- Technical SEO only matters up to the point where search engines can crawl, render, and index what you want them to rank. Past that, most gains come from content, links, and product.
- AI search and large language models will reward real brands that do real work in the real world, not just sites that tweak meta tags and schema.
- Programmatic and AI-driven content can work, but only if you combine it with real expertise, data, and editing instead of spamming out thousands of thin pages.
- Internal links, redirects, and content refreshes quietly move more traffic than many flashy tactics SEOs love to argue about.
If you want to grow with SEO today, you have to stop obsessing over clever hacks and start thinking like a marketer who happens to understand search. Fix what is actually broken, publish content people can use, build a reputation beyond your website, and let SEO plug into that. That is the shift most people resist, and it is also why they stay stuck.
Why most people are stuck chasing the wrong SEO problems
When I talk to founders, in-house teams, or agencies, I keep seeing the same pattern. They are sweating over subdomain vs subfolder, schema experiments, or some new AI-only endpoint, while basic issues like indexing, internal linking, and real demand capture are half-done at best.
And I get it. The shiny stuff feels smart. It gives you something to tweet about. But if you care about traffic and revenue, the boring fundamentals still carry most of the weight.
Technical SEO is the most important thing on your site until search engines can crawl, render, and index you properly. After that, it quickly becomes the least important lever.
So I want to walk through a more grounded way to think about SEO in 2024 and beyond. Not theory, but what tends to move numbers across different types of sites: SaaS, local, content sites, and bigger brands.
Technical SEO: where it actually stops mattering
There is a point where more technical work is just polishing the hood while the engine is missing. That point comes much sooner than most people admit, especially tool vendors and technical SEOs.
The real checklist that matters
You do not need a 120-page audit for most sites. You need to answer a short list honestly:
- Can search engines crawl your key URLs without hitting weird blocks or loops?
- Is the important content rendered without relying on fragile JavaScript tricks?
- Are your main pages indexable and actually indexed?
- Is the site reasonably fast and stable on mobile for humans?
- Are error states (404, 500, etc.) under control, not out of hand?
If the answer is yes to all of that, you are past the phase where technical tweaks are your main growth driver.
| Technical situation | What to do | Impact potential |
|---|---|---|
| Key pages not indexed | Fix crawl blocks, canonicals, robots tags, sitemaps | Very high |
| Migration with missing redirects | Map and reclaim old URLs with proper 301s | Very high |
| JS-heavy site hiding core content | Render content server-side or in simple HTML | High |
| Large site with broken hreflang or geo routing | Fix the logic and let recrawling catch up | High, but slower to show |
| Everything above solved | Shift focus to content, links, product, brand | Technical impact now low |
When technical really matters (and when it does not)
Technical SEO matters a lot in edge cases: large international sites, big migrations, complicated JS apps, and anything with messy legacy systems. In those worlds, one bug can burn actual money, not just rankings.
An example that sticks with me: a marketplace that showed the wrong country version in search, then blocked checkout for users on that wrong version. People clicked, browsed, then got blocked at payment. They were losing serious revenue every day.
The fix was not heroic. It was logic and patience. They added a clear warning and a country switch prompt. Then they waited for recrawling to catch up, which took months, not days.
If a single technical bug can freeze millions in daily revenue, fix that first. If your biggest problem is a CSS warning in a tool, you probably have more useful things to do.
For most normal sites though, once you pass the basic threshold, each new technical task has shrinking returns. That is often the moment people should pivot to content and links instead of trying to squeeze another 2 percent from Lighthouse.
Redirects: the small change with oversized impact
If there is one technical lever that keeps outperforming its effort, it is smart redirects. Not fancy patterns, just basic reclaiming of what you already earned.
- Old blog posts that got links but now 404.
- Legacy product URLs from past designs.
- Acquired sites that were half-merged then forgotten.
When you connect these properly to relevant live pages, you are often pulling years of link building into a single afternoon.
| Scenario | What people do | Better move |
|---|---|---|
| Hundreds of 404s with links | Ignore them, focus on new links | Map to closest relevant URLs with 301s |
| Expired campaign microsites | Leave them dead | Redirect to current category or resource hubs |
| Old domain from an acquisition | Keep it parked | Audit and redirect quality URLs into your main site |
There is still a myth that 301s or 302s do not pass value or pass only a tiny fraction. That was shaky years ago and is even weaker now. In practice, you can watch a page hand off its rankings to the new target with a proper redirect in place.
Is it perfect? I doubt it is perfect. But it is close enough that ignoring it is just throwing away equity you already paid for.
Authority, internal links, and why most sites waste their strength
People love domain-level scores. They look simple. They feel like a single answer to a complex problem. But they also trick teams into ignoring where the real advantage sits.
Domain metrics vs page strength
Most third-party authority scores are side effects of something more basic: how many strong pages a site has and how those pages connect internally. In other words, page-level strength that spills over into a crude domain picture.
The problem is that many brands stop at that picture. They chase a higher domain number instead of using the strength they have. That is why even big sites with great press still struggle to rank on bread-and-butter terms.
If you have a strong domain and weak internal links, you are driving a powerful car in first gear.
I like to think about authority with one question: if you publish a new page and give it a few good internal links, how competitive is that page by default? That is the real value behind the metric everyone stares at.
How to internal link without making a mess
Internal linking does not need to be art. It just has to be thoughtful and visible. The mistake I see is either near-zero linking or absurd overuse with every second word turned blue.
- From high-traffic pages to important money pages or hubs.
- From related articles to core guides or service pages.
- From navigation and footers to the pages that truly matter, not random ornaments.
If you are not sure where to start, pick your top 20 pages by organic traffic and ask one question for each: what should this help rank better?
| Page type | Typical mistake | Better linking pattern |
|---|---|---|
| Blog posts | Only link to other posts | Link to posts and related product / service pages |
| Category pages | No links back to top guides | Include 2-5 text links to deep content and FAQs |
| Homepage | Huge slider, few real links | Lean layout with clear links to main money pages |
If your pages start looking like a sea of links, pull back. You can also shift some links into neat sections below the text. For example, add a small box titled “Further resources” instead of stuffing five links inside one paragraph.
Why subdomains vs subfolders is overblown
The subdomain vs subfolder argument keeps coming back. The reality is that it mostly reflects one thing: how you connect sections of your site.
When a blog lives on a subdomain and is not linked in the main navigation, it is basically treated like a separate site by users and by crawlers. When you move it to a folder and wire it into menus and internal links, you suddenly see a jump. People then blame the subdomain when the actual fix was linking.
If you treat a subdomain as a real part of your property and link to it in a meaningful way, it can perform very close to a subfolder setup. The bigger problem is that almost nobody does that. They throw content on a subdomain and forget about it.
So yes, folders are usually safer, cleaner, and simpler. But not for magical reasons. Just because they push people to treat content like part of the main site instead of some separate island.
Content that ranks today: length, quality, and what actually matters
There is a strange tension in SEO around content length. Some people still cling to the idea that long posts must rank better, others swear word counts are irrelevant. Both views miss the nuance a bit.
Does word count matter?
Word count is not a direct ranking factor in the neat way some people hope. There is no “2,000 words beats 800” rule. But length often tracks something else: whether you covered the topic properly or padded it with filler.
My rule is simple and boring. Say what you need to say in the amount of words it actually takes. No extra paragraphs just to hit a number.
- A definition page might be 250 words and still be perfect.
- A complex how-to might need 1,500 words with clear sections and visuals.
- A buying guide might split into a main page plus several detailed subpages.
If you catch yourself repeating the same point with new phrasing three times in a row, that is not depth. That is padding. Search systems are getting better at seeing the difference.
What quality content really looks like
Quality is not some vague vibe. You can break it down into a few clear traits you can check:
- Readable: short paragraphs, simple sentences, headings that make sense.
- Direct: the main answer appears early, not buried six scrolls down.
- Grounded: real data, examples, screenshots, quotes, or workflows.
- Focused: one main intent per page, not six half-baked ones jammed together.
- Useful: you can apply it to do something or decide something.
If a reader needs to skim past three anecdotes before they see a clear answer, the content is serving your ego, not their query.
One trick I like is to do a quick “bluff” check on every important piece. Bottom line up front. Would someone get the main value if they only read the first 4-5 lines? If not, I rework that intro.
Refreshing and pruning: boring work that adds up
On many established sites, new content is not the main win. Refreshing and pruning is. You often get a faster return by fixing what you already have than by adding one more new post.
That might sound dull, but when you do it with intent, it compounds over time.
- Rewrite outdated sections on pages that still get impressions.
- Merge two thin posts into one solid guide with a redirect.
- Remove sections that no longer match how your product or service works.
- Improve structure with clearer headings and better internal links.
Sometimes people are scared to cut or merge because they feel every URL is precious. It is not. If two pages fight for the same query and neither really wins, combining them and sending all signals to one can be a net gain.
Programmatic content without wrecking your site
Programmatic SEO is one of those areas that can be brilliant or can quietly poison the rest of your property. The difference sits in what you feed it and how much you care after launch.
Most failed attempts look similar:
- Tens of thousands of pages generated from the same generic template.
- Thin content where each page is only a city name or a minor phrase swap.
- No real data, no visuals, no clear reason for the page to exist.
- No ongoing improvement once the initial build goes live.
That pattern might gain traffic briefly. Then algorithms tighten the screws on site-wide quality and the bump turns into a dip, sometimes dragging good sections down with it.
Better versions of programmatic content look more like small products than static pages. They blend data, logic, and content, then get improved over time.
- Directories that show live metrics, not just a list of names.
- Tool pages that wrap real calculations or comparisons, not simple text.
- Location pages that pull in local reviews, partners, or event data.
In some cases, it even makes sense to host riskier programmatic experiments on a separate domain. That sounds harsh, but if you suspect a project might look spammy to a strict reviewer, you probably do not want it next to your core product or docs.
AI content, LLMs, and what actually helps you get surfaced
AI changed how content gets produced and how search works. Both sides matter. You can use AI to write faster, and you are also now competing inside AI answers where links do not look like they used to.
How much of your content should be AI-written?
People want a neat answer here. There is no single percentage that fits every site, but I think most people are scared of AI in the wrong way.
A huge chunk of any article is basic knowledge everyone already knows or can look up. That part does not need your personal flair. It just needs to be correct and readable. AI can handle a lot of that if you guide it properly.
The real edge is not “pure human” or “pure AI”. It is how well you mix them and how much real expertise you inject on top.
The pattern that works best for many teams looks something like this:
- Let AI draft the structure and basic sections using a good brief.
- Ask AI to suggest examples, then replace most of them with your own.
- Add data, stories, screenshots, and opinions from your team.
- Edit heavily for tone, clarity, and honesty.
So yes, many good articles might be 70-80 percent AI-assisted at the sentence level. But the important 20-30 percent (direction, data, details) still comes from people who know the topic and the audience.
Programmatic + AI: where people go wrong
Using AI inside programmatic setups is tempting. You can generate thousands of pages for the cost of a team lunch. But cheap does not mean smart.
The real risk is not speed. It is sameness. If you generate 50,000 variations of the same thin explanation, you are signaling that your site is comfortable flooding the web with noise. Search systems do not reward that in the long run.
Signs your AI-programmatic project will probably cause trouble:
- You cannot explain why a human would ever bookmark one of those pages.
- You do not have a plan for updating or improving them after launch.
- The only value is that they contain a keyword and some generic text.
If, instead, AI is one layer on top of strong data or workflows you already have, your odds are much better. For example, you might use AI to turn structured data into readable summaries, then layer tools and filters on top for users.
LLM.txt and other shiny distractions
Lately there has been a rush to create special files for AI crawlers: LLM.txt, custom “AI” endpoints, entire separate content versions that only bots will see. Most of this is a distraction at best.
Large models read raw text. If they can crawl your HTML, they can see your content. They do not need a separate markdown-only URL just to understand you. And some of the “standards” that popped up are not standards at all, just blog posts that spread fast.
So far, I have not seen solid evidence that adding an LLM.txt file by itself does anything measurable for how often you are cited in answers. That might change at some point, but I doubt that is where the real leverage will be.
Your time is better spent on things like:
- Making sure your key pages are cleanly structured and easy to crawl.
- Reducing JS bloat that hides text behind complex rendering.
- Publishing content that gets linked, talked about, and referred to.
How to show up in AI answers as a small business
If you run a local business or a niche site, the advice for LLM visibility is almost boring. It sounds like offline marketing mixed with clean SEO basics, which is exactly what it is.
- State clearly what you do, where you operate, and who you serve.
- Keep your details consistent across your site, maps, directories, and profiles.
- Get real reviews and mentions from real people and real organizations.
- Participate in your community: events, partnerships, local media, charities.
LLMs lean on what the rest of the web says about you. If the only trace of your business is a thin homepage and a half-filled profile somewhere, you are easy to ignore.
Good SEO has always bled into PR, product, and brand. AI just made that overlap harder to hide.
The nice side effect is that this work does not only help AI answers. It helps classic search, social proof, and conversions as well.
The myth of schema as a secret AI signal
Schema has turned into another area where people promise more than it can reasonably deliver. I keep seeing claims that structured data will suddenly make LLMs respect your brand or cite you more often.
Schema has clear value where it triggers real features: stars, FAQs (when they exist), product details, event snippets, and similar. That part is not in question. Outside of that, its impact is much less dramatic.
From a machine reading point of view, schema is just extra hints that sit alongside the raw text. If your regular content is a mess, no amount of JSON will save you. And if your content is already clear, schema is more of a nice add-on than a magic key.
Growing new sites vs competing with giants
Starting from zero is very different from defending or growing an established brand. The mistakes also differ. New sites chase huge terms too fast. Big sites waste their authority on half-hearted assets.
If you are starting from scratch
If I had to grow a brand new site without leaning on an existing domain, I would not start by debating whether I need a complex content calendar. I would start by understanding where the money can actually come from.
- List the core problems your product or service solves.
- Map those problems to search intents across the funnel.
- Pull keywords from a handful of direct and indirect competitors.
- Group them by topics and estimate opportunity using traffic potential, not just single-term volume.
Then I would pick a mix of:
- High-intent pages that can make money if they rank.
- Mid-funnel guides that build trust and internal linking power.
- A few broader “learn” pieces that can earn links.
In parallel, I would do real-world link work: industry directories, local press, genuine partnerships, suppliers, and complementary businesses. Not guest posts on random sites that share no audience with me.
And I would keep expectations grounded. If a behemoth owns a vanity keyword like “mobile” or “CRM” with a massive page and deep internal linking, a new site is not going to take that soon, if ever. That time is better spent capturing the thousands of specific, often neglected queries where intent is clearer and competition is lazy.
If you are a smaller player against huge brands
Competing with big companies is mostly about picking your battles and choosing angles they ignore. Big teams are busy. They have politics, processes, and guardrails. You can use that slowness to your advantage.
- Target longer, more specific queries with clear purchase or signup intent.
- Answer questions big brands think are too small to care about.
- Create better, more focused pages around those gaps instead of generic summaries.
Large brands often publish pages that technically cover a topic but clearly did not have a passionate editor behind them. That is your opportunity. If they phone it in with 600 bland words, you can beat them with something sharp, opinionated, and actually helpful.
Programmatic risk, site-wide quality, and when to separate domains
One area where I see bigger teams get in trouble is when programmatic projects pollute an otherwise strong main site. The traffic from the new section looks exciting for a few months, then quality systems catch up and the broader site takes a hit.
Sometimes the fix is as drastic as moving the experimental section to a separate domain and cleaning up the main one. That is a painful call. But if analytics show that your legacy tools, docs, or blog were doing fine until you added a large set of low-value pages, it is at least worth testing the split.
Personally, I would rather keep the core domain focused on things I am proud to show a manual reviewer: product, docs, real guides, high-utility tools. Anything that feels like it might age poorly or look spammy under a harsher lens can live elsewhere.
SEO tactics that deserve less of your time
There are a few habits I wish the industry would let go of, or at least push to the bottom of the list.
- Endless schema experiments that never connect to real SERP features.
- LLM-specific hacks with no evidence of consistent benefit.
- Obsession with daily ranking swings for single keywords instead of overall visibility and conversions.
- Ritualistic audits full of minor warnings nobody will ever act on.
If a tactic does not help users, does not clearly support your business model, and cannot be tied to any measurable change, it probably does not deserve hours of your week.
I am not saying these things are always useless. Some schema is useful. Some technical tuning matters. Some experiments are worth running. But they should not crowd out the basics that quietly produce most of your results.
What to actually do next
If you feel overwhelmed by conflicting SEO advice, simplify your next steps. You do not need a perfect plan. You need a clear starting point and a bias toward doing work that has a chance to matter.
- Audit your indexation and fix clear technical blockers.
- Clean up redirects and reclaim links from old URLs.
- Map your top keywords to real business outcomes, not just traffic.
- Write or improve a small set of pages that truly answer those intents.
- Add thoughtful internal links from your highest-traffic pages.
- Set up a simple process for updating content that already ranks.
If you do these things honestly, you will already be ahead of many competitors who are too busy chasing the latest tactic to touch the work that moves the needle. It is not glamorous, but it is sustainable, and it is closer to how search actually works than most of the noise you see online.
Once that foundation is in place, you can get a bit more ambitious. Try a small programmatic experiment that actually delivers value. Blend AI into your workflow where it saves time instead of where it creates thin duplicates. Test, measure, adjust. And do not be afraid to say no to “tricks” that sound clever but do not pass a basic sanity check.
SEO is not dead, but it is also not magic. It is just a very sensitive way of measuring how well your content and brand line up with what people care about. If you keep that in mind, most of the strategic decisions get a lot simpler.
Need a quick summary of this article? Choose your favorite AI tool below:
