Instinct has taken over my life the last couple of months. Car shopping, groceries, books, the random stuff I used to hand to an EA. I keep giving it more because it keeps getting shit done, and does so seriously well.

Now I use it to run a big chunk of the offsite AEO work at my multimillion-dollar agency.

I didn’t plan this. I used it for something, it worked, and I gave it something harder. Eventually I was handing it work I’d been hiring contractors to do.

Here’s a practical use case you can steal.

The best AEO comes down to what happens off your website. Getting your brand into the articles, comparisons, and industry roundups AI pulls from when someone asks about your category.

It’s also an insane amount of work. By hand, each target goes through these steps:

Multiply that by a list of sites and several clients, and somebody has to keep all those conversations moving.

A few months ago, I was assigning that work to contractors. With Instinct, the steps are:

All of it runs through iMessage, which is a ridiculous place to find out you can run this much of a business.

The useful part starts before I text anything.

I use tools like Profound to find the pages that keep getting cited in answers that matter to a client. Then I look for the ones where that client is missing.

I want the actual page and the actual buying question. “We should get into this publication” is too vague to act on. A comparison that keeps showing up for a use case the client serves, names the relevant competitors, and leaves the client out gives me something specific to chase.

That’s a much better reason to pursue a placement than a domain authority score on its own. I can explain why we care about the page and what we hope inclusion will change.

I still need to read it. I check whether the client belongs in the comparison, whether the article describes the products accurately, and whether the client would want a prospective customer finding them there. A citation report can point to candidates. I make the buying call.

Not every gap needs money thrown at it. An article might need an update. The author might be open to evaluating another product. Or the publisher might offer paid inclusion. Those are different conversations, and outreach tells you which one is on the table.

Once the targets make sense, the brief to Instinct is simple:

That number is only an example. Set yours based on the opportunity and what you’re willing to spend to test it.

If I send an agent a vague instruction like “get us placements,” I shouldn’t be surprised when it comes back with something I don’t want.

Say the target is an existing comparison that keeps showing up in relevant answers. The publisher replies with an offer for a new sponsored article somewhere else on the site. That might be a fair offer. It’s still a different purchase. I wanted inclusion on one specific page for one specific reason.

Settle that distinction before anyone spends a dollar.

Same goes for negotiation. My ceiling is the most I’m willing to consider paying. Keep it out of the opening email. Ask for the rate, pin down what it includes, and negotiate from there. A lower price means little if the placement is temporary, takes a pile of extra work, or lands somewhere other than the page we wanted. I want the terms side by side so I can decide.

Instinct handles the back-and-forth. I pick the targets and approve the purchase.

That’s the split I’m comfortable with…Instinct has enough direction to keep the conversation moving without checking with me on each reply. I keep the decisions that affect the client and the budget.

What surprised me is how much the outreach itself decides which targets are worth chasing. The fee is only part of the cost. The rest is time spent finding a contact, chasing them, clarifying the offer, and negotiating. Sometimes there’s copy to write, and then approval.

A small placement can eat real time before you’ve even confirmed it’s available. And when the answer finally comes back as “too expensive” or “wrong placement,” you’ve still paid for all that effort.

Cut the manual work and more of these targets are worth a look. That lets me evaluate an offer without spending the same human hours on every conversation.

With less human time per target, I can afford to be curious about more of what the research turns up. I still approve every purchase.

A placement doesn’t guarantee the client shows up in the answer. The page can keep getting cited while the answer recommends somebody else.

After a placement, I check two things: whether the agreed inclusion went live correctly, and whether the client’s presence in the relevant answers changed.

The first is easy to confirm but the second takes repeated checks.

I’d define the questions before the placement goes live and check them repeatedly, on both sides of the change. I care about the share of answers where the client appears, how it’s described, and whether it’s actually being recommended for the use case we targeted.

One screenshot won’t settle that. Even a sustained improvement needs scrutiny. Other content is changing, competitors are working, and the engines keep moving. Questions the placement shouldn’t affect make a useful comparison. If everything improves at once, be careful about crediting the placement with all of it.

Also record the placement type, the cost, and the live date. Otherwise, six weeks later, we’re trying to remember what we bought and convincing ourselves whatever happened next was because of it.

Keep the reporting light, but clear enough to inform the next decision. I need to know whether to buy another one at the same price, try something different, or wait for more evidence.

I can have a strong opinion on the final offer without sending the third follow-up myself. I can stay accountable for the spend without running every pricing call.

Instinct lets me hand off that work with specific instructions and get a decision back.

I started with personal errands. Now it’s doing agency work I was recently hiring people to do. It happened one assignment at a time, as I got comfortable handing over more.

If I were trying this, I’d pick one client category and a small set of pages I could justify. I’d give the agent a clear brief, a spending ceiling, and an approval step. Then I’d review what comes back.

I’d learn more from working through real offers than from another afternoon staring at a citation dashboard.

P.S. Yes, my phone is dying and I have 262 unread texts.