting. Revenue is behind plan, and someone proposes the obvious correction: hold marketing flat, or cut it, and put the money into sales headcount. Sales closes business. Marketing makes things.
It is a rational position if you believe buying is something that happens when you ask people to buy. It mostly isn’t.
Because it is correct about the part of the market you can see.
At any given moment only around 5% of buyers in a category are actually in the market. The other 95% are not, and will be, later [3]. A sales-first org is optimized almost entirely for that 5% — the people already raising a hand — which is why it produces fast, legible results and why it feels like the efficient choice.
The problem is the other 95%. They are not ignoring you because your pitch is weak. They are not in the market yet. When they do enter it, they will not run a neutral evaluation of every vendor in the category. They will start from a shortlist of names they already recognize and already have some reason to trust. That shortlist was formed months or years earlier, by everything that happened while nobody was buying anything.
Which means the real competition is not for the deal. It is for the shortlist. And you cannot win a place on it at the moment the deal appears.
This is where the acronym gets useful, and also where most people misuse it.
Google’s quality rater guidelines describe E-E-A-T: experience, expertise, authoritativeness, and trustworthiness [1]. The extra E — experience — was added in late 2022, to ask whether content was produced by someone who had actually used the product, visited the place, or done the thing [2].
Two details matter more than the acronym itself, and both cut against how it usually gets talked about.
The first: Google is explicit that “of these aspects, trust is most important. The others contribute to trust” [1]. Experience, expertise and authority are not four boxes to tick. They are three inputs to one output.
The second, which almost every agency deck gets wrong: “E-E-A-T itself isn’t a specific ranking factor” [1]. There is no E-E-A-T score. You cannot optimize for it directly, and anyone selling you an E-E-A-T package is selling you a proxy.
What it is, is a description. Google wrote down the thing human buyers were already doing — preferring people who have obviously done the work — and built systems to approximate that judgment. As the announcement put it, “these are not fundamentally new ideas” [2].
So when I argue for authority I am not arguing for an SEO tactic. I am arguing that the behavior search engines are trying to imitate is the behavior your buyers already have. Demonstrated experience is what makes a shortlist.
Because of the shape of the curve, not the size of the return.
A sales hire produces an attributable number inside a quarter. You can point at it in a board deck. Authority produces almost nothing visible for several months, and then it stops being linear — each piece of work lands on an audience the previous work already warmed, and the same effort starts returning more.
Those two shapes are not comparable on a quarterly reporting cycle, and the quarterly cycle is what most companies use to decide. So the compounding asset loses to the linear one, repeatedly, on a comparison that was never fair.
Worse, the cut is self-justifying. Reduce authority work and nothing bad happens for two or three quarters, because you are still harvesting the shortlist position you built earlier. The bill arrives later, as a slow rise in acquisition cost that nobody traces back to the decision. By then the team that made the cut has usually moved on.
This is the same reason frequency is the lever everyone underfunds. Memory is built by showing up repeatedly over time, and the payoff arrives on a schedule that no quarterly review is built to see.
The example I know best is one I am responsible for: a managed IT services provider, organic search only. I took it over in April 2025.
Through 2023 the site averaged about 350 organic visits a month, worth roughly $160 a month in equivalent ad spend. In the twelve months before I started, about 540 visits and $336. In the twelve months after: about 765 visits and $988 — traffic up roughly 40%, the value of that traffic up nearly 200%.
The last six full months average about 1,160 visits and just under $2,000 a month. Against the 2023 baseline, roughly three times the traffic and twelve times the value. Paid search across that entire period: zero.
Now the parts a chart on its own would let me hide.
These are Ahrefs estimates, not measured analytics — directionally sound, not accounting. The current month shows a spike I have excluded from every figure above, because partial months lie. And there is an earlier rise, spring to autumn 2024, a year before I arrived, which reached about 690 visits and then gave almost all of it back by January. Anyone looking at the same chart will see that bump, and they should. It is the control condition: a lift that did not hold.
Here is the detail I find most persuasive, and it is the one that would be easiest to leave out. The steep part does not start when I did. The first nine months after April 2025 are respectable and unspectacular — a plateau in the 600s and 700s, dipping to 530 in November. The acceleration starts around January 2026 and has not stopped since.
That lag is not an embarrassment to the argument. It is the argument. Nine months of work that would have failed any quarterly review, followed by a curve that bends and keeps bending. If someone had cut the program in October 2025 — six months in, numbers flat, perfectly defensible decision — they would have killed it one quarter before it started paying, and the chart would have proved them right.
What I cannot give you from this data alone is causation. What I can give you is the mechanism: which pages, which questions they answer, what we stopped publishing, and why the compounding shows up when it does. If I could not explain that, the numbers would be worthless no matter what shape they made. The shape is not the evidence. The shape is what the evidence produced.
That distinction is the entire point. The discipline that makes you trustworthy to Google is the same discipline that makes you trustworthy to a buyer. A marketer who volunteers the flat nine months, and the earlier bump that wasn’t theirs, is demonstrating the exact quality the framework is trying to measure. You cannot claim trustworthiness. It is only ever inferred from behavior — which is why AI visibility is a leadership problem before it is a technical one.
There is an obvious objection to everything above: if authority is built by publishing, and AI has made publishing effectively free, then authority gets commoditized along with everything else.
Google’s own guidance answers this more usefully than most of the commentary has.
Its position on generative AI is not a prohibition. The guidance opens by saying AI “can be particularly useful when researching a topic, and to add structure to original content” [4]. What it draws a line at is volume without value: using these tools “to generate many pages without adding value for users” falls under the spam policy on scaled content abuse [4].
Then comes the part worth reading twice, especially if your team is already shipping AI-assisted work:
“Generative models don’t retrieve facts, but predict a likely sequence of words based on their training data. Because of this, generative AI outputs may contain inaccuracies (also known as hallucinations). It is critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing.” [4]
And the review is not just the body copy. Google explicitly extends it to title elements, meta descriptions, structured data, and image alt text [4] — the exact places a busy team is most likely to let a model’s output ship unread.
Read that next to the E-E-A-T guidance and the strategy writes itself. If models predict plausible text rather than retrieve facts, then anything a model can produce unaided is, by definition, the average of what already exists. It is commodity content. It cannot demonstrate experience, because no experience went into it.
So AI does not weaken the case for authority. It removes the last place to hide. When producing competent, generic content costs nothing, competent and generic stops being worth anything — and the only remaining differentiator is the one thing that cannot be generated: what your team has actually done, seen, tested, and got wrong.
Used well, these tools are genuinely good at research, structure, and first drafts, and Google says so. Used as a replacement for having something to say, they produce exactly the material its spam policy describes. The dividing line is not whether AI touched the work. It is whether a human with relevant expertise stood behind it before it shipped.
Nothing here requires a bigger budget. It requires surviving the quarter.
Separate the two scoreboards. Stop measuring awareness work on conversion metrics in the quarter it runs. Measure the 5% work on pipeline and the 95% work on whether you are becoming known — branded search volume, direct traffic, share of the conversation, who shows up in answers about your category.
Publish from actual experience. The E in E-E-A-T is the cheapest advantage available to most businesses, because it is the one thing neither a competitor nor a model can generate. Your team has done the work. Most companies simply never write it down.
Show your working, including what went wrong. Every caveat you volunteer makes the claims around it more believable. This is counterintuitive to anyone trained on marketing copy and it is the fastest way to build the T.
Hold the line for four quarters. This is the real ask and the one that usually fails. The curve is flat before it is steep, and the decision to cut almost always lands in the flat part. Mine stayed flat for nine months.
Marketing is not overhead and it is not a document factory. It is the only part of the business that compounds — and the only one that builds an asset you own rather than renting attention by the deal.
Sales-first works. That is what makes it dangerous. It works well enough, for long enough, that the cost of skipping the other 95% never appears as a line item. It appears as a slowly rising cost of acquisition, a pipeline that depends entirely on outbound, and a brand that has to re-earn the right to be heard in every single conversation.
The companies that treat authority as infrastructure end up having a different conversation. Not “who are you,” but “we’ve read your stuff — can you help.”
That gap doesn’t show up in a quarter. It shows up in a chart four years long.
Mike Bloomstine helps leaders navigate the shift from activity-led marketing to outcome-based systems. To talk through what this looks like for your business, visit mikebloomstine.com.
I help B2B and B2C teams build marketing engines that compound — strategy first, systems underneath.
