Stop trying to rank for keywords
21st February, 2026
Almost all SEO briefs, ambitions, and ‘strategies’ begin with a sentence like, “We want to rank for X”.
It sounds reasonable. Commercial. Measurable. Sensible, even.
It is also the wrong place to start.
That sentence assumes that search visibility is something you can directly pursue. That rankings are goals you select from a menu. That if you optimise hard enough, you will be rewarded with placement.
That is not how this works. It never really was.
Rankings are an output
You do not choose what you rank for. You are inferred.
Search engines, and now AI systems layered on top of them, build models of entities. They infer what you are about, what you are good at, who you are relevant to, and how much you should be trusted. Pages are inputs. Rankings are outputs.
If your strategy begins with “how do we rank for X”, you are trying to manipulate an output without changing the system’s understanding of the underlying entity.
That is why so much SEO work feels unpredictable, and often like magic. Tweak a title here. Add some copy there. Acquire a handful of links. Sometimes it moves. Often it does not. Rarely does it stick.
I made a similar argument in There’s no such thing as a backlink. We fixate on the visible artefact and ignore the system that interprets it. The same mistake is happening here. A keyword is not a strategy. It is a label applied to a pattern of demand.
The landing page delusion
The standard playbook is familiar:
- Find a high volume keyword.
- Create a landing page precisely matching it.
- Write 1,000 words of broadly applicable advice.
- Optimise headings.
- Internally link.
- Wait.
In a simpler web, this sometimes worked. Marginal gains were enough.
Today, most of those pages are interchangeable. They do not introduce new insight. They do not expose new data. They do not reflect lived experience. They are competent summaries of what already exists.
Search systems are very good at collapsing competent summaries into a single, canonical answer.
AI systems are even better at doing so.
If your strategy is to produce a marginally better version of the same page your competitors have, you are building raw material for someone else’s synthesis. The system will quote the most credible, consolidated source. Not your carefully optimised paragraph.
Most keyword landing pages are not resources. They are bait. And increasingly, bait is ignored.
This is not just an SEO problem
It is tempting to treat this as a tactical issue. Tidy up your content. Improve your technical SEO. Work harder.
But “we want to rank for X” reveals a deeper problem.
It centres the channel, not the audience.
It centres the metric, not the outcome.
It treats demand as something to capture, not something to understand or shape.
A more serious starting point looks different.
What are we uniquely positioned to produce?
Which audiences are we trying to help make progress?
Where do they struggle?
What could we publish, build or expose that would genuinely reduce friction for them?
This is not new thinking. Frameworks like Jobs To Be Done have been reminding us for years that people “hire” products and services to make progress in a specific context. They do not wake up wanting to interact with a keyword.
Keywords describe how people articulate their situation. They do not define what you should build.
Differentiation is the strategy
If ten companies sell similar products and all publish identical “ultimate guides”, no amount of optimisation will manufacture distinctiveness.
Search systems model sameness. Sameness does not get recommended.
So the uncomfortable question becomes, “What can you create that your competitors cannot easily replicate?”.
Not longer content. Not better optimised content. Something structurally different.
- Original datasets drawn from your own operations.
- Tools that solve a recurring problem.
- Detailed breakdowns of real-world projects, with numbers and trade-offs.
- Opinionated frameworks that reflect how you actually think and work.
- Public documentation of how your product behaves in edge cases.
These are expensive to produce. They expose how you operate. They require real expertise.
That’s the point.
If your honest answer to “what do we uniquely know or see?” is “not much”, then your SEO problem is upstream of marketing. It sits in positioning and capability.
You cannot out-optimise commoditisation.
“But we have the best product”
I hear this a lot. It doesn’t matter.
Search engines and AI systems are not omniscient judges of product quality. They are inference engines. They infer “best” from signals; like:
- Reviews and ratings.
- Mentions and citations.
- Depth of discourse.
- Author credibility.
- Consistency of expertise across contexts.
You cannot declare yourself the best on your own domain and expect the system to take your word for it.
Being the best and being understood as the best are different problems.
SEO sits squarely in the second.
If the wider web does not reflect your superiority, if credible experts are not discussing you, if your own site does not demonstrate depth and experience, the system has no reason to believe you.
And in an AI-mediated world, that inference is made at the level of entities, not pages.
Authorship and experience are not cosmetic
Adding an author box to a generic article does not create expertise.
In a world of synthesis, anonymous content is weak. Corporate bylines with no traceable footprint are weak. Content written by “Marketing Team” or “Site Admin” is averaged out.
Systems increasingly evaluate:
- Who is speaking.
- What they have demonstrably done.
- Where else they are cited.
- Whether their expertise is consistent.
If you are publishing advice about complex topics, the people writing it should have real, relevant experience. That experience should be visible. It should connect to other artefacts, talks, interviews, citations.
This is not about gaming E‑E-A‑T guidelines. It is about being a real, legible entity on the web.
In Why semantic HTML still matters, I argued that clarity for machines is a long-term investment. The same applies here. Clear authorship, explicit relationships between people, organisations and content, and consistent narratives help systems build an accurate model of who you are.
Without that, you are just another interchangeable document.
Start somewhere else
Instead of “What keywords should we rank for?”, start with something harder:
- Capabilities
What do we know, build or see that is non-trivial? - Audience progress
Who is trying to achieve what, in what context? - Unique utility
What can we publish or build that materially helps them, in a way others have not? - Evidence
How do we demonstrate experience, not assert it? - Amplification
How do we ensure this work is seen, referenced and embedded in the wider ecosystem?
Only then should you look at search behaviour.
Keywords become a map of language. They show you how different audiences describe their problems. They reveal geographic and cultural nuances. They highlight fragmentation and confusion.
They inform how you label and structure what you create.
They should not dictate what you create.
What about startups and commodities?
Two common objections.
First, early-stage brands.
You do not need decades of authority to publish something valuable. In fact, startups often have an advantage. They are closer to the problem. They are learning in real time. They can document trade-offs, failures and discoveries openly.
Specificity beats generic authority.
Second, commodity e‑commerce.
If you sell printer ink or phone cases, you may not be publishing essays on philosophy. But you can still differentiate.
- Detailed compatibility guides based on real customer queries.
- Transparent sourcing and quality testing.
- Comparison tools.
- Real installation videos from real jobs.
- Community Q&A.
If your product pages are templated descriptions and stock photography, you are indistinguishable from marketplaces and aggregators. In an AI-summarised world, indistinguishable suppliers are abstracted away.
Differentiation is harder in commodities. That does not mean it is optional.
AI has removed the illusion
None of this is new.
Search engines have always modelled entities and reputation. We just got used to exploiting the slack in the system.
AI has removed some of that slack.
When systems synthesise answers, collapse similar documents, and privilege canonical sources, the gap between “optimised” and “obvious” widens.
If your strategy is to produce slightly better versions of what already exists, you are feeding a machine that will compress you out of view.
I wrote in Standing still is falling behind that the web rewards those who compound. The same is true here. Distinctive, experience-backed artefacts compound. Generic landing pages decay.
Stop trying to rank
If your first question is “what do we want to rank for?”, you are optimising the mirror, not the muscle.
A better question is, “What can we build or publish that would make us the obvious answer for a specific audience, in a specific context?”
Search visibility is a byproduct of:
- Being useful.
- Being legible.
- Being referenced.
- Being remembered.
Keywords still matter. They describe demand. They help you understand language. They can expose gaps.
But they are a diagnostic tool, not a strategy.
Stop trying to rank for keywords.
Start trying to become the answer.