Smarter Law Firm Marketing: AI Tools That Actually Work, with FirmPilot

Episode Notes

Transcript

Zack Glaser:

Hey y’all, it’s Zack and this is episode 580 of the Lawyers Podcast, part of the Legal Talk Network. Today we have a sponsored episode with Jake from Firm Pilot and we are talking about using artificial intelligence in your law firm marketing. Now, we’re not talking about artificial intelligence to just create content that nobody will read or sounds like a robot wrote. We’re talking about using ag agentic tools to create content that is actually intended to get you in front of the clients that you’re trying to get, using artificial intelligence to figure out what it is your potential clients are trying to read, what it is your potential clients are looking for, and get your website out in front of them. Now, we also talk about a EO or GEO or basically trying to get your law firm optimized, your law firm website optimized to get in front of people who are using chat, GBT, Claude or Anthropic to search for people like you. Now, if this sounds interesting, stick around and listen to my conversation with Jake from Firm Pilot.

Jake Soffer:

Hi, I am Jake. I’m the CEO and founder of Firm Pilot, and we use AI and data science to help firms do way better marketing.

Zack Glaser:

Jake, thanks for being with me. I appreciate it. Obviously, marketing in law firms is a big important topic in and of itself has been for years, but right now it’s a relatively hot topic because of the artificial intelligence and artificial intelligence has really adjusted how lawyers are doing their marketing. How is it affecting people, let’s say in a general sense, and then we’ll get a little bit deeper into this here in a second. How’s it affecting law firms in a general sense? The advent of chat, GPT and such

Jake Soffer:

Firms can just be so much smarter about what they’re marketing. Legal marketing isn’t anything new, just it’s turned from an art into a science, which is really cool to see and is kind of why firm pilot exists. It’s our background. My background’s in artificial intelligence and machine learning. Studied it back in school a long, long time ago, had another AI company. The way we started is kind of funny. My brother has a law firm, he’s a PI attorney in Miami. Throw a rock, you’ll hit a billboard, hyper competitive market. He worked with a few agencies and unfortunately just didn’t get lucky with the ones that he had worked with. So he asked me for help and just our natural approach was to take a data-driven approach to it. So he did something pretty funky. We used what are called transformer models,

Took high ranking content out on the internet from his competitors, other law firms, and we turned it into numbers or what we call vectors where we could almost grade the relevancy of content or quality of content. And instead of guessing or what I subjectively Jake or Zack would think was good content, we created content using AI for him that aligned with target numbers and it worked. A few months later, he was ranking number one on Google for Miami, Uber accident lawyer. He still does to this day, and his firm just took off. And that was kind of the light bulb moment for firm pilot, but more importantly of being able to use data to say this is how you should market, not, I think we should do it this way because it worked for a different client in a different market three years ago.

Zack Glaser:

Right. I think that’s a fascinating thing to dig into with artificial intelligence and marketing because part of that obviously is the creation of the content and we as lawyers have all seen the LLMs and you can go into chat GPT and somebody could easily say, I need a blog article on blank. But what it sounds like you’re saying is kind of using the artificial intelligence, using that on the front end and saying, what’s working? What are we trying to measure? And using it to help you have better reports, but kind of like front end reports, right?

Jake Soffer:

Yeah, so I mean, anyone can go to chat GPT and say, write me a piece of content, and it’s a really cool parlor trick.

Yes, it’s

Awesome. I use chat GBT all the time for stuff, but I wouldn’t use it for SEO. I would use it for a piece of the SEO process to assume it can do. The entire process is insane. That is an insane amount of electricity. If you truly understand how large language models work, they’re not as smart as you think they are. It’s more memorization than it is actual cognitive thought.

Zack Glaser:

That sounds like a lot of lawyers I know.

Jake Soffer:

So the way it’s working, when you say, Hey, chat GBT, write me a blog post. It’s been trained on billions of documents across the internet. It includes Reddit and Wikipedia, stuff that’s incorrect or you disagree with or in other states or outdated, and it just regurgitating that based on the probability of the next, not even word a token, which is sometimes a full word or less. So you don’t know if what you’re going to get is optimized for SEO to help you rank for those keywords. What you’re getting is chat GPT saying, based on my entire training corpus, this is what should be the next token in this sentence based on the probability from my training data. That’s all it’s thinking about. Now, there are reasoning models, and that’s another topic for another time, but yeah,

Zack Glaser:

Yeah. Yes. We throw around the word AI and we’re generally speaking about a very specific thing. So yeah, we’re kind of talking about LLMs here as opposed to anything else. So how does somebody get around that? That’s kind of a twofold thing. That’s like a garbage in, garbage out sort of thing. And it’s also, it’s not built to do the thing you’re asking it to do specifically. Right, exactly. So how do we get around that?

Jake Soffer:

Exactly. So could, if you want to spend, I think we’ve spent like $7 million and you don’t want to see our GPU bill to build this, but if you can either do it manually or build the infrastructure to have a multi-step agent orchestration system, we don’t say, Hey, LLM, write us a piece of content.

We

Have so many different steps. Think of it like a factory or I don’t know, Zack, if you’re a vodka, tequila, gin guy, whatever your drink of choice is, you want the top shelf stuff. What you don’t want is the moonshine. We run it through filter after filter, after filter, after filter after filter to get you the gray goose, Belvedere, Don, Julio, whatever that is, meaning it’s not, Hey LM write a piece of content.

We built agents that do competitive research. We do agents that do research strictly on schema markup. We do agents that do research strictly on formatting images, metadata, keyword insertion, all of this to fill a, let’s call it a job, to then create that content, which then goes through additional filters to get you that final output. So you could build it, you could use an LLM to help you if you know what each of those steps should be. Or of course there are end to end solutions that have that entire factory created, like IEA firm pilot.

Zack Glaser:

I like that metaphor there, but I think you forget that I’m from Tennessee, and so we are generally okay with our moonshine, but I will say you take it through filters, you put it into the appropriate barrels and you wind up with some pretty nice Jack Daniels. So yeah, I see what you’re talking about there. So it’s a smarter way to do kind of the same things, but you’re also talking about having to look at what is my goal, what’s my goal on the front end? And I think that in my opinion is where that’s that like a miracle occurs sort of thing that attorneys want to have happen. I don’t want to have to worry about a goal, I just want to have an LLM write my content, and then it just goes,

Jake Soffer:

If it was that easy, everyone would do it.

Zack Glaser:

Right, right. That’s exactly right. Okay, so there’s another side to this though, and I want to use your knowledge on this too. We did a little bit of the how do you create this content. On the other side is people have adjusted how they’re searching for lawyers in a way, and you and I were talking right, right before we started recording, some people call what we’re doing a EO, some people call it GEO, but it’s the optimizing for really how people are adjusting their search. Now, what are you thinking about when you’re thinking about a e, o and GEO and such?

Jake Soffer:

So we get asked about this every freaking day and we say the same thing. It’s always not the sexiest answer that people want to hear, and it’s probably not what you would expect from the AI enthusiasts and people who’ve been working in AI don’t abandon the fundamentals.

There are slight changes that we make when we’re trying to target certain chat. GPT mentions we’ve gotten really good at getting our clients mentioned for certain queries and chat, GPT and Claude and Gemini, but it’s not a completely different world, more conversational content, slightly different use of schema. You don’t want these 3000 to 5,000 word articles. Just answer a question, don’t give a whole background with legal jargon because again, you’re thinking about the electricity consumption of a chat GBT that’s crawling all these pages, but at the end of the day, do not abandon the fundamentals of SEO.

That is the core of how these things work. And what we also stress to clients are, well, do you remember extended text ads call only ads or no, they don’t exist anymore. Our point is, is chat GPTA fad? I don’t think so, but honestly, who knows or what’s going to come next? So what we try to stress to clients is don’t get tunnel vision of I have to rank on chat. GPTI have to rank on chat GPT. Is it valuable? Yeah, we have tons of clients who get cases from it, but don’t rely on one marketing channel. We use data to go through many channels. So yes, we love it, but it shouldn’t be your all in everything is what we try to stress. Don’t get over intense about it is what we stress to

Zack Glaser:

Clients. That makes a lot of sense because right now in the data that I’m seeing, and I don’t live in the marketing world, but I do live in the legal tech world, but the data that I’m seeing is that about 20% of people are searching first with using chat GPT or perplexity or something like that, which is not an overwhelming number. Most people are still using Google to search or something like Google to search. Well, I’d say most people are still using Google, but one of the things that you’re hinting at is A, that’s going to change a hundred percent. B, we don’t know exactly how it’s going to change.

Jake Soffer:

There’s two things I say to that is I try not to look too much at those numbers. It’s large enough of a market share that it’s a marketing channel worth paying attention to.

Zack Glaser:

That’s

Jake Soffer:

About it. But I tell clients, let’s not go crazy analyzing that because it’s one, it’s such a small amount of data.

I’m a huge fantasy football guy and I lost week one, but throughout beginning of the game it said I had a 50% chance of winning. Then it said I have 10, then 80, then 20, then 90 I lost. The point is, if you truncate the time span, it’s so volatile, we can zoom into anything in the world that has some probability measurement and find a point where it was 80% and find a point when it’s 10%, five years from now, we can look at the market share of chat GPT, and I’ll be able to point to a place where it had 5%, maybe 80 20 versus 60. The point is we can’t just zoom in on one data point that’ll eventually be a speck of dust in five years of data. So anyone who says 20%, that’s massive. I need to see that hold for an extended period of time, number one. Number two, public markets are usually a really good indicator of what’s happening in the real world and look at Google stock. So if SEO was dying, if search really was dying, Google stock wouldn’t be doing what it’s doing right now at, I mean, look at their revenue breakdown and how much is attributed to search and ads. So that’s the other thing I look

Zack Glaser:

At. That’s a good point. So with that, like you’re saying, a lot of your advice is in relation to this specifically is do the SEO fundamentals with a couple of little changes. What’s one of the little changes that you’re thinking about when saying, alright, yeah, let’s do good SEO,

Jake Soffer:

But

Zack Glaser:

With a shade of GEO or a EO?

Jake Soffer:

We have slightly modified the way our models will write content, especially if we’re trying to get a certain query in chat, GPT or Gemini for our client and being answer first content. So if you look at legal blogs from four or five years ago about the same exact topic, can I sue Uber if I was hit by an Uber driver and today, can I sue Uber if I hit by an Uber driver? Chances are the content from six, seven years ago is going to be way longer. And within the article, when you have different subsections of how to report it, what kind of damages can you get? The first few sentences are just stuff that doesn’t matter. Legal jargon, statistics. Today we tell our clients we need to just answer the damn question. The first sentence needs to answer that because something that’s crawling your site just wants to know, can I get the answer here quickly? If not, I’m out. So if it has to go through three, four sentences of historically X percent of people have blah, blah, blah, blah, just answer the question.

Zack Glaser:

That’s what you mean by these are using a lot of electricity. They’re trying not to use much electricity by reading stuff that has a low probability or potentially a low probability of actually getting them the answer.

Jake Soffer:

I mean, think about it, by the time we’ve been on this call, Zack, how much content has been created on the internet, not just in the legal world, but everywhere and to crawl and crawl and crawl and crawl. That is expensive.

Zack Glaser:

Yeah. Hell, this is content that’s going to be on the internet.

Jake Soffer:

Exactly.

Zack Glaser:

So at least this. Yeah. Okay. We could obviously dig into this a lot further, but before we go into how other people can dig into it further with you, what would you suggest? Somebody that is a lawyer that is like, I hear about all the AI that I can use in my marketing. I’m looking at it. I want to jump in, but I don’t want to dive headlong. What would you suggest that they do? Kind of first step,

Jake Soffer:

There’s a balance of one on one hand, you just have to take a leap, but you have to set up the expectation of this is an experiment, not a guarantee. There are no guarantees in any change management for your business. So it’s the same thing when you sign up with a marketing agency or get a new va, there’s no guarantee this is 100% going to be the thing that accomplished your goal. So you have to have that expectation going in that I’m going to be open to the idea that this experiment for my business could fail, and that’s how you innovate, fail fast. So you have to have that expectation. Number two, do your research. If a client wants to talk to references of ours, we have tons. It’s not a fluke what we’ve been doing. We’ve now done it for more than a hundred firms. We raised over $12 million from notable investors like Thomson Reuters and HubSpot. Clearly, we’re onto something. We like to think so look into who you’re going to work with. Are you the first client of theirs? And it doesn’t mean, hey, don’t try it, but just again, back to point number one, set your expectations. We had a first, a second, a third client at some point

Who took a chance on us. So just know what you’re getting into. Do your research on not just the product but the company.

Zack Glaser:

I like that. And inherent in that advice is go to the experts. You are not in this alone. You don’t have to do this on your own and go out and create a complex series of agents to distill everything through and all of a sudden come up with this. Go to the experts that have already done that at firm pilot or something like that. I like that. Well, then the last question, Jake, is where can people connect with you at Firm Pilot if they want to learn more about this?

Jake Soffer:

Yeah, absolutely. It’s easy firm pilot.com. They can book a call to see a demo of the platform live or even just get a whole bunch of free resources from our website. If you’re not ready to scale your firm and you’re just in the early exploratory stages, no pressure. We’re not super salesy guys. We just love what we do and we’re passionate about it, and we post a lot of free content about things you could do for free just to kickstart your marketing if especially you’re a brand new firm or a firm that’s large, but you just want to experiment a little bit.

Zack Glaser:

Yeah, love it. Okay, so firm pilot.com and we will put a link to firm pilot in the show notes. Jake, thank you so much for being with me and for sharing your information.

Jake Soffer:

Thank you, Zack.