What to do as SEO traffic drops over the next 3 years

Hey, it's Amit here again.

Over the next three years, SEO traffic is likely to become less predictable.

Not because search is dying  - but because discovery is spreading across more surfaces.

The mistake most teams will make is trying to protect traffic, instead of reinforcing the signals that still compound when traffic fragments.

The Reuters Institute recently published its 2026 Journalism, Media & Technology Trends report.

A key theme running through it:
many publishers expect a 40% decline in search traffic over the coming years.

Source:
https://reutersinstitute.politics.ox.ac.uk/journalism-media-and-technology-trends-and-predictions-2026

That expectation isn’t unreasonable.

AI answers reduce clicks.
Social referrals are less reliable.
Discovery is no longer concentrated in one place.

But this isn’t a prediction about relevance disappearing.

It’s a prediction about distribution changing.

People still:

They’re just doing it across more places:
Google, AI tools, YouTube, editorial sites, communities.

Which changes the question from:

to:

Not to mention that much of the traffic sites were getting - were probably not converting anyway. I’d argue much of it was as a result of SEO strategy that was a bit “vanity metric” heavy - understandable in a world where we were all being KPI’d on YoY traffic growth. 

Gaetano Di Nardi summed it up well in his article on SearchEngineLand about “The Great Normalization”:

In this environment, surface-level content doesn’t hold up.

Covering the obvious answer isn’t enough.
Repackaging consensus isn’t enough.

To maintain visibility  -  human and AI  -  content increasingly needs to:

This is no longer about rankings alone.
It’s about teaching systems what your content is actually for. This is straight from the patent for information gain.

Even strong content often struggles on its own in competitive spaces.

Modern search systems don’t just ask:

“Is this good?”

They ask:

“Who else treats this as credible?”

I’d argue that for the foreseeable future, it's not possible for search (be it AI search or traditional search)to be completely reliant on on-page signals - it MUST always have something external to the website’s own self.

That’s where off-page signals come back into focus. 

As clicks become less predictable, new link heuristics are popping up.

Now, since HCU and when AI came onto the scene, one of the loudest is the idea that “links on pages with traffic are the best links” -  because traffic supposedly signals quality.

It sounds neat.
It also starts to fall apart quickly.

If SEO traffic declines or fragments into AI interfaces, how reliable is any traffic estimate long-term?

So, traffic is an output, not a cause.

Search systems don’t reward pages purely because they get traffic.
They reward pages because they’re relevant, trusted, and well-positioned within a topic  -  which may result in traffic.

Optimising for traffic as the proxy means optimising for something increasingly unstable. So be careful not to fall into that trap.

You’ll also hear people talk about semantic links  -  links placed in relevant content.

That direction is broadly correct.
But many stop too early.

If anyone can place a link on that page via payment, templating, or mass outreach, the semantic value decays fast.

The real question is:

Can this link only exist if the content genuinely belongs there?

That constraint  -  editorial judgment, audience fit, topical depth  -  is the actual signal. 

And of course, if the content itself follows the same practices we already discussed in our previous newsletters like information gain, its on a site with brand signals, etc. 

AI systems don’t need clicks to infer relevance.

They learn from:

A highly relevant link or mention on a low-traffic page can still matter  -  because it teaches the system what you are about, not just who visited you.

That’s why obsessing over traffic alone is fragile.

And why the safest approach remains:

Build links and mentions on pages and domains that are hard to access unless your content genuinely fits.

Recently, I refreshed one of my higher-competition pages around link building courses. Not only did I explain every course, but I talked about specifics, that NOBODY else seemed to talk about on their listicles. It was stuff only I could have known, because I either took the course, or I know people who have.

That on-page work helped.
But the real movement came only when I did off-page work as well.

The page or entity was also reinforced with:

That’s when things shifted.

Rankings improved.
And more interestingly, LLM-driven click-through increased too AND it appeared on the AI Overview.

Not instantly.
Not cleanly.
But cumulatively.

(Making on appearance on the AI Overview on Google USA for “link building training”)

If traffic becomes less predictable, the response isn’t panic.

It’s resilience.

That means:

I still recommend you or someone on your team getting hands-on with things like HARO-style pitching  - because it teaches you what editors and systems actually reward.

Traffic may fragment over the next three years.

Authority still compounds.

The teams that understand that difference  -  and act on it - will feel the shift far less than everyone else.

Amit Raj
The Links Guy

P.S.
The people who adapt early won’t “lose traffic”  - they’ll just stop depending on it.