Strategy

The uncomfortable truth: you don't have a top-of-funnel anymore

If AI answers the question, your "traffic strategy" isn't a strategy. Buyers are asking for a shortlist and getting an answer without clicking.

If AI answers the question, your "traffic strategy" isn't a strategy.

That line will irritate people who've spent a decade building sensible demand engines. It should. Because it's not a critique of effort - it's a critique of the underlying assumption that discovery happens on your website.

It doesn't. Not reliably. Not anymore.

A growing share of buying journeys now start - and in some cases finish - inside an AI surface. ChatGPT. Gemini. Perplexity. Copilot. The specific brand varies by company and by role, but the behavioural shift is consistent:

buyers are asking for a shortlist and getting an answer without clicking.

No visit. No form fill. No attribution.
Just… a decision made upstream.

And if your strategy depends on the click, you're measuring the wrong thing.

Your website hasn't disappeared. Its job has.

Let's be precise: your website still matters. It may matter more than ever.

But its role has flipped.

For years, the model was:

Search → click → browse → compare → convert

So you optimised for:

rankings
CTR
traffic volume
lead capture
retargeting pools

Now the model increasingly looks like:

Ask → answer → shortlist → verify

That last word is the giveaway.

Your website is no longer where discovery starts.
It's where claims are checked.

Which means your old funnel isn't "evolving". It's collapsed.

The top has been compressed into a single screen: an AI answer that names a few options, frames the trade-offs, and cites a handful of sources.

That's the gate now.

The new KPI isn't rankings. It's recommendation.

Here's the uncomfortable part marketers don't want to say out loud:

You can "rank" and still not exist.

You can be present on page one and still get zero consideration because the buyer never sees page one.

AI surfaces compress choice. They don't show ten blue links and let the buyer do the work. They do the reduction for them.

So the question isn't:

"Are we ranking for the right keywords?"

It's:

"When buyers ask AI for the best option… do we get named?"

And if we get named:

are we described accurately?
are we cited with credible proof?
are we compared fairly vs competitors?
are we recommended for the right use-cases - not the ones that hurt us?

Because two brands can "rank"…
…but only one gets named in the answer.

That brand makes the shortlist.

Why "more content" doesn't fix this

When teams feel the ground shifting, the default response is to publish.

More blogs. More landing pages. More SEO "coverage".

That can help - but only in the way adding more leaflets helps when the real problem is you're not on the shelf.

AI doesn't reward volume. It rewards signal.

Signal looks like:

clear positioning (what you are / aren't)
decision pages (who you're best for, and why)
comparisons (honest, specific, evidence-backed)
proof (case studies, certifications, data, reviews, listings)
authoritative third-party references
implementation and security clarity (especially for B2B)

If AI can't find those signals, it fills the gaps with guesswork or with competitor narratives.

And it doesn't do it maliciously. It does it because that's how the system works: it predicts a plausible answer based on what it can retrieve and what it has seen.

Your job is to make "plausible" unnecessary.

The new job: control what AI says when people ask

This is the punchline:

You don't need more content.
You need to control what AI says when people ask.

Not with tricks. Not with hacks. With clarity and evidence.

That means you need visibility into three things:

What AI engines say about you (across real buyer questions)
What they cite (and what they don't)
What to ship next to improve those answers (content, proof, comparisons, PR/listings, technical)

Without that, you're flying blind - optimising pages for clicks that never happen, and celebrating rankings that don't translate into consideration.

So what does a modern "funnel" look like?

It looks less like a funnel and more like a loop:

Measure what AI answers across your buyer prompts
Identify where competitors are being recommended instead
Ship the smallest set of changes that improve accuracy, proof, and framing
Re-test and track what genuinely moved
Repeat until your brand is consistently named for the right questions

It's not a one-off SEO project. It's a new operating cadence for discovery.

Try it on your brand

If you want to see what this looks like in practice, Odyssiant runs structured buyer prompts across engines, stores every answer, scores visibility, and turns weak results into a prioritised action plan - so you know what to do next, and can re-test to see what worked.