Rand Fishkin is right. “Inimitable product” is the wrong frame.

Rand Fishkin's piece yesterday calling for businesses to build "inimitable products" as the answer to Google's AI-powered enclosure of search is the most circulated SEO take of the weekend. It deserves the attention. The diagnosis is sharp, the anger is earned, and the strategic conclusion — that businesses need to offer something an AI interface cannot reproduce — is correct as far as it goes.

But it doesn't go far enough. And the framing is going to mislead a lot of small businesses who read it.

Because "inimitable product" sounds, to most UK business owners, like a problem of *what you sell*. Ultrasonic chef's knives. Made-to-measure suits. Pottery refined over thousands of years. The examples Rand uses are all product-level differentiation — distinctive physical objects or artisan services that obviously can't be summarised inside an AI Overview.

That's not the actual problem most of my readers have. And it's not the actual lever.

The product is rarely what's imitable

Most UK service businesses I work with don't have an imitation problem at the product level. A decent accountant in Leeds is functionally inimitable — the work is bespoke, relational, contextual, and regulated. A plumber in Bristol is inimitable. A solicitor handling commercial property in Manchester is inimitable. A B2B SaaS tool with three years of customer feedback baked into its workflow is inimitable.

The product was never the problem. What's imitable is the explanation of the product.

The *product* is fine. The product was never the problem.

What's imitable — what AI Overviews and AI Mode are eating alive — is the explanation of the product. The "what is conveyancing and what does it cost" page. The "five things to consider before hiring an accountant" guide. The "how to choose a plumber" listicle. That layer of educational, comparative, informational content was the bridge between a business and the customers who didn't yet know they needed it. And it's the layer Google is now atomising into summary boxes that never click through.

So when Rand says "build inimitable products," the small business owner reading it thinks *but my product is already inimitable, that's not my problem*. And they're right. The problem isn't the product. The problem is that the discovery layer above the product just got nationalised.

The actual lever is the relationship, not the artifact

The thing AI cannot reproduce isn't a chef's knife. AI can already describe ultrasonic chef's knives perfectly well, summarise the reviews, compare prices, and link to the cheapest seller. What it can't do is be the *trusted source* the customer thinks of when the need arises.

That's brand. Specifically, it's the bit of brand that lives in the customer's head before they open a search interface at all.

This is the editorial position I've held in piece after piece on this site, and it gets sharper every time the data confirms it. AI systems cite brands they've heard of. Customers ask AI systems about brands they've heard of. The discovery surface has compressed, which means the upstream signal — *does this person already know who you are* — has gone from "useful" to "load-bearing."

Inimitable product is the wrong frame because it points businesses at the artifact. The actual asset is the recognition. The relationship. The reason a customer types your name into ChatGPT instead of asking ChatGPT for a generic recommendation.

What Rand gets right, and what to take from it

Where the piece is genuinely useful is in killing off the "just make great content" delusion. That advice was always partly survivorship bias dressed up as strategy, and it's now actively harmful. If you spend 2026 writing thoughtful 2,000-word explainers hoping they'll rank, you are donating training data to the companies eating your traffic. Rand's right about that. Mike King's been right about that. The math has been clear for two years.

Where the piece misleads is in what businesses should do instead. "Build an inimitable product" is great advice for a founder considering what business to start. It is useless advice for the 90% of businesses already operating, whose products are inimitable enough but whose *brand* is invisible enough that AI systems route around them.

The actual to-do list is less romantic. Earned media in publications the AI systems index heavily. Podcast appearances. Real customer reviews on platforms that get cited (Reddit, Trustpilot, industry-specific directories). PR that generates third-party mentions of your brand alongside the categories you serve. Structured data that makes your entity machine-readable. Internal architecture clean enough that when an AI does decide to cite you, it can find the right page.

It's the SEO playbook with the emphasis shifted: less keyword targeting, more entity building. Less ranking, more being known. Less content volume, more citation worthiness. The pieces of this I've written about recently — the machine-readability problem, Reddit's structural role in AI citations, the end of the keyword as the organising unit — are all really the same argument from different angles. The unit of optimisation has shifted from the page to the brand.

What this means for the next twelve months

Brand is the only compounding asset left.

If you're a UK business reading Rand's piece and wondering whether you need to invent ultrasonic chef's knives, the answer is no. You probably already have a fine product. What you need is for more people to know about it before they next open an AI interface. That is the entire game now.

Practically, that means budget shifts. Less spent on volume content production, more on PR, partnerships, podcast appearances, and the unglamorous work of being mentioned by other people on the open web. It means treating Reddit, YouTube, LinkedIn, and trade publications as distribution surfaces in their own right, not as places to repurpose blog posts. It means measuring brand search volume and AI citation appearance, not just organic sessions.

It also means accepting that this is slower and harder to attribute than the old SEO playbook. There's no Search Console dashboard for "appearances in ChatGPT responses about your category." The measurement layer is being built by outsiders, and it'll take another year before any of it is clean enough to defend a budget with.

Rand's frustration is the right frustration. Google has, in effect, expropriated the discovery layer it spent 25 years inviting businesses to build on. The response can't be to build a better widget and hope. It has to be to build the only asset Google can't intermediate: a customer base that knows your name before they ever open a search box.

That's the work. It's less satisfying than building a chef's knife. It's also the only thing that scales.