Over the previous few months, I’ve deeply analyzed how Google’s AI Overviews deal with long-tail queries, dug into what makes manufacturers seen in massive language fashions (LLMs), and labored with manufacturers making an attempt to future-proof their Search engine optimisation methods.
As we speak’s Memo is the primary in a two-part collection the place I’m masking a tactical deep dive into probably the most neglected mindset shifts in Search engine optimisation: optimizing for subjects (not simply key phrases).
On this difficulty, I’m breaking down:
- Why keyword-first Search engine optimisation creates surface-level content material and cannibalization.
- What the precise variations are. (Isn’t this only a pillar-cluster strategy? Nope.)
- Ideas from different professionals throughout the net.
- How you can discuss by these points along with your stakeholders, i.e., shoppers, the C-suite, and your groups (for premium subscribers).
And subsequent week, I’ll cowl construct a subject map, and operationalize a topic-first strategy to Search engine optimisation throughout your crew.
In the event you’ve ever struggled to persuade stakeholders to assume past search quantity or puzzled develop authority, this memo’s for you.
Enhance your abilities with Progress Memo’s weekly skilled insights. Subscribe without spending a dime!
Sooner or later during the last 12 months, it’s doubtless you’ve heard the steering that key phrases are out and subjects are in.
The Search engine optimisation pendulum has swung. In the event you haven’t already been optimizing for subjects as a substitute of key phrases (and you actually ought to have), now’s the time to lastly begin.
However what does that truly imply? How will we do it?
And the way are we supposed to observe topical efficiency?
With all this speak about LLM visibility, AI Overviews, AI Mode, question fan-out, entities, and semantics, after we optimize for subjects, are we optimizing for people, algorithms, or language fashions?
Personally, I feel we’re making this harder than it must be. I’ll stroll you thru optimize for (and measure/monitor) subjects vs. key phrases.
Why Optimizing For Subjects > Key phrases In 2025 (And Past)
In case your crew remains to be centered on key phrases over subjects, it’s time to elucidate the significance of this idea to the group.
Let’s begin right here: The standard keyword-first strategy labored when Google primarily ranked pages based mostly on string matching. However in at the moment’s search panorama, key phrases are now not the atomic unit of Search engine optimisation.
However subjects are.
The truth is, we’re dwelling by what you (and Kevin) may name the dying of the key phrase.
Consider it like this:
- Subjects are the muse and framing of your website’s natural authority and visibility, just like the blueprint and construction of a home.
- Particular person key phrases are the bricks and nails that assist construct it, however optimizing for particular person queries on their very own with out optimizing for the subjects to anchor them, properly, they don’t pull a lot weight.
In the event you focus solely on key phrases, it’s like obsessing over selecting the correct brick shade with out realizing the blueprint is incomplete.
However while you plan round (and optimize for) subjects, you’re designing a construction that’s constructed to final – one which search engines like google and LLMs can perceive as authoritative and complete.
Google now not sees an excellent search end result as a direct match between a consumer’s question and a key phrase in your web page. That’s some previous Search engine optimisation considering that all of us have to let go of utterly.
As an alternative, search engines like google interpret intent and context, after which use language fashions to increase that single question into dozens of variations, a.ok.a. question fan-out.
That’s why a piecemeal strategy to concentrating on Search engine optimisation key phrases based mostly on search quantity, stage of the search journey, and even bottom-of-funnel (BOF) or pain-point intent could be wasted time.
And don’t get me fallacious: Concentrating on queries which might be BOF and remedy core painpoints of your viewers is a sensible strategy – and you need to be doing it.
However personal the subjects, and you may see your model’s natural visibility outlast massive algorithm adjustments.
Key phrase-Solely Pondering Limits Progress
And in any case that, if it’s nonetheless a problem convincing your stakeholders, shoppers, or crew to pivot to topic-forward considering, clarify the way it limits development.
Groups caught in keyword-first mode usually run into three issues:
- Floor-level content material: Articles turn into skinny, narrowly scoped, and straightforward to outcompete.
- Cannibalization: Content material overlap occurs usually; articles compete with one another (and lose).
- Blind spots: You miss associated subtopics, tailoring content material to personas, or exploring issues inside the matter that your viewers really cares about.
Alternatively, a topic-first strategy means that you can construct deeper, extra helpful content material ecosystems.
Your purpose is to not simply reply one question properly; it’s to turn into a go-to useful resource for your entire topic space.
Understanding The Subject Maturity Path: Outdated Method Vs. New Method
Let’s take a more in-depth take a look at how these two approaches are completely different from each other.
Picture Credit score: Kevin Indig
Outdated Method: Key phrase-First Search engine optimisation
The traditional strategy to Search engine optimisation centered round choosing particular person key phrases, assigning each a web page, and publishing content material that aimed to rank for that phrase.
This mannequin labored properly when Google’s rating indicators had been extra literal (assume string matching, backlink anchor textual content, and on-page optimization carrying many of the weight).
However in 2025 and past, this strategy is exhibiting its age.
Key phrase-first Search engine optimisation usually seems like this:
- Minimal inner cohesion throughout pages; articles aren’t working collectively to construct matter depth or reinforce semantic indicators.
- Content material selections are sometimes pushed by common month-to-month search and tool-based key phrase problem scores, slightly than intent or persona-specific wants.
- A high-effort, low-durability content-first Search engine optimisation technique; posts might rank initially and maintain for some time, however they not often stick or scale.
- Monitoring efficiency is usually centered on site visitors projections and completed by web page kind, Search engine optimisation-tool-informed intent kind, question rankings, and (sure) even generally matter teams.
However even when groups undertake a matter cluster-first mannequin (like grouping associated key phrases into matter clusters or deploying a topic-focused pillar + cluster technique), they usually keep tethered to outdated key phrase logic.
The end result? Floor-level protection for single key phrases, frequent content material cannibalization, and a website construction that may appear organized however nonetheless lacks strategic matter optimization.
With out persona insights, or a transparent content material hierarchy constructed round core subjects, you’re constructing with bricks, however no actual authority blueprint.
Wait a second. Is optimizing for subjects any completely different from the traditional pillar + matter cluster strategy?
Sure and no.
A pillar + cluster mannequin (a.ok.a. hub and spoke) is a framework that may arrange a topical strategy.
However strategists ought to shift from matching pages to precise key phrases → masking ideas deeply as a substitute.
This traditional framework can help matter optimization, however provided that it’s carried out with a topic-first mindset.
Listed below are the first variations:
- Key phrase-driven pillar + cluster mannequin: Pillar = covers seed key phrase goal(s); clusters = cowl long-tail variations of seed key phrases.
- Subject-driven pillar + cluster mannequin:Pillar = provides a complete information to the subject; clusters = present in-depth help for key ideas, completely different personas, associated issues, and distinctive angles.
Merely deciding on high-volume key phrases to optimize for in your pillar + cluster technique plan doesn’t work prefer it used to.
So, a pillar + cluster plan can assist you arrange your strategy, however you’ll have to cowl your core subjects with depth and from a wide range of views, and for every persona in your target market.
New Method: Subject-First Search engine optimisation
Your future-proof Search engine optimisation technique doesn’t begin with a give attention to key phrases; it begins with focusing in your goal folks, their issues, and the subjects they care about.
Subject-first Search engine optimisation approaches content material by the lens of the real-world options your model offers by your services and products.
You construct authority by exploring a subject (one you could instantly converse to with authority) from all related angles: completely different personas, intent varieties, ache factors, trade sectors, and contexts of use.
However bear in mind: Subject-first Search engine optimisation isn’t precisely a web page quantity sport, though the breadth and depth of your matter protection are essential.
Subject-first Search engine optimisation entails:
- Masking your core, focused subjects throughout personas.
- Investing in “zero-volume” content material based mostly on precise questions and desires your target market has.
- Producing content material inside your matter that gives completely different views and scorching takes.
- Constructing authority with data acquire: i.e., new, recent information that gives distinctive insights inside your core focused subjects.
And guess what? This strategy aligns with how Google now understands and ranks content material:
- Entities > key phrases: Google doesn’t simply match “search strings” anymore. It understands ideas and audiences (and the way they’re associated) by the information graph.
- Content material constructed round folks, issues, and questions: You’re not answering one question while you optimize for a subject as a complete; you’re fixing layered, real-world challenges in your viewers.
- Content material journeys, not remoted posts: Subject-first methods map content material to completely different consumer varieties and their stage within the journey (from studying to purchasing to advocating).
- Extra sturdy visibility + stronger hyperlinks: When your website deeply displays a subject and tackles it from all angles, it attracts each natural queries and pure backlinks from folks referencing actual perception and utility.
- That E-E-A-T we’re all alleged to give attention to: Kevin discusses this a bit extra when he digs into Google High quality Rater Pointers in constructing and measuring model authority. However that is an absolute no-brainer: Taking a topic-first strategy actively works towards establishing Expertise, Experience, Authoritativeness, and Trustworthiness.
I needed to know the way others are doing this, so over on LinkedIn, I requested in your ideas and questions.
Listed below are some that caught out to me that I feel we are able to all profit from contemplating:
Lily Grozeva asks: “Is masking a subject and establishing a model as an authority on it nonetheless a quantity sport?”
My reply: No. I feel Backlinko is an efficient instance. The location constructed unbelievable visibility with just some, however very deep guides.
Picture Credit score: Kevin Indig
Diego Gallo asks: “Any recommendations on resolve if a query ought to belong to a web page or be its personal web page?”
My reply: “In my expertise, one technique to decide that’s cosine similarity between the (tokenized, embedded) query and the principle subjects / intents of the pages you could choose from.”
Diego additionally left an excellent tip for masking all related intents: Construct an “intent template” for every web page (e.g., product touchdown web page, weblog article, and so forth.). Base the template on what works properly on Google.
Picture Credit score: Kevin Indig
Matthew Mellinger referred to as out that you need to use Google’s Folks Additionally Requested inquiries to get readability on which inquiries to reply on a web page.
Picture Credit score: Kevin Indig
By the best way, you may also use the intent classifier I constructed for premium subscribers for this job!
Gianluca Fiorelli put a cool analogy on the desk:
Picture Credit score: Kevin Indig
Not Strategizing With A Subject-First Mindset? You’re Outdated
Subsequent week, we’re going to take a deep take a look at operationalizing a topic-first Search engine optimisation technique, however listed below are some closing ideas.
Whereas there are such a lot of unknowns within the present search panorama, there are a number of truths we are able to floor ourselves in, whether or not optimizing for search engines like google or LLMs:
1. Your model can nonetheless personal a subject within the AI period.
As proven within the information through the UX examine of AIOs, model/authority is now the primary gate customers stroll by when contemplating a click on off the SERP, search intent relevance the second; snippet wording solely issues as soon as belief is secured.
If individuals are going to click on, they’re going to click on on the acquainted and authoritative. Be the topical authority in your areas of experience and choices.
Have your model present up once more, and once more, and once more in search outcomes throughout the subject. It’s easy, but it surely’s laborious work.
2. I don’t assume specializing in a topic-first mindset might backfire in any means (in 2025 or past).
Demonstrating to your core ICPs – whether or not they discover you through paid advertisements, natural search, LLM chats, socials, or word-of-mouth – by authoritative, branded web site content material that you simply perceive the subjects they care about, questions they’ve, and supply the options for his or her wants particularly solely builds belief … regardless of how your model is discovered.
3. Construct matter techniques, not simply articles or pages.
Integrators have to take a web page out of the aggregator’s product-led Search engine optimisation playbook: Create a complete system (just like TripAdvisor’s hundreds of thousands of programmatic pages supported by user-generated content material (UGC) opinions, however you don’t want hundreds of thousands 😅) constructed round your subjects of experience that deal with views, options, and questions round every persona kind for every sector you serve.
Construct the organizational construction inside your website that makes these subjects and personas simple to navigate for customers (and straightforward to crawl and perceive for bots/brokers).
4. Persona or ICP-based content material is extra helpful, much less generic, and constructed for the subsequent period of customized search outcomes.
‘Nuff stated. In the event you strategize matter optimization by the lens of personas (even to the purpose of together with actual interviews, surveys, feedback, and ideas from these persona varieties), you’re including to the dialog with depth and distinctive information.
In the event you’re not constructing audience-first content material, does optimizing for LLMs and search bots even matter? You’ll acquire visibility, however will you acquire belief when you lastly earn that click on?
Featured Picture: Paulo Bobita/Search Engine Journal
