DEV Community

Cover image for SEO in 2026 feels broken. So I’m trying something else
Andrew Rozumny
Andrew Rozumny

Posted on

SEO in 2026 feels broken. So I’m trying something else

I’m not an SEO expert.

Just a developer trying to get at least some users to the things I build.

And lately it feels… strange.


What I’m seeing

I published several posts on Dev.to.

Some were original — 20–80 views

Some slightly adapted — maybe 100–300

But overall, nothing really moves.

At the same time:

  • Google traffic → almost zero
  • My own site → barely indexed
  • Social → small spikes, then nothing

Feels like you start from zero every time.


The weird part

When I write a normal article — nobody reads it.

When I build a simple tool page for one very specific use case —

it has more chance to be indexed and actually used.

Not huge numbers. But different behavior.

This doesn’t look like classic SEO anymore.

Feels more like:

content that ranks → vs → answers that get extracted


The moment it clicked

I asked ChatGPT:

"regex for email validation"

It returned:

  • working regex
  • explanation
  • edge cases

Everything in one place.

No need to open anything else.

So why would someone click search results?


What I’m trying now

Instead of writing more content, I started building tools.

Small ones. Very simple.

  • JSON formatter
  • Regex tester
  • Base64 encoder
  • SSL checker
  • IP lookup

Right now I’m building towards ~100 tools.


Early signals

Still early, but I see:

  • tool pages get indexed faster
  • people actually stay on them
  • some long-tail queries appear
  • structured parts sometimes show up in AI answers

Again, small numbers.

But not the same pattern as blog posts.


How I structure a page now

I stopped thinking in terms of:

tool + some SEO text

Now each page looks more like a small answer system:

  • clear title for one exact problem
  • working tool
  • short intro
  • 2–3 useful sections
  • examples
  • FAQ
  • glossary
  • related tools
  • links to nearby tools

Data structure behind it

Instead of writing everything manually, I model pages like this:

{
  slug: "json-formatter",
  title: "JSON Formatter",
  searchIntents: [
    "format json online",
    "pretty print json",
    "json beautifier"
  ],
  useCases: [
    {
      title: "Debugging API responses",
      description: "Make JSON readable"
    }
  ],
  glossary: [
    {
      term: "JSON",
      definition: "Structured data format"
    }
  ],
  relatedSlugs: ["json-validator"],
  clusterSlugs: ["json-validator", "json-diff"]
}
Enter fullscreen mode Exit fullscreen mode

Shared layout

All pages use same structure:

<ToolPage>
  <ToolHeader />
  <ToolUI />
  <HowToUse />
  <UseCases />
  <Examples />
  <FAQ />
  <Glossary />
  <RelatedTools />
  <ClusterTools />
</ToolPage>
Enter fullscreen mode Exit fullscreen mode

Only the tool itself is different.

Everything else is predictable.


What I think is changing

Not sure yet, but looks like:
Content is no longer the main unit.
Structured answers are.

Blog posts are easy to summarize.

Tool pages with clear structure are harder to replace.


Not sure where this goes

Honestly, I don’t have a conclusion.

Just testing.

I’ll keep building tools, structuring pages, and see what happens.
If something works — will share numbers.

If not — also useful result.


Curious about others

Are you seeing any traffic from AI tools?

Or still mostly Google?

Top comments (3)

Collapse
 
jonathanapplebaum profile image
Jonathan Applebaum

I think there is real truth in this.

A lot of blog-style content is now easy for LLMs to compress into a usable answer. Something like “how to extract IPv4 and IPv6 addresses from plain text using regex” does not really require a click anymore if the model gives a solid answer directly.

A tool page is harder to replace because it gives utility, not just explanation. That feels like a real shift.

Collapse
 
andrewrozumny profile image
Andrew Rozumny

Yeah, this is exactly the line I’m trying to understand right now.

Anything that can be reduced to “question → answer” is basically owned by LLMs now. No real reason to click if the answer is good enough.

But tools are different because they’re not answers — they’re actions.

What I’m experimenting with is pushing even further: not just tools, but pages that expose intent (use cases, inputs, patterns) in a structured way, so both users and models can interact with them, not just read them.

Feels like the shift is from “content that explains” to “interfaces that do”.

Still figuring out where that boundary actually is though.

Collapse
 
sebhoek profile image
Seb Hoek • Edited

Thanks for writing this up! When I discuss SEO with some chat, it comes up with the same approach: write some little tools and offer it to others on my website.

But should we all put regex testers on our website now just to promote our hobby projects? I am not sure that this is the right approach for me. But the current times are weird anyway, so maybe that's the way to go this year.

Curious to learn more about the results of your experiments - keep them coming!