Pawel Grabowski. Independent consultant

You don't forget messages like this.

My client, Jane, Slack-messaged me their GSC screenshot one morning. It was showing a big jump in, well, pretty much everything. Impressions, clicks, positions... Basically, it's almost like one day, Google decided to finally show them in search.

She and her co-founder Benedikt were so excited. I could easily read between the lines of their ongoing chat... "this thing is working!"

It does, of course. But I remember it because of another thing:

That feeling of seeing that slow and hard effort to finally pay off. It's incredible. It gives you so much boost, and fills your head with visions of success.

I guess you can only feel it once in your life, and you rarely forget it. Or actually, I should have said, you were able to feel it once in your life.

Because if you base it on what metrics show, then that becomes almost unreachable with AI.

Almost everything that created that feeling is missing with AI, after all.

• There are no rankings to get to the top of (even though many platforms tell you otherwise.)
• There is far less traffic to win, too. AI systems increasingly educate users about their problems and offer solutions the way your content used to do. So, if you look at this from the numbers game point of view, there is not much traffic to gain here.
• Hell, there's even no GSC or any other dashboard as such. And I know that Google recently added AI visibility graph to GSC but let's face it; it delivers zero actionable data.

This poses another problem. Any brand needs to justify the investment in AI search, after all.

It needs to be able to "report" on progress (I've added the quotation marks on purpose.) Tell the bosses that something is happening, at least.

Yet we can't measure this the old way and report with a "one-pager with graphs..."

Feels like catch-22, doesn't it?

Well, I may have a solution. And it includes first looking at what really makes up AI visibility, and try to quantify that.

I've been thinking about with this for a while and I think I'm close. Perhaps not fully there yet. But close.

And my system is based on tracking changes and shifts across five elements that make up the AI visibility:

Brand identification

This is the most fundamental element, answering whether AI systems identify the company correctly, distinguish it from similarly named businesses, understand the product or products it offers, and connect the right facts to the right entity. And most importantly, can they do it without relying on extracting information from your website?

Positioning

This element evaluates whether AI understands what you are and where you belong.

Does AI associate you with the right category, problems, customers, capabilities and competitive set?

I like to think of it as a commercial mental model AI has constructed of the business.

Discoverability

Discoverability is all about whether you surface when the conversation starts with something other than your brand, like a problem, particular feature, use case, category or anything else that, logically, should lead to you.

Presence in buying situations

Initially, I considered this part of discoverability but I no longer think it is.

Discoverability is about whether AI thinks of you in relevant contexts. Presence in buying situations is about whether you show up, and with what information, when customers ask AI to help choose what to buy.

Does AI include you in appropriate shortlists? Compare you with the right competitors? Recommend you for the situations where you're genuinely a good fit?

This is where AI visibility starts getting much closer to commercial relevance.

Narrative

Finally, narrative covers the picture AI presents, and I analyse it from two angles:

• What AI says about you, and
• Whether what it says matches what you say about the product.

The most important thing is that all five can be measured.

Not like rankings, of course. There is no position to track or percentage improvement to report.

But you can establish a baseline for each element and track how it changes.

What's more, those shifts can be tested, recorded, and mapped over time. And then, used to track progress, spot challenges and inconsistencies, and identify opportunities.

Pretty much just like we used to do.

Just with a different data model.