Financial Brand - Presentation
===

Wil Reynolds: [00:00:00] Howdy. There are friends Will Reynolds here and ah, I imagine that if you are at the financial brand forum and you're sitting in all these sessions talking about AI, life feels a little bit like a big game of whack-a-mole, right? All the things are changing. Every week a new model comes out. Last week it was gamma four, uh, this week it is whatever Claude's new model is, they're not releasing to everybody.

And my job today is to bring some sense to Whackable. Okay? And I think the first way I can help make sense of all the things you're hearing is to give you a framework to think through all the things that you hear. Because what we know is from the invention of time, marketing has required you to do three things.

First you wanna be visible or known or seen. And then you take all that visibility, all that being seen, and you try to push customers into believing that when you, when they have a problem, you might be a good solution for it. [00:01:00] And then after believing you about that, only then can they choose you. And that's how we get new account signups and new members in our credit unions.

Right.

So one of the great debates around generative engine optimization is this concept of are you gonna try to do it fast? Which so often speed ends up hurting, trust, and ultimately think about all the decisions you've made, all the vendors that you work with. Did you really pick the person that just got back to you fastest, or did you pick the person that you felt cared and that you felt that you could trust them to try to help you hit these goals?

And I think the best way to start is not by talking about banking, because I like to go back to the 19 hundreds. Michelin literally drove all over France to look at a bunch of restaurants and say, these are the best ones in the French countryside. And then they made a book and then they went all around and gave it to chauffeurs so that on the hope [00:02:00] that if you built something that was quality and then these drivers went all around.

That they would wear down their tires and they would remember the brand Michelin. And today a lot of us here in Vegas are definitely looking for like, what's the Michelin Star restaurant? And can they fit me in? So it was built in 1900 and they gave them out. So they were seen. But the fact that today, Michelin star restaurants have become a filter for quality.

Means that not only do people believe them, but then when you look at how hard it is to get into those places, people are definitely choosing the restaurants. Now, I bet you, I think the bet that they're making is that somehow this thing is worth it as an asset to get all that visibility on their brand name.

And we all know that stuff happens back here in human brains. It isn't always what you think it is. Um, so another company, this one's a B2B finance company, was inspired by this and you know, they're [00:03:00] in the AI age and they said

We're inspired by this and even though we're in B2B finance, we can also do something like they did. Let's review a bunch of restaurants, and what they did is this, they created content like the top five restaurants in Minnesota for business dining.

Do you guys realize that this search is on Minnesota business restaurants. The title is Top Restaurants in Minnesota. Let's just remind you folks, Minnesota's freaking massive and it can take you up to seven hours to drive from one end of Minnesota. To the other, so no humans are gonna search that way.

This is what happens when we turn our brains off to AI and we just go after everything at scale. We produce zombie content, we produce the kind of content that you just repeat the model over and over again. You flood the index with slop, you do a little bit of customization, but ultimately, we all see this stuff coming a mile away.

It's 100% formulaic to me. It's a [00:04:00] hundred percent slop and it's a hundred percent ignorable. Nobody wants that content. So it's interesting how when RAMP had the opportunity to drive around and taste food and give people real recommendations that actually mean something to them. They said, we're going to use AirOps and build 50 pages automatically.

And what's really sad, I think about that, is ramped up so much high quality content that in this one area of their business, they just loaded up a bunch of pages that really don't connect with how humans operate. And I'll tell you, I don't think any of us tonight are gonna try to find out or look up a ramp rated restaurant.

Are we? Alright? So I know a lot of us are smaller brands competing with bigger brands and for so often, if another company had bigger dollars than we did, getting out spent used to mean that sometimes we would just get out marketed or there were certain places we just couldn't compete. [00:05:00] But in an AI world.

I don't think that applies anymore. You know the other thing I see, and this is my data, you can see that Seer is posting less and less content over time, but yet when I create a bar for content, humans love direct traffic, social traffic, all the different parts that tell me that people, humans loved this.

Traffic from Microsoft Teams. All the things that I can track still, we are doing the best work we've ever done. Getting the pages that we do right to actually perform for our business. People love them, but we had to choose to produce a lot less content and not more. And our pipeline is growing. Okay? So for many of us, the saddest part about working in a smaller organization is you're like, if they're bigger, they're gonna win.

If you're Banana Republic, you're gonna beat Nudie jeans. 'cause everybody in here has heard of Banana Republic. How many of you in you have heard of Nudie jeans? [00:06:00] But let me show you a little bit of more of the story. So one of these two brands ranks well for the word ethical genes. Now Nudie

they have a free repair station on all their genes because they don't want you to buy new jeans 'cause they know how it affects the environment. But then Republic has nothing like that. Nudie jeans will fly all over the world to their partners and they prove it on their website to say, we made sure that they are.

Using like the right kind of employment practices with the people that do it. If you buy from Banana, you probably don't want to know the employment practices of how people are treated that make your freaking cute little pencil skirt.

So given that, which one do you think ranked number five? And which one do you think ranked number 55 for the word ethical jeans? Well, you'd be surprised, my friends Banana Republic, because they're the bigger brand, won that search over Nudie jeans, and all they had to do was put a block of AI generated content at the [00:07:00] bottom of the page to win.

That's a heck of a lot cheaper than getting on planes to make sure that the people who make your clothes are actually being treated fairly and they beat them for that. Right Now in an AI world, when you go to ChatGPT. And you type in ethical genes companies or ethical genes or ethical brands, guess which brand has 0% visibility and which one's got 90% visibility?

Yeah, my friends, they do a great job. Now, something that you've gotta start to realize, and I want to train all of you to see this when you do prompting. Is, do you see when there's little links to the, to the, to the right of the answer? What that means is that that search triggered a web search.

So whatever you typed in, in your prompt, it triggered the AI made a decision to say, Ooh, I should probably search the web, goes out and brings the data back and gives us a bunch of brands there. But do you see where there's no link next to it? That part of the answer is coming from [00:08:00] training data, which now means it's in the actual way that the AI has been trained.

It would give you the answers, nudie, genes, Everlane and outland denim, which is interesting 'cause if they ever chose not to use the web to answer this question, if you're nudie, you're gonna show up on the web and you're also gonna show up in the training data. Alright, so I think for a lot of small bank marketers, this is your moment.

Your arms are sweaty, knees, weak mom, spaghetti because you've always put community at the center. And let's be honest, a lot of your bigger competitors had a bigger budget but not a bigger heart. They saw community as pretty much a commodity. You're in your communities and I think that's an advantage.

And I think you also don't have to deal with all the bureaucratic crap that the other marketers are gonna have to deal with. When I start showing you how to win in these areas, you see marketing built around channels instead of customer [00:09:00] problems. It's kind of dying. You know, channels will always change.

TikTok didn't exist. You know, AI didn't exist. Algorithms change the problems that people have, don't, and the way that we choose to trust brands or not trust brands, it's pretty much stayed the same. This is another company's, uh, approach on ai, and I don't agree with their timetables at all. I think they made it way too simple.

But I think there's something to be said for the four boxes. Where are you not visible? And let's make sure that you're detectable. And then where are you detectable, but not necessarily citable yet. Let's make sure that we get those citations increasing. And then how do we become a default answer? A default answer is an example to me of when you are part of the training data where it now knows to associate your brand with certain things.

And I think that's the path to AI dominance. And I will tell you that AI is broadly better [00:10:00] than Google search at rewarding smaller players like us. For everything that I've seen, there are pockets where we can win, and I think it's gonna matter more than ever that we win in those pockets because 53% of customers.

Distrust, AI powered search results. So I think we're gonna talk about this a little bit later, but there's gonna be areas where you can win because you've always been connected to community. So if you're ready to go, let's rock and roll. It's gonna get a little technical. So when you do a prompt on ChatGPT or whatever, many times when they go out to the web, they are doing searches.

Isn't it interesting that for these queries that were done, these prompts, I think it's ChatGPT said, restrict. The answer to these sites, you can see it site, colon, greenhouse, ash by hq, workable. So what this tells me is somewhere in ChatGPT five four thinking model when a prompt around the best applicant tracking systems [00:11:00] was entered.

They said, we know the brands, so let's go search those brands specifically so that we don't fall for all the low quality, low content that people are putting out there that's telling. The more you become a brand, at least in ChatGPT five, four thinking, and I think this is an indication of where they're gonna go, the more likely you'll show up in prompts.

But ooh, that's brand building. And for the first time, one of the major players has actually released some information on what people are searching for on ai. Now, these probably all look like keywords. What they really are is the little queries that that copilot is running. When people put in their prompts, I don't get to see the person's prompt.

I get to see the refinements of the searches that were done. A lot of brand in there, right? A lot of brand. So now what I can start to do is look at something like geo testing methodology, and at [00:12:00] the bottom of Bing's webmaster tools, you can see the pages that showed up for that. And you can see our case study page showed up.

Now I'm like not happy about that because what that means is, is that for our geo methodology, it's most often citing my case study page, which is not the right page. So now all of you can share this with your generative engine optimization team or the person doing SEO for you and tell them, go into Bing.

We now have bing webmaster tools. Ability to get some of this information and ask yourself if Bing runs all these prompts as part of the searches to answer the prompt rather, and it brings back pages that are not really the right page that we want. We should probably try to do something about that.

Alright, so one of the most fun things that we do is we have a UX team that does a lot of research watching real humans at the keyboard use AI to try to solve their problems. [00:13:00] And this is, I'm gonna start sharing some of that with you. So we looked at overdraft around different businesses, not just Wells Fargo.

We looked at overdraft and we said, okay, if I were to ask chat GBT, what are the overdraft fees for our Wells Fargo account? Do you see how often Wells Fargo is there? Seven out of nine of the answers are Wells Fargo, which means they own the content that drives the answer, and we've got examples. Of ChatGPT literally updating content within 36 hours of, of us changing it.

And because Wells Fargo owns Wells Fargo, they can make changes to make sure they show up appropriately. And I did some diving and you're gonna get, just look at this slide. It looks like just a content slide. But remember, look at the right side. We're gonna talk about that in a little bit Bank of America has a page about overdrafts and draft protection, and as you can see here, [00:14:00] go banking rates, overdraft apps. They're driving all the answers, which means for Bank of America at this time, for this prompt. I'm not saying every time, it did not bring in any of their content to answer the question.

And when you see me do the search or when you see me get to their site, this is their page. Look at the box at the top. This is this page for overdraft and overdraft protection, but you don't see any of the content. We're gonna talk about that in a second. The reason why you don't see the content is something we can all find right now.

If you hit F 12 on your keyboard and then you hold control shift and P and start typing in disabled JavaScript, you will have a box pop up. That enables you to then see what does the search engine or the, um, AI bot see when JavaScript is disabled, because they do not always index JavaScript. Well, that's all you gotta do.

That's how I found that Bank of [00:15:00] America's page does not show up with any of their resources. So now we're talking about a battle, right? Because if you're Wells Fargo, good for you if you're them. O so. Let's do a, let's do a Dallas versus Eagles kind of fight here. So at the top we've got a rate table. Our research shows us rate tables is what most humans do when they go to chat.

GPT. Second most popular, uh, thing that they do. So we've got Philly at the top, Texas at the bottom, both look like normal tables to all of us, right? There's nothing different about either one of them, but this is what the tables actually look like for Phil, for one of the. Pennsylvania credit unions. If you go to their site and you click on check the, and you go to the page that has all the rates, you gotta click the little plus button to expand it.

Unfortunately, it doesn't show up at all. It it, it doesn't show up in any of the search engines. It doesn't show up [00:16:00] appropriately in entity AI answers because it's in JavaScript and it's not indexable. Whereas Dallas just has the box. The Texas company, the Texas Bank just has the box, no pluses, none of that expansion stuff.

And they are always showing the right interest rates for them. Now, if you're thinking, yay, Dallas one, I'm gonna remind you. Little Super Bowl thing. Yeah, go Birds. But it's different guys like, so you have to start to hit that con, that F 12 hit control shift P type in disabled JavaScript, and see if your rate tables are no longer visible.

Alright. And also you have to, you have to refresh the page after you do that. Alright, so, um, I'm gonna show you something that's a, that, that really took me by surprise when I first read about it. And most of us in here, our AI metrics are around visibility and traffic. Never have I seen a bank marketer say, I, I, I monitor AI to help protect us from fraud.

So what we found is that, um, there are literal fraudsters [00:17:00] who will register phone numbers, uh, that show up in chat, GPT and other AI models for bank brands. And so what they do is they run a bunch of prompts against them. They see what phone numbers show up and if they're ever available. They acquire those phone numbers because they know that some people will type that in to chat GPT and call them and they can scam them 'cause they think they're calling Wells Fargo or Bank of America or your bank.

So 36% of the time, the phone number's not right in AI models and financial services was the area where we saw this had the biggest impact. And what's really crazy is, um, as a lot of the customer service aggregation sites get human, et cetera, and GetHuman, I was like, oh my God, they haven't posted a new post on, on their press section in four years, guys, four years.

So that's how outdated some of their stuff is, and they're responsible for a [00:18:00] lot of people in this room's phone numbers showing up to start asking, what is the phone number for in your brand? And run that every month. C or run it every week or run it a hundred times at once and see if the number always matches up.

Because what we found is that only 41% of the time when you ask about a phone number, does the brand actually own the answer. Oftentimes it's third party sources that you do not control, and I bet you that doesn't get included in your ROI. Now, I was doing some other. Research, this one is super new, super rough.

Um, but I was looking for credit union domains that were able to win on best in compare queries. And I'm just gonna tell you that the America Teachers, credit Union, Alliant, credit Union, Navy, federal, and Pentagon are a few. You might want to look at if you're like trying to model, uh, after a, uh, a credit union that's been able to beat some of the big banks on some of the terms.

But I think we gotta get rooted [00:19:00] again in like, you wanna show up in ai. There's a lot of tricks that are short term and there's a lot of small things you can do that are easy to do that aren't tricks. But you really have to start to think about AI as a brand channel so that they think about you in the future.

And I, I, I, I got the data to prove why that's important. I think for so long, we as marketers have spent years compromising on our designs and having five panels and whatnot so that everybody's offer can be seen. And now that's a problem because all those features sometimes have been in implemented with things that are not indexable to help your rates and help people to find you easily.

So it's hallucinating your rates or it's getting your old rates from somebody else's website and we don't want that. Alright, so let's get a little bit more human here. I think the first thing I'm gonna remind you all of or tell you all is most of us out there that are doing AEO or GEO are tracking the wrong prompts.

We have specifically done the research in banking and we've seen people have anywhere from four words in their prompt to [00:20:00] 119 for the same task. This is what happens when you actually watch people do it. It's like, oh my God, we forgot the humans. Um, and here was the resource I was mentioning earlier. You can see that people, what people are likely to do.

We talk to real humans, compare banks and compare products and find rates and fees. You know what? They don't do a lot of apply for the loan or open an account. It's just one of those things where people do not feel yet like, oh, I'll just let an agent open up my account. No, I'm not doing that. But I will use it to compare banks, compare products, and find your rates and fees.

We interviewed 30 real people. We evaluated 114 different prompts. It only cost us $800 using a tool called outset.ai to manage the watching of people as they did these tasks. I think last year I might have presented this part, but like, you know, once we started realizing that people are putting brands in their prompts, it's like, wait a second, how do they know the brands to [00:21:00] put in their prompts?

Because they're asking their friends. They've seen your brand in places they've heard about you somewhere, and that is not what that, that is like how people know to put the brands in. So you gotta be kind of careful of assuming that just showing up for some generic query, like checking account is worth it.

Here's a bunch of great starters from my friend Harpreet, start typing these things into chat, GPT for your own brand and your competitor's brand. Because what I've realized is people are putting so many brands in that you're gonna have to be a brand people have known and heard of, or that your friends pass on to other people.

You know, the old way is this, you type in checking account. None of it's personalized to what's happening in my life. Whereas AI is completely personalized to what's happening in my life. I gotta click on all these websites and review them all, and it's just a bad experience. You know, I clicked on sun east.org, went to this page after typing in checking account, and I see why this guy on that side is like, ah, you know why he is like that?

[00:22:00] Because there's nothing on this page about checking. That's what the guy over there is, is, is holding his head arms up for, and the, the old days of like it being all NerdWallet and those kind of sites are kind of over. Um, you got Reddit there now, which obviously we'll talk about in a little bit. But then up here, if I were to search for Sunny East versus PFCU, if I was comparing these two, now we got an AI overview, right?

And now we've got Reddit there with a, ah, it's 14 years old. Ah, but. This is what this result is starting to look like. And I will tell you, people do this a lot. Now, how many of your prompts are you tracking like that? Lucky for you. We built a tool. It is open source. You can download it and install it and play with it yourself.

But I can put in some attributes about being a CEO of a company of a certain size. I can then say, Hey, here are the concepts that that matter most to this persona. And then boom, you get a spider [00:23:00] chart that shows how your bank performs against another bank someone might compare you to with the actual answers from each one of the different LLMs if you so choose.

So now you can see things like. Hey, Navy Federal Credit Union. This answer says they have limited commercial treasury management and syndication capabilities. Well, that came from somewhere. If you're Navy Federal and you're like, that's not true. We just launched that last month. I'm like, well, we need to track that.

We need to make sure that we're mentioning that in the right places so that ultimately it shows up over and over again so that three or four, or five or six months from now, it will, uh, have changed.

One of my biggest concerns about people who buy generative engine optimization services is that the geo math isn't SEO math, it's a different thing.

You can't rely on affiliates anymore. I mean, look at some of these major publications on how much traffic they're losing, and this is an old post. I can only imagine it's worse. And I think for too long we've all tracked the easily trackable [00:24:00] KPIs, and I think that it's the smaller banks that have realized, like sometimes it's the person whose hand just shake.

Sometimes it's the person who just showed up at a tough time in their life because they're your neighbor that ends up getting your brand mentioned in a place that you can't even track anymore, and you can clutch your pearls all you want. SEO is a performance channel. GEO, in my opinion, is not. Back in the day, you would get a keyword with a monthly search volume, and then if you can get ranked, you would get a certain amount of visits and you could, you could project these things out in geo.

We don't know how the personalization works. We don't get search volumes. We don't get the click through rates anymore. I mean, it's so bad that when I said, how do I open a checking account? You can see here there's no links, so there's no clicks to be had. But on some prompts there are different sites that are showing up.

What are the downsides of opening a joint checking account? Some answers are coming from websites, some are coming from training data, so some people can get clicks and some people can't get clicks, right? Good job chase. If you're in the house. [00:25:00] Um, mostly you don't. Most of your other competitors don't show up for stuff like this, but these things matter.

And then you have to remember that AI memory these days builds a whole knowledge graph around you. And when they know your life stage, your income, your past questions, when they know all that they are. Totally optimizing those recommendations. So, you know, best credit card means something different for someone who's planning a trip to Japan versus somebody who's renovating their home.

And those are great examples from my buddy here from Yext. So what I would tell all of you to do is go to ChatGPT or Claude or whatever you use and drop this prompt in and see how much it knows about you. For me, I was surprised and it was almost all true. Alright, the next thing about the math of generative engine optimization is.

We, we don't always, you always have to know when somebody asks you tough questions. You gotta know the launch date against the training data date. So, you know, Gemini [00:26:00] 3.1 was launched on March, 2026. So you could have somebody say, oh, I heard Gemini launched a new model. Uh, how do we show up? And you need to also know that the training data for that model.

The training that I cut off when it stopped training was January, 2025. That means we are now, 15 months of time has expired, where the training data side has nothing but 15 month old data in it. So you can't always expect the work that a Geo team is doing to show up instantly for anything that's being answered from training data.

All right. Another thing that we noticed when we were doing some research and specific in the finance space, the average words per answer over doubled on ChatGPT. So for a lot of you, if you said, oh my God, my generative engine optimization's working so great, uh, in December or January, it's not working any better, they just changed more words per answer, so therefore more brands showed up [00:27:00] alright, the next thing is. How they're changing the layouts of these pages. Literally, guys up at the, I did this search, I did this same search three different times on the same day. One time when I did it, all the links at the top, LinkedIn blue, you click on 'em, takes you to the website.

All the links down here under, with the little dotted lines under them, under credit unions. Um, in Texas. Those all, when you click on them, they don't take you to websites, they take you to Google searches, which is how Google's growing their search traffic. Then I did it again. Some were underlined in blue, some were underlined with dots.

The ones in blue took me to the website. The ones with the dots took me to Google searches, and then I did it again and nothing was highlighted up top all in the same day. This is gonna wreak havoc on you trying to attribute that traffic. So I know a lot of, you're probably a bit overwhelmed and you're like, alright, well help me figure out how to get help.

Right? I think a lot of people are saying that GEO and SEO are the same. I think my Nudie Jeans [00:28:00] example shows you that it's not, and I think it's lazy to say that people owe you an explanation on what is actually different between GEO and SEO. And if they're not willing to give you that, it's tell that they don't really know what they're talking about.

So if you've been struggling a bit, you're like, oh my God, how Well help me. I'm gonna give you some ways to get help. You know how like you got this guy, this Chris dude on the ketchup predator. I was like, hmm. I wanna catch a shady SEO. I'm sick of you guys talking to people and not being equipped with ways of making sure that you don't get ripped the freak off.

So let's show you how that happens. First, make them explain to you how visibility leads to you being believed and chosen. 'cause your. Numbers are about assets under management. How many new accounts did we open? How many funded accounts did we open? How many accounts got passed fraud after six months?

You care about that stuff. A lot of these AI optimization people are just focused on your AI visibility, and that's not [00:29:00] enough because visibility. Means like, oh, well, because I live in Philly and I'm always at the stadiums, you would think that my banks would be either Wells Fargo citizens or Lincoln Financial for insurance.

But the reality is none of them have my business. Now, I'm not saying that naming a stadium isn't worth it, but what I'm saying is, is that is an example of how just being visible over all this time does not push people into believing or choosing your brand all the time. And Lincoln Financial watch what they did here.

See their description. They provide resources and solutions that help empower Americans to take charge of their financial lives. That could be anyone writing that, that is going to be a problem. I don't have time to get into all the research behind that, but you really wanna make sure that you're, um, using stuff that's unique about your business and your descriptions and whatnot.

Now, what you can see here is a lot of these companies that rank for best agencies for generative engine optimization, which I would love to show up for, but you know what? They all wrote. Listicles, like top this, top that, whatever. And if you look at the answers, it's [00:30:00] all First page, Sage. First page. Sage.

First page. Sage. You know why? 'cause they build a bunch of low quality slop and the AI has not figured out how to get it outta their index. So. You run this prompt on any agency you're talking to, you will see how much of their content has best in top in it. 'cause they're gonna try to do that crap with you.

We're the best credit union and here's the other eight ones and uh, we're number one, that's not even marketing, right? So you start asking them, Hey, I noticed some of these. Can you open up your Google Analytics and show me how it's performing and watch 'em choke. Right. And then I'd say, show me your pipeline and show me it's actually working.

I see your visibility went up. Did your pipeline go up at the same time? Oh, well, uh, it's like, so I'm not in the job, I'm not in the business of just being visible. I need people to also choose my damn business. Right, right. Alright, so make sure you don't fall into the build trap of just judging success by visibility and how many pages we built and how little time we built.

Also be careful that just showing up isn't the whole job. Clickup [00:31:00] left the instructions from ChatGPT in their freaking posts. Do you trust them more or less? I think we all know the answer to that. So I'm asking all you to take the same stand that I've taken, which is I'm slow playing AI visibility for my own business.

And you might think that sucks for Seer, but it's not true because ultimately our pipeline is growing. It's because of this. In a world where everybody can now produce infinite amounts of content, being a trusted advisor in your community is gonna matter more than ever. That's why I wanna win in these places where humans share things with other humans, and this is what it looks like when you're believed.

My number of contacts is growing. My direct traffic is growing. My referral traffic from sites that trust us enough to send traffic from their sites to ours is growing. Major brands in our space are starting to send us so much more traffic than they used to send us. That's why I feel strongly that we are winning the right battles.

[00:32:00] Humans, the problem with trust channels is that while they're more efficient where humans share with other humans, they're also the hardest ones to track. You see direct social and referral all outperform pretty darn significantly search so you get more bang for your buck when you win in those places.

But they're not easy to win on because you gotta do real high quality stuff and they're definitely the hardest ones to track. So start making sure your good news is spread everywhere. The things unique about your business is spread everywhere. Alright, let's go over a couple things. You can do one interview your SMEs.

Every company has, especially in banking, people have been there for a long time. They've been in the community for a long time. So don't just write ai slap. Take these people that have been tenured in your company for a long time, and if you're the marketer whose job it is to get information out of them, just interview them.

Put a recorder in front of them and ask 'em four or five questions about whether or not they think you should have a joint [00:33:00] checking account. And if they can't do that, they can't say the account's best or not, or whatever. I know some of you can't. Interview two different people with two different points of view and say, our bank is not saying one of these people is right or wrong, but there are two different perspectives that they're giving you.

And now you get all that rich content from real people that deal with other, deal with humans on the other side, not just tools and algorithms, I want you to review your prompts. Tell me about in my business, or tell me weaknesses about my business and just run them manually once a week or once a month and just start to train your gut to know like what they typically like and don't like.

Start taking some of your prompts that you've been tracking. Run them against your sales calls or your paid search keywords. Is the prompts that you're tracking representative of the kinds of things that people are saying on sales calls in those transcripts or in those paid search terms that you're paying for?

That's a better way to get your prompts. And you gotta review your reviews because AI's gonna say something about you, so you might as well make sure it's the right stuff. As my friend Amanda said, [00:34:00] you know, one of the saddest things is that I was doing an analysis for a, for a, for a doctor, and, um, there's all these great reviews about her.

And when I hit F 12 and looked at the JavaScript, I found out that none of them were indexable, all that great content. Never saw the LLMs at all. Everybody wants me to talk a bit about Reddit, so let's do it. I'm gonna go back to this example 'cause I think it's best. When Clickup built this page, one of the things they never really did was check to see what people think on Reddit.

So they built content saying they were the number one 'cause they built a listicle. But guess what? When I took all the different Reddit threads that ranked. No one ever mentioned them as a good solution for this problem, but that didn't stop them from building a page saying, not only are we one of those solutions, we're the number one solution.

Oh God. And I think it's time to turn your team into influencers. What do I mean by that? Well, LinkedIn, my friends, is driving a ton of citations [00:35:00] right now for these AI models, posts that offer advice, and I know that's hard, so that if you don't wanna offer advice. Do a side by side comparison of two people giving advice, do that right and say, we're not picking either one, but here's two different ways of looking at this.

We have found one bank took very well to that approach. Because the reason why influencers are important is because they give you internal distribution. So turn those people that have been into your company for ages into influencers who distribute your content.

What I will tell you is at least on perplexity, according to this study company, pages do do better than individual pages, but for all the major ones that most of your customers are gonna use, people like to hear from other people, and the AI is showing that bias. Now, I know some of you might be thinking, but what if I build an internal influencer group and they leave?

You know, it's like, well, what happens if they don't join you in the first place? 'cause this my friends, everything I'm seeing is the future. And I don't mean influencers in the role, your eyes, influencers. I mean people inside of your company who have [00:36:00] been there forever that have been working with customers in your community.

Just interview them and ask them more questions and get their views on the freaking web. And I hope that I built trust with you all today to where you might say, Hey, next time I do a prompt about AI optimization, I'm gonna type in what might Wil Reynolds say, because I hope that all of you double down on the trust that you've already built starting tomorrow.

Thanks so much.