Nobody needs another newsletter explaining that AI search has changed how buyers discover brands.

You’ve read it. We’ve written it. You may have written it, too.

Here’s the more useful question: Can AI actually find you when your buyer is looking?

We asked Andrew Dawson, Head of Innovation at Edelman Intelligence, how a lean marketing team can find out without building an enterprise intelligence stack.

Here's his Marketing 101 Master Class to ease your AI search anxieties.

Let’s get into it.

Andrew builds proprietary AI systems at Edelman for exactly this kind of problem.

So we assumed his answer would involve something similarly … proprietary.

We were wrong. It doesn't have to be that complicated.

“There’s a number of great platforms and tools that a marketer of 1 can tap into,” he says.

Those tools can help you understand your footprint: where your brand appears, which channels matter, which publications keep showing up, and which sources AI engines pull from when they assemble an answer.

But Andrew says the tool isn’t the system.

The system is what you do with the information. 

“It’s not ‘human in the loop.’ It’s ‘human IS the loop,’” Andrew said, quoting Gabe Michael, who leads global AI production at Edelman.

For a lean team, that comes down to 4 moves.

Play #1: Map your AI search footprint

First, figure out where you actually show up.

Run the buyer questions that matter to your business through an AI visibility tool. Don’t ask generic questions like “What are the best software companies?” 

Instead, try questions your customers would actually ask when they are close to making a decision.

Andrew says the goal is to understand what channels and publications matter.

Pay attention to:

• Which prompts surface your brand

• Which ones don't

• Who shows up instead

• What sources are being cited

• Which publications and channels keep showing up

If the same handful of sources keep appearing, you're starting to see which parts of the information ecosystem are shaping the answers in your category.

Don't start optimizing yet. Figure out what you're optimizing from first.

Play #2: Get the data out of the platform

Once you've mapped your footprint, get the data out of the platform.

Andrew's advice is simple: Export it.

The point isn't to build your own fancy dashboard. It's to stop letting someone else's dashboard decide what you can compare.

Put the data somewhere you control. A spreadsheet is enough.

At minimum, track:

Now you have something you can sort, compare, and come back to. 

You can see whether the same competitors keep beating you, whether certain sources show up repeatedly, and eventually, whether any of it is changing.

Play #3: Ask better questions

The tool found the pattern. Your job now is figuring out what the pattern means.

This is where Andrew says the "human application" comes in. Once the data is yours, you can start asking why you show up where you do, why a competitor shows up where you don't, and which sources seem to carry more weight than others.

Andrew has a rule for data in general: He says that “who, what, where, and when” only get you so far. The useful questions are usually why and how.

So instead of stopping at “Competitor X appears for this prompt,” ask why.

• What sources are supporting them? 

• Do those same sources keep showing up across other prompts? 

• What are they associated with that you aren't? 

• Where are you missing from questions your buyers are likely to ask?

You're looking for the patterns that tell you where to act. Maybe one publication keeps getting cited across your category. Maybe your competitors consistently appear for a topic you've barely covered. Maybe you're showing up, but for questions that aren't valuable to your business.

One audit can tell you where you stand today.

The problem is, "today" doesn't last very long.

Play #4: Keep checking

Here’s the annoying part: The answer you just worked so hard to understand won’t stay put.

Andrew says the citation mix can change quickly, on roughly a monthly cycle in the systems he tracks.

“It’s all about tracking, making sure that you’re checking it day over day, week over week, month over month,” he says.

For a lean team, that doesn’t mean rerunning the entire exercise every morning. It means putting enough of a cadence around it to catch what’s moving.

The important part is that you run the loop again.

Map. Export. Ask better questions. Check again.

Because an AI visibility audit isn’t useful if it tells you where you stood 6 months ago.

What does ‘better’ look like?

More citations would be nice. But 100 citations for questions your buyers never ask aren't useful.

The goal is to show up more often in the answers that matter to your business.

That might mean appearing for more high-intent buyer questions. Closing the gap with a competitor that keeps making the shortlist. Getting cited by sources that consistently influence answers in your category. Or simply seeing fewer important prompts where you're nowhere to be found.

The exact scorecard will look different for every company. The most important thing is that you have one.

—

When someone asks ChatGPT for a platform like yours, do you come up? We'll check. And hand you a 90-day content plan to make sure the answer is yes.

Book time with Abby + Shaye to learn about storyarb's 30-day, AEO-informed strategy sprint!

—

Campaigns that got us talking: The product insight is that Genesis’ yet-to-be-revealed SUV has a unique way for passengers to get in and out. The output is a 17-minute Maggie Gyllenhaal film about Marilyn Monroe, starring Dakota Johnson and Ellen Burstyn and premiering at the Venice Film Festival. The public won’t see it until the entire festival run is over. We would pay actual money to read that creative brief.

AI spotlight: 47% of marketers say senior leadership is the least AI-savvy group in the org. Which tracks with the number of AI mandates being issued by people who have never opened the tool. Hard to set a policy on output quality you’ve never personally evaluated.

Stuff that makes us scroll back up: Speaking of “human in the loop,” we loved this skit from Varun Rana on tech companies rediscovering the need for human labor. Hits a little close to home 😅

A politician went viral for reading a speech that included AI instructions. 

"Here's a more natural flowing version of that section.” And the speech went on.

Yikes. Maybe next time, proofread what the robots are suggesting. Better yet, hire a human editor?

Remember when you could Google yourself and call it competitive research?

AI search made knowing where you stand a little more complicated. But Andrew's point is that it didn't make it impossible.

You don't need to know everything the algorithm knows. You just need a system for noticing when the answer changes.

See y’all next time. 

— the storyarb writers’ room 🫡

P.S. Once you've got a baseline and want to address the other half of the problem — actually influencing what the models cite — our AEO guide covers it.

We help brands reach thousands of B2B marketing leaders, and we turn great experts into editorial features worth their inbox space.