What Are We Optimizing?

For years, I’ve struggled to answer a question I should be able to answer easily: what is SEO?

Not how to do SEO. Not how to explain its outputs. Not how to defend its budget, build a roadmap, audit a website, write a title tag, or produce the familiar theatre of recommendations. I mean the more basic question: once you strip away the tactics and the visibility outcomes, what is left?

The word that bothers me is “optimization.”

What exactly are we optimizing?

Is it the experience? The content itself? The infrastructure? The path someone takes through a website? The way information is organized? The way a search engine interprets a page? Or are we, somehow, claiming to optimize the search engine itself?

Depending on who you ask, you get different answers. That has always been part of the problem. For many practitioners, SEO is still framed as the work of placing X better on platform Y. Get this page ranking. Get this keyword moving. Get this snippet appearing. Get this brand mentioned. The object changes, the platform changes, the acronym changes, but the underlying reflex stays the same.

Do the things. Hope the outcomes follow.

This is why SEO strategies so often become lists of tasks. Fix these titles. Rewrite these descriptions. Add these headings. Consolidate these pages. Implement this schema. Publish this content. Monitor these rankings. It is not that these things are always useless. Many of them are necessary. Some are even valuable. But a list of actions is not a theory of change.

And that, I think, is where my discomfort begins.

The problem is not that SEO has tactics. Every discipline has tactics. The problem is that SEO often mistakes its tactics for its theory of change.

I feel this most acutely in enterprise work. Recently, a colleague asked whether we had a way to programmatically retire pages that had received zero clicks or impressions. Technically, it was a reasonable question. Strategically, it felt like a symptom.

Because even if we could build that system, even if we could convince the right team to prioritize it, even if we could automate the retirement of underperforming pages at scale, what would we really have solved? Would we have improved the customer experience? Clarified the information architecture? Helped the business understand demand? Increased trust? Made anything more useful?

Or would we simply have found a more efficient way to clean up after years of undisciplined publishing?

On the days I do tactical SEO, I often feel like I’ve failed.

Not because the work is beneath me. It isn’t. Page titles, meta descriptions, headings, internal links, redirects, canonical tags, structured data — these things matter in the way maintenance matters. They help keep the machine legible. They prevent avoidable failure. They make systems cleaner than they would otherwise be.

But maintenance is not the same as strategy.

There is something deeply numbing about spending too much time on the small theatre of SEO best practice. Reword this heading. Adjust that title tag. Mention the target query earlier. Make the meta description more compelling, even though it may not be used. Add another recommendation to the backlog. Wait for impact. Explain why impact is difficult to isolate.

So much of this work rests on inherited assumptions and correlative observations. We act with confidence because confidence is easier to sell than uncertainty. But do we truly know how Google ranks and re-ranks documents, passages, entities, brands, and sources? Do we understand the weighting, the thresholds, the interactions, the feedback loops? Or have we built an industry around observing surfaces and naming the ripples?

Schema markup is a useful example because it should have been one of the more intellectually interesting parts of SEO.

In theory, structured data is about meaning. It asks us to define entities, attributes, relationships, and context. It forces a business to say what something is, how it relates to other things, and why it matters. Done well, schema is not decoration. It is a semantic layer.

But much of the time, we treated it as another visibility lever.

Add FAQ schema. Add Product schema. Add Article schema. Validate it. Ship it. Report that it was implemented. Hope for a rich result. Then, when the rich result disappears or Google changes how it displays things, the supposed strategy disappears with it.

That is the pattern in miniature. Something with conceptual depth gets reduced to an implementation task because implementation is easier to sell, easier to execute, and easier to measure.

Schema should have pushed us toward clearer meaning. Instead, too often, it became another checklist item.

And yet, I don’t think SEO is meaningless. I think the most meaningful parts of the work are often the least recognizable as SEO.

Recently, the most rewarding work for me has been solving localization.

Not the shallow version, where English keywords are translated into other languages and placed into the same content template. Translating keywords is hard enough. Translating entire pages reveals something more uncomfortable: the original page was never as universal as we assumed.

A US page carries US assumptions. It assumes a certain reader, a certain level of context, a certain commercial rhythm, a certain tolerance for persuasion, a certain way of structuring proof. Move that same page into French, German, Japanese, or Korean and those assumptions become visible very quickly.

What sounds clear in one market may sound blunt in another. What feels concise in English may feel underdeveloped elsewhere. What counts as trust, authority, clarity, humility, evidence, or usefulness changes with language, culture, and expectation.

This is the kind of work that still feels alive.

Not because it is more glamorous than metadata or schema, but because it forces better questions. Not “how do we rank this page in another country?” but “what does this audience need in order to understand, trust, and act?”

To answer that, we had to move beyond keyword translation and into research: linguistics, cultural expectations, market maturity, search behaviour, content design, and scholarly work on how people read, evaluate, and make decisions across languages. The task was no longer to replicate English at scale. It was to rebuild our model of international content from first principles.

That, to me, is where search work becomes strategic again.

It is not about placing X better on platform Y. It is about helping an organization confront the fact that meaning does not travel automatically.

And maybe that is the part I had been missing.

The more I think about it, the more SEO feels like marketing. Or at least, it feels like using empathy to help marketing help sales.

Not empathy in the sentimental sense. Empathy as a practical discipline. The ability to understand what someone is trying to do, what they already know, what they are afraid of misunderstanding, what evidence they need, what language they use, what language they distrust, and what would make the next step feel obvious rather than forced.

A keyword is only a trace of a need. Traffic is only evidence that something was visible. A ranking is only a position inside someone else’s system. None of these things tell us, on their own, whether we helped a person make sense of a problem or helped the business earn the right to be considered.

Seen this way, SEO becomes more useful again.

It sits between marketing and sales as a kind of translation layer. Marketing wants to shape demand. Sales wants to convert demand. Search reveals how demand expresses itself before either team has control over the conversation.

SEO, at its best, is the work of understanding how people seek, compare, doubt, and decide — then helping the business become more useful at each of those moments.

Which is why the current conversation around GEO feels so familiar.

Generative engine optimization, AI search optimization, answer engine optimization — whatever acronym wins, the pattern is already repeating. One camp insists it is just SEO. Another insists it is something entirely new. But when you ask either camp how they intend to influence outcomes, the answers often converge into familiar habits: structure the content, clarify the entities, build authority, improve technical access, earn mentions, answer questions, create useful material, make the brand legible.

The vocabulary changes faster than the underlying work.

What worries me is not that people are experimenting. We should be. What worries me is how quickly certainty has arrived. There is no shortage of experts claiming to understand AI search, brand visibility in LLMs, or how to optimize for systems whose inner workings they do not meaningfully understand.

But history repeats itself. In the early days of SEO, mystery created opportunity. The less we knew, the more confidently some people sold answers. Now, with large language models, the same reflex has returned. We are once again trying to make ourselves legible to systems we do not fully understand. And once again, the first instinct is to ask what tactic might make the system notice us.

But if localization has taught me anything, it is that meaning is fragile. It changes across language, culture, context, format, and expectation. If we still struggle to understand how meaning travels between people, we should be cautious about claiming mastery over how meaning is interpreted, compressed, retrieved, and synthesized by machines.

So maybe my issue was never really with SEO.

Maybe my issue was with the version of SEO that became too comfortable with surfaces. The version that confused activity with impact, implementation with strategy, visibility with value. The version that could describe what changed in a search result but struggled to explain why the business deserved to be found in the first place.

I still don’t know if “search engine optimization” is the right name for the work.

Maybe it never was.

Because I don’t think we optimize search engines. And I’m no longer satisfied saying we optimize content, either. That feels too small.

At its best, SEO is commercial empathy under technical constraints. It is the practice of understanding how people express need, how systems mediate that need, and how a business can become easier to find, understand, trust, and choose.

That definition gives me something sturdier to hold onto.

It explains why the tactical work can matter without becoming the whole discipline. It explains why schema is more interesting as meaning than markup. It explains why localization feels more strategic than keyword translation. It explains why AI search should make us more intellectually humble, not more eager to invent another checklist.

Maybe SEO is not a channel. Maybe it is not a set of tactics. Maybe it is not even a stable profession in the way we often pretend.

Maybe it is the work of making a business legible to demand.

And if that is true, then the question is not how we optimize for Google, or ChatGPT, or whatever interface comes next.

The question is simpler, harder, and much less comfortable:

Are we helping people understand why we deserve to be found?