The Provability Gap: What AI Is Really Doing To Agency Value

Walk into almost any agency pitch this year and you'll hear some version of the same claim:

We're ahead on AI.

We understand AI search.

We're already helping clients navigate the shift.

That's not necessarily wrong. But when almost everyone in a category starts making the same claim, the claim itself stops being differentiation.

What matters next is proof.

And right now, there's a significant gap between what agencies are saying about AI and what many can actually demonstrate.

I think of it as the provability gap.

The Claim Everyone Is Making

AI-mediated search has moved quickly from an emerging capability to something clients are actively asking agencies about.

That's creating understandable pressure for agencies to establish a position.

The problem is that positioning language can move much faster than organizational capability.

“AI-enabled” is easy to say.

What's harder is explaining exactly what the agency does differently, what that changes for the client, and how the agency knows it's working.

That's where differentiation begins.

The Agency's Own Business Is a Useful Test

Recent RSW/US research illustrates the disconnect.

85% of agency principals believe their agencies are well-positioned and uniquely differentiated.

Yet only 11% say their own websites are “very optimized” for AI-mediated search.

And while 66% say they're helping clients optimize their websites for this new search environment, relatively few appear to have fully applied the same discipline to themselves.

That's a version of the old cobbler's-children problem, but there's a larger issue underneath it.

If AI search is becoming part of the agency's value proposition, the agency itself should eventually become one of its strongest demonstrations of that capability.

Not because every experiment has to succeed.

Because firsthand experience creates something positioning language can't:

credibility.

The Measurement Problem Is Real

There's another reason proof remains difficult.

AI-mediated search is still developing as a measurable marketing channel. Attribution isn't as mature as it is in traditional search or paid media, and agencies are still adapting to changes in organic visibility and ROI measurement.

For example, recent AgencyAnalytics research found that 41% of agencies report new challenges measuring ROI as traditional attribution becomes less reliable.

That's important context.

The answer isn't to pretend measurement is more precise than it is.

It's to become very clear about what can be measured now, what can't, and what evidence you're using in the meantime.

That honesty can actually strengthen the client relationship.

Why This Matters for Agency Value

This isn't ultimately a story about search technology.

It's a story about what an agency can credibly claim to be good at.

Every emerging capability follows a similar cycle.

At first, simply having the capability creates differentiation.

Then competitors adopt it.

Then everyone claims it.

Eventually, the market asks for proof.

That's the point at which real capability separates from positioning.

And that's where agency value gets interesting.

An agency with a documented methodology, firsthand experience, measurable results where measurement is possible, and a clear understanding of where the limits are has built something more defensible than an agency with an AI page on its website.

The claim is easy to copy. The capability behind it isn't.

Closing the Provability Gap

Agency founders don't need to build an enormous AI-search practice overnight.

I'd start much more simply.

First, use your own agency as the test case. If you're advising clients about AI-mediated discovery, understand how your own agency appears in that environment.

Second, separate what you know from what you're still learning. Don't turn uncertainty into marketing language. Clients are sophisticated enough to understand that this market is evolving.

Third, document the method. What did you change? Why? What happened? What did you learn? A documented process becomes intellectual property even before it becomes a perfect case study.

Fourth, build proof before you build the claim. One credible example is more valuable than another paragraph explaining that your agency is “AI-forward.”

The Bigger Point

AI isn't eliminating agency differentiation.

It's creating another test of it.

Right now, everyone is racing to develop an AI story.

That's temporary.

Eventually clients will stop asking, “What are you doing with AI?” and start asking, “What can you show me?”

The agencies preparing for that question now will have something much more valuable than a good answer.

They'll have proof.

CTA — DIAGNOSTIC

If AI has become part of your agency's positioning, ask one simple question: what can we demonstrate today that proves the claim? If the answer isn't clear yet, that's the gap worth working on.