This week’s newsletter is sponsored by Stacker and Semrush
Paid subscribers can download this template to evaluate agency proposals
I will be speaking at the Shenzhen SEO conference in two weeks; reply if you will be there.
Refer friends as subscribers and earn rewards: a free month of a paid subscription, a free Product-Led SEO audiobook ($19 value), or a 30-minute intro call ($900 value)
See the Stacker offer at the end of the email for a free PR story just for booking a meeting
I see a lot of SEO proposals and reports; most SEO firms are not being truthful to their clients. It’s 2026, and an agency is definitely not needed for the same reason it was in 2022 before ChatGPT popped onto the scene.
[Sponsored by Semrush]
New research: Comparing ChatGPT & AI Mode
We analyzed 126M+ prompts to get clarity on the key levers and characteristics of ChatGPT and AI Mode, helping you to grow your brand’s AI visibility in each. Discover the differences in the brands being mentioned, the sources being cited, and the competitive sets they create across 22 industries.
The pre-AI-era agency was driven by a content model that drove the entire agency industry: a team of freelance or full-time employed writers churning out 500-word posts targeting long-tail keywords. The bulk of the agency margin was driven by this model because the agency could tack on 50%-100% on what the actual content cost them to produce. To adapt to AI, many of these same agencies have done nothing more than rename their content teams “AEO specialists.” These new experts do the same work as before, but they just swapped the word “keyword” for “prompt” because the client is not sophisticated enough to notice.
The truth is, if an LLM can write that article faster and cheaper than the agency writer, why do SEO agencies still exist? The answer is that it shouldn’t if this is all the agency does. This isn’t a new reality; rather, AI just made it impossible to hide.
Content alone is not SEO
When it came to SEO, content was never the value an agency provided. In-house teams could also hire freelance writers and manage them. What the client really wanted to pay for was growth. The agency sold the monthly packaging, claiming that their agency process led to growth. Agencies delivered content because it was easy to scope and easy to hand over as a deliverable. Twenty articles a month, briefed, written, and delivered, is a lot easier to scope than the ambiguous “growth.” Some agencies sold guest posts for links, which is also just content, but on someone else’s site. This was just as easy to scope and deliver.
(I have already discussed technical SEO as an AEO deliverable in this post.)
For the growth KPI, they created and pointed to rankings dashboards. This appeared to be progress because it was measurable, and agencies kept their jobs when the dashboard was green. When the dashboard turned red, the agency was replaced with another that could deliver green dashboards. In reality, strategy was supposed to be the real value beneath it all, but agencies never had to prove that because the content itself was the deliverable everyone could see and count.
With the introduction of AI, the primary deliverable of content is worthless. Anyone with Claude access can produce better, meatier mediocre articles nearly autonomously than a typical agency team can ship in a quarter. This is why there are so many social media posts about firing agencies and replacing them with Claude. The core agency deliverable of dashboards and content can truly be replicated by a $20 Claude subscription.
To combat this client churn, now is the moment for agencies to pivot their offering and for clients to demand something new. For the agency model to survive, agencies should stop selling content immediately. A business paying an agency in 2026 purely to churn out blog posts is paying for something with less differentiated value than an LLM subscription. There is plenty of real work left that AI can’t touch, and given how many companies are distracted by AI FOMO, the right work is more valuable now than it’s ever been.
Strategy as THE deliverable
Strategy is the primary asset agencies should be selling. Real strategy means understanding why a business, with its ICP, should own visibility in problems its competitors haven’t yet figured out. This requires a deep understanding of the customer, the product, and the jobs-to-be-done. Doing this is hard work the internal team might not be able to execute on its own, which is why an agency should be brought in to do that heavy lifting.
An organization’s internal team might have customer support data, product feedback, and brand research, but they don’t know how to turn it into a compelling strategy they can execute to drive growth. The lazy way out of just using a keyword tool is the wrong approach. The same internal team can also purchase a keyword tool and make color-coded keyword plans.
They need an agency to do what they can’t do on their own and, most importantly, to turn it into meaningful growth. Companies can certainly bring these tasks and skills in-house, but there are many inherent benefits to using an agency, including instant access to talent (assuming the agency has it), flexibility, and scalability. Additionally, agencies working across multiple accounts will have insights that no single team or employee could access if they focus on only one company. Ultimately, the best option for many organizations might be to in-house these functions, but that is not an option for many.
PR as an agency effort
PR is the second focus area where agencies can help, and most SEO agencies have never done this. They bought links and guest posts, but didn’t deliver meaningful PR work, because that is hard. More than ever, PR is important because it will shape the brand perception reflected in LLMs as citations. An answer engine builds its response from sources with a track record of being cited elsewhere, so the same signal that earns a journalist's trust ultimately feeds an LLM
Getting mentioned in the answer an LLM gives someone requires the same thing as getting mentioned in the New York Times: a third party with credibility decided your brand was worth citing. You can’t fake your way into that with better technical SEO or an audit.
This is earned by being the source a journalist trusts, the founder who says something worth quoting, or the company whose data a researcher cites because it’s actually useful. SEO agencies spent years producing fake outcomes. I have seen past clients pay their way into the NY Times, but it was in places that never mattered for SEO and certainly not for AEO. Fake efforts no longer work when the goal is revenue. A guest post on a page no one, not a crawler or human, will ever see does not move any needles.
The agency of the future
To construct this sort of agency, it will look completely different. There are no writers on staff, or if there are, they’re doing original reporting and customer research, not grinding out keyword-stuffed posts that no one will ever read. This agency's output is not measured by articles written or links acquired; the actual deliverables are strategies, execution plans, and performance feedback. This agency conducts research and provides answers on how to win in ways the client could not achieve on their own.
The value of this agency is measured by its relationships with journalists and analysts, the way a PR firm would, except now that relationship directly determines whether an AI model treats the brand as a trustworthy source on a given topic. The agency coaches the client’s actual subject matter experts on what to say publicly, because those experts talking in their own voice on a podcast is worth more to visibility now than a hundred blog posts ever were.
The agency of the future is much smaller than the one it replaced, and it bills more like a strategic consultancy than one measured by words written or minutes in a meeting. Most agencies will run from this model because it's harder to scale and doesn't fit the model of driving margin with cheap labor.
These agencies will have to be truly creative and more agile than their clients to justify not being fired when clients in-source their basic deliverables or simply use a Claude subscription. The agencies that remain profitable in two years will be the ones that have figured out how to layer PR outcomes onto SEO logic and provide strategies that work in real life, not just in a slide deck.
Yes, agencies should exist, but the best ones will look nothing like today’s agencies. As always, reach out if you want recommendations.
[Sponsored by Stacker]
Earned Media Is How Brands Get Cited in AI Search
See why 84% of AI citations come from earned media coverage.
Learn how brands earn placement across 3,000+ vetted news publishers.
For twenty years, SEO budgets centered on content visibility and backlinks. AI search reshaped both within months.
Large language models gravitate toward brands that are already established and widely cited. That recognition comes from earned media. Muck Rack’s analysis of 25 million+ AI citations found earned media drove 84% of results, with paid content responsible for a mere 0.3%.
Stacker turns brand-funded stories into editorially reviewed content, placed natively across 3,000+ vetted news publishers, with publisher-verified reporting your team can audit. One well-built story on a trusted, high-authority publisher builds more model-visible brand authority than most backlink campaigns combined.