The first move to show up in AI answers (it's not writing content)

I've been working closely with companies recently to help them with their SEO and AEO (not just a list of vowels, but also means helping products and businesses appear in AI results). So I figured this week, I'd try something different and walk you through the first thing you need to do to show up in AI summaries.

Also, before we go any further, some context. I last bought cooking pans 5 years ago when I first moved to New York. They've treated me well, but they are your standard non-stick pans and like...maybe…toxic? So I'm in the market for new pans. Stainless steel pans intrigue me and make me feel as if I'm a real chef, like Remy from Ratatouille, but they scare me.

So naturally, I asked ChatGPT. It gave me a list of brands, along with explanations for why they are the perfect pans for me. Which is funny, because that list of brands the AI decides to show is the thing I help companies fight their way onto. So let's talk about how you get on it.

TL;DR

Every founder wants AI to recommend their product, but most of them react the way they did in the early SEO days. They panic and start writing. Blog posts, FAQ pages, a glossary nobody will ever read. It feels productive, and because there's no real scoreboard, you get to believe it's working.

Unfortunately, it usually isn't. You're writing in the dark, guessing what to say before you know what anyone's actually asking. Instead, you should flip the order. Before you write a single word, make a list of the real questions your customers ask AI, then run those questions and see who the AI recommends back.

That list is your prompt set, and it's the North Star for everything that follows. Once you can see the exact questions where you're missing, and exactly who's beating you there, writing content becomes much easier.

Kevin Indig, who's run growth for Reddit, Ramp, and Shopify, puts it the same way: start with the prompts, then figure out whether you can write a better answer than whatever's winning today.

So let's build that list, then run it.

The three kinds of questions

Problem. They've got the itch and no idea who you are. "Best non-toxic cookware." "Why does my pan keep sticking?" Start here, because this is where a total stranger meets you for the first time, and it's the hardest slot to win. Your name isn't in the question, so the AI has zero reason to bring you up unless you've earned it.

Comparison. They're picking between a few options. "Always Pan vs Caraway." Put plenty of these in to see how AI talks about you versus your competitors.

Objection. The last flicker of doubt before they pay. "Does the Always Pan coating wear off?" “How long will these pans last?” I personally know that your product is perfect, but others might not, so it’s important to include them.

Also, branded (your name is in the question) and unbranded (it isn't). Track both, but the unbranded ones tell you more. Anyone can show up for "is [my brand] worth it," because the person already knows you exist. Showing up for "best [category]," where nobody typed your name, is the best test. That's where new customers come from.

Two more things that sharpen the list:

• Go specific, not broad. "Best CRM for real estate agents" is winnable. "Best CRM" is a war you will not win. The narrower the question, the better your odds.

• Skip the pure definitions. "What is ceramic cookware?" gets answered straight from the AI's training data, so it doesn’t need to do any web research. These are usually much harder to show up in.

Use your customer's words, not your marketing

To make this real, I ran the whole thing on Our Place, the company behind the Always Pan (the 10-in-1 pan, just like the 10-in-1 shampoo you have). Big brand, great-looking product, millennial branding, buckets of money. Exactly the kind of company you'd assume AI adores. We'll see.

When you write these questions, you'll be tempted to use your own marketing language. Don't, because your customers don't talk the way your marketing does.

Our Place sells "one pan for your tiny kitchen." So my first draft of questions had things like "best cookware for a small apartment." Sounds reasonable, but it’s not a common search. Type "best cookware for" into Google, and it fills in stove types, induction, gas, glass top, and never kitchen size. The words a brand uses about itself, and the words its customers use, are seldom the same.

After digging into other people looking for pans, I found they were asking if the coating wears off, if it's really non-toxic, and why people are obsessed with “seasoned” cast-iron pans.

A few places where you find the language that your customers use:

• Google autocomplete.

• People Also Ask. That dropdown of questions on any Google results page.

• Reddit.

• Your reviews. Check sites like Trustpilot and Amazon.

My Test Subject: Our Place

I built a small set of eight questions and ran each through ChatGPT, Perplexity, and Google's AI answers. Eight is a demo size, just enough to show you the shape. For your own brand, you'd want more.

Here's the set:

On the three unbranded questions, the ones where the user has just started research, Our Place was the top recommendation zero times, on all three engines. GreenPan won "best" and "most durable" everywhere. Caraway won "prettiest."

On one comparison question, Perplexity filed it under "what to avoid." On another, ChatGPT said, word for word, "I'd probably skip the Always Pan." The robot, unprompted, talks a shopper out of the sale (your biggest opp might be AI).

A brand called Alva, which I’ve never heard of, beat Our Place for a small-kitchen recommendation on Perplexity. How does a small fry beat a giant? Alva published one blog post titled "Best Cookware for a First Apartment." It answered the exact question, in plain words, on its own site. The AI found it and rewarded it. Our Place never wrote that page and missed out on a huge opportunity.

The Sauce

This is the part you can do today. Three prompts, back to back. Before you can control the narrative of your product, you need to understand what the current narrative is, and this will help you.

Step 1: Get the language. Before you write anything, find out how your customers actually talk, in their own words, not your marketing's. The first prompt sends an AI to dig through Reddit, reviews, and the questions people really search, then hands you back a plain report of their exact words and worries.

Step 2: Turn the language into questions. The second prompt takes that report and writes your test set: fifteen real buyer questions across problem, comparison, and objection, weighted toward the comparisons where brands actually get named.

Step 3: Run the scorecard. Take those questions through ChatGPT, Perplexity, and Google, in an incognito window so your history doesn't skew things. For each answer, note whether you're the top pick, who beat you, and what got cited. The third prompt runs the whole sweep for you if you've got an AI tool that browses the web.

→ All three prompts, ready to copy: The Prompt Pack

This is just the first step to showing up in AI summaries, but it’s one you shouldn’t skip. It's easier than ever to build products, but harder than ever to stand out in the noise. And for making it this far, I thank you and gift you a meme below.

Reply to this email and let me know what you thought of it!

— Amaraj (aka my eggs always stick to my “non-stick” pans)

The Meme