--- title: "AI in PPC: Strong Case… But Some Weak Matches" serpTitle: "AI in PPC: Strong Case… But Some Weak Matches" description: "Frederick Vallaeys and Anthony Higman explore why AI Max and PMax struggle in high-cost legal PPC and compliance risks in automation." date: "2026-04-22" url: "https://www.optmyzr.com/ppctownhall/ai-in-ppc-legal-automation-risks/" --- # AI in PPC: Strong Case… But Some Weak Matches > PPC Town Hall 127 Frederick Vallaeys and Anthony Higman explore why AI Max and PMax struggle in high-cost legal PPC and compliance risks in automation. **Published:** April 22, 2026 **Watch:** [YouTube Video](https://www.youtube.com/watch?v=QGhc7LCLslM) **Apple Podcasts:** [Listen](https://podcasts.apple.com/in/podcast/ai-in-ppc-strong-case-but-some-weak-matches-ppc-th-127/id1508399985?i=1000763108638&l=mr) **Spotify:** [Listen](https://open.spotify.com/episode/0zmCa8nTwd5WFKcweAk2pV) --- ## Episode Description Managing Google Ads for personal injury law firms isn't like managing campaigns for e-commerce brands. When a single click costs $300, the stakes change completely. Frederick Vallaeys, CEO & Co-Founder of Optmyzr, sits down with Anthony Higman, founder of Adsquire, to explore how AI and automation are changing Google Ads, especially in high-cost, high-risk verticals like legal PPC. Anthony has been managing campaigns since 2011, back when everything was keywords and manual bidding. He's watched the industry evolve from ultimate advertiser control to increasingly black-box systems, and he's tested every new automation feature Google has released. The episode also digs into broader questions about where AI actually helps versus where it creates new problems. If you're navigating AI in Google Ads and want a practical, experience-based perspective that isn't just cheerleading for automation, this episode offers grounded takeaways from someone operating in one of PPC's most unforgiving verticals. Here's what you'll learn: * Why AI Max turns specialized lawyers into generic "lawyer" searches * The compliance nightmare: when AI writes your ads (and breaks the rules) * Why first-party data doesn't actually differentiate * How conversational search complicates everything * Are advertisers just unpaid testers for platforms that aren't ready? * Why AI creates winners and losers in the legal industry * Why Anthony shifted from training juniors to hiring veterans * Why most people stop at rebuilding old systems * The risk of automation: when people stop paying attention, platforms shake the cushions * Using meetings as prompts: one practical AI workflow that actually works --- ## Episode Takeaways Anthony Higman started in the mailroom of a personal injury law firm after college. He worked his way up to managing all their Google Ads. That was 2011, back when everything was keywords, manual bidding, and complete advertiser control. Now, 15 years later, he runs Adsquire, an agency that manages Google Ads exclusively for law firms, primarily in the personal injury space. Fred opened the conversation by asking how someone who's spent 15 years with ultimate control over every keyword feels about being asked to run AI Max campaigns, which remove keyword-level visibility entirely and operate as black boxes. Anthony's answer was immediate: "Personally, I hate it." The conversation that followed was a detailed unpacking of why automation that works in general e-commerce contexts often fails catastrophically in high-stakes, regulated verticals. Anthony's clients operate in one of the most expensive advertising environments in PPC. Average cost-per-click can hit $300 or higher. A single irrelevant lead doesn't just waste budget; it erodes client trust fast. And compliance isn't optional. Saying the wrong thing in an ad can trigger bar association complaints. Anthony has tested every major automation push Google has rolled out. He's run AI Max campaigns across his client base. He's dealt with local services ads dynamically pulling website content and making claims that violate attorney-client privilege. And he's watched platforms push features into sensitive ad categories before those features are ready. And this PPC Townhall episode is entirely about those perspectives and real situations. ### Why AI Max turns specialized lawyers into generic "lawyer" searches Fred asked Anthony about AI Max specifically. Google pushed it hard, positioning it as one of the only ways to access AI-powered ad placements. So Adsquire tested it across all its clients. "The results were pretty bad, to be honest with you," Anthony said. "And I think it's another consequence of automation, especially in what we do." The core problem is that PMax and AI Max tend to collapse nuance. Legal practices specialize, like a firm handles truck accidents, another handles medical malpractice, and another does workplace injury. These are all subsets of personal injury law, but they're distinct practice areas with different client needs, different case economics, and different expertise requirements. > "What it tends to do is turn every specific nuanced vertical of law into 'lawyer' or 'law firm,'" Anthony explained. "And lawyers don't like that. They're all specific law firms—medical malpractice, personal injury, truck accident. There are variations of personal injury law that certain law firms specialize in. > > And PMax, all the automation types for the most part, tend to turn law firms into the lowest common denominator, which is just 'law firm.' And then they get landlord-tenant dispute calls and just all sorts of wild stuff, and they hate it." This isn't a minor annoyance. When a truck accident firm gets a call about a landlord-tenant dispute, that's a lead they can't take. It's a wasted budget. And at $300 per click, those mistakes compound fast. Fred pointed out that this problem gets worse in regulated industries, where you can't say whatever you want in ads. If Google is automating ad copy and dynamically generating messaging, there's a loss of control that can lead to compliance violations. ### The compliance nightmare: when AI writes your ads (and breaks the rules) Anthony shared a recent example that illustrates the stakes. Google's local services ads started dynamically pulling content from law firm websites and inserting it into ads. The system is a black-box that advertisers can't control what gets pulled or how it's used. > "There was an instance where on local services ads it said 'everything is confidential,'" Anthony explained. "And the whole premise of local services ads is Google's recording and listening to all the conversations, and nothing's attorney-client privilege." So the ad promised confidentiality, while the platform itself violated it. That's not just bad marketing. In legal advertising, that's a potential bar association complaint. > "I get it. Over time, given enough bandwidth and runway, I'm sure these things will work themselves out," Anthony said. "But I am exceedingly angry at the pain that we have to endure while it gets worked out, to be honest with you." Fred connected this to the broader reality of high-cost verticals: when your average client feels miffed about an irrelevant click in e-commerce, that's frustration. When a law firm pays $300 for a click that leads to the wrong case type or triggers compliance risk, that's anger, and it shortens client retention dramatically. > "Law firms don't hang long when things are going south," Anthony said. "They're very quick to be like, 'Okay, what else can we do if Google is not working anymore?'" ### Why first-party data doesn't actually differentiate One of the sharper critiques Anthony offered was around the industry's shift from keywords to first-party data. Google positions this as progress: moving from static keyword targeting to dynamic audience signals that reflect real user behavior. > Anthony sees it differently. "I feel like it's a scam in that they're really just trading keywords for first-party data," he said. "But what does that really change, honestly?" His point: if you're a personal injury law firm, your first-party data is people who submitted a contact form, called your office, or scheduled a consultation. If your competitor is also a personal injury law firm, their first-party data is… people who submitted a contact form, called their office, or scheduled a consultation. > "Everybody's first-party data, if you're of a certain business type, is going to be the same," Anthony explained. "We're all going after the same conversions. We're all going to have the same audience data. So what is it really doing besides trading this thing that had more control for this thing that has less control?" Fred acknowledged the point but pushed on conversion efficiency: maybe the differentiation comes from how well you convert those prospects into paying clients. That becomes a first-party data signal fed into the bidding system, allowing more efficient converters to bid higher and still hit profitability targets. Anthony brought up a proposal he'd made to Google: "Okay, I get it. It's shifting to this. It's not going to stop. I can yell as much as I want, but they have their internal plan. So then give me more control over the audience data. Give me more levers to pull there—what I can target, when, why. Make first-party data more like keywords, if that makes sense." The underlying frustration is that if everyone's targeting the same audiences with the same data, how does automation actually create competitive advantage? And if it doesn't, why remove the control advertisers had with keywords? ### How conversational search complicates everything The conversation shifted to AI overviews, ChatGPT, and the broader move toward conversational, prompt-based search. In the past, someone searching for "personal injury attorney" triggered a clear signal. Now, someone might have a multi-turn conversation with an AI about their car accident, their medical bills, whether they have a case, and what kind of lawyer they need and the final prompt might not even include the phrase "personal injury attorney." > "How do you know what is the trigger point when eventually the conversation someone had with a chatbot becomes that clear signal that your law firm would have been a good ad to present?" Fred asked. Anthony acknowledged the shift is happening. Lawyers are tech-savvy and love shiny objects, so they're already asking about ChatGPT referrals. > "ChatGPT referrals are novel, so they stick out like a sore thumb, and people get excited about them," Anthony said. But here's the problem: legal ads aren't showing up in most of these AI-powered platforms yet. And Anthony understands why. > "They're not going to do it for sensitive ad categories—finance, injury, pharmaceutical," Anthony explained. "And the reason is because LLMs still hallucinate. They can't have a weird query relate to something way far out there and then show an ad, because that would be a media nightmare for them." So the platforms are treading carefully. Google might work legal ads into AI mode eventually, which is more grounded in the Google knowledge graph and less prone to hallucination. But for now, legal advertisers are stuck in a strange middle ground: AI is the future, but they can't access the AI ad formats yet. ### Are advertisers just unpaid testers for platforms that aren't ready? Anthony made a broader point about testing. He suspects Google used to test new features primarily on non-logged-in users. Now, with login requirements more widespread, everyone is part of the testing ground. > "It seems like the testing ground used to be not-logged-in users, and now it's just the world," Anthony said. "And so we have seen weird anomalies and performance dips. It seems like it's geo by geo, but again, there's not a lot of info being passed to us in this weird testing ground that is now just real life." Fred agreed that nobody fully knows where this is all heading because it's genuinely being figured out in real time. The transition from keyword-based advertising to prompt-based, conversational advertising is happening on the fly. But Anthony's frustration is rooted in the terms and conditions of Google Ads, which essentially say: "We can do whatever we want at any time, and there's no recourse for you." > "Why I'm angry about AI for the most part is that it just seems like a wild testing ground that really nobody knows what the direction is yet, or all these bugs that are still trying to be worked out," Anthony said. "And we're just guinea pigs for it all. And that hurts sometimes." Fred acknowledged the pain but pointed out the competitive pressure Google faces. ChatGPT represents both a monetization risk (people finding answers there instead of Google) and a monetization opportunity (ChatGPT can sell ads). So Google can't take a slow, methodical approach. They're moving fast because they have to. > Anthony responded, "You're not wrong. I mean, it is all driven by the code red from ChatGPT coming out. I know they have more data and more awareness of these things before anybody else does. So I don't disagree with you. It just kind of sucks half the time." ### Why AI creates winners and losers in the legal industry Recently, Claude Code released plugins specifically for contract review and legal research. Some legal stock entities saw huge valuation drops. But in the same week, the Wall Street Journal reported that some lawyers now charge $3,500 per hour, and clients don't flinch. > "You've got this one place where lawyers are becoming unnecessary, and this other place where lawyers are just commanding higher fees than ever," Fred said. "How do you see that shift, and how do you think that's going to impact people marketing on behalf of law firms?" Anthony's answer was broader than just legal. He sees this pattern across industries. Social media and AI-generated headlines drive market movements without people digging into the details. A karaoke machine company released a white paper claiming breakthroughs in autonomous trucking, moved markets, tanked stocks and the white paper can't be found on the internet anymore. > "We're at such a weird spot in society where social media is a really big part of everybody's lives right now, and lots of people still read things intentionally and take that time," Anthony said. "But AI is kind of accelerating this—'Give me the AI summary of what happened'—just these little short bursts where people don't dig deeper into the details." He suspects most AI-powered legal tools follow a similar pattern: people get excited, dive in, spend AI credits, build something to 60% completion, hit a wall, and it fades. "But the AI companies win because everybody's diving in to spend their credits," Anthony said. As for the $3,500-per-hour lawyers? Anthony sees that as the natural outcome of AI lowering barriers to entry. When tools democratize basic legal work, the truly expert practitioners become more valuable, not less. > "I think it makes the good people more valuable," Anthony said. "All these companies are putting AI stuff right into the platform, which is lowering the barrier of entry for people. But I think that makes the good people more valuable, if that makes sense." Fred agreed and connected it to hiring. In his company, the most valuable people are becoming even more valuable because they can use AI to amplify their expertise. Meanwhile, junior work that used to require a human can now be done by AI and reviewed by a senior person. > "Why not just have the senior people review what Claude or GPT or Gemini was doing?" Fred said. "That kind of makes a challenge for entry-level folks." ### Why Anthony shifted from training juniors to hiring veterans Anthony acknowledged he's shifted his hiring strategy, maybe unintentionally. He used to hire smart, curious recent college grads and train them up. Now he's hiring more experienced people. > "I recently shifted towards I want more experienced people," Anthony said. "Maybe unintentionally I've made that shift based on everything that's going on." But he's still looking for the same core trait: curiosity. "I want someone that wants to peel the onion. What's under this? Why is it like that? This doesn't make sense," he explained. Fred had a harder question: if AI is doing entry-level work, what do those people do? "We're just going to put all of these people out of business—for what?" Anthony said. "I guess it's for advancement. It just—again, there are so many bugs that I don't see the advancement to where I think it needs to be." Fred pointed out the pace of improvement. Tasks that were buggy or slow a year ago are now orders of magnitude better. Engineers at AI companies are using AI to build the next generation of LLMs, creating a compounding effect. > "These things are perhaps just a couple of months away from that next iteration where it just blows our mind again at something we didn't think it could do before," Fred said. ### Why most people stop at rebuilding old systems Fred shared his own experience about vibe coding extensively to build things like an Advent calendar for PPC advice, a lightweight Udemy clone for Optmyzr training, and holiday-specific social media content generators. But the broader point Fred made was about shifting how marketers think about their time. If you're no longer managing keywords and bids manually, what should you be doing instead? > "What I've been spending all this time managing keywords and managing bids, but if I didn't have to do those things and I could actually be a marketer, what would I want to do?" Fred said. That reframe is what unlocks creativity. Not "how do I rebuild the thing I was doing before," but "what have I always wanted to do that I never had time for?" Fred expressed frustration with people who start vibe coding, rebuild the previous system, and then stop. They can't imagine what comes next. > "I am very disappointed usually in people when I give them a vibe coding tool or they start vibe coding, but they kind of stop at having rebuilt the previous thing and then they just can't imagine what is next," Fred said. He blamed a lack of creative training. "Too many of us have just not grown up being asked to be creative because it becomes about, 'Well, you want to work in digital marketing, so you need to know what buttons to push in Google Ads to do the thing that's necessary.'" But those buttons are just the interface layer. The underlying system is what matters. And if you had access to the underlying system—not just the five buttons Google gives you—imagine what you could do. Fred gave an example: with API access, you can generate reports that go far deeper than the Google Ads UI. You can pull keyword-plus-ad-level performance data, something Google doesn't expose because it's overwhelming for most users. > "If you're creative in this field and you ask that question, you can now build it very easily," Fred said. "And that's what excites me. But I also need people to really think beyond what they've historically done." Anthony appreciated the perspective but raised a counterpoint that's been gnawing at him. ### The risk of automation: when people stop paying attention, platforms shake the cushions > "A lot of stuff that gets automated, you pay less and less attention to it over time," Anthony said. "Everybody's got an AI note-taker now. I think that's cool because when something comes up, you can go back and reference it. But I think the amount of time that people actually go back to their AI note-taker dwindles down." Smart bidding is an example. Google pushed it hard. The message was: spend less time looking at bids, we'll handle it. And advertisers did stop looking. > "I think there's more BS that gets pushed in when things stop being paid attention to," Anthony said. "More crap searches get put in. More manipulation happens when you're not looking at the thing that's really important anymore because you're off coming up with new ideas." He's not saying platforms are inherently malicious. But they're businesses. Their goal is profit. And at global scale, tiny inefficiencies—irrelevant search matches, slightly inflated CPCs—add up. Fred acknowledged two realities: first, people consciously choose to stop paying attention when they decide the time cost of monitoring outweighs the potential savings. Second, platforms do obfuscate data, making it harder to validate what's happening even if you want to. > "Even if I wanted to go and validate what they were doing, I just can't," Fred said. "With prompt-based advertising, it's not tying back to a keyword. They're never going to give me the full history of why that ad was shown. So I just have to at some level trust the system." That's hard. Especially when platforms have a profit motive that doesn't always align perfectly with advertiser interests. ### Using meetings as prompts: one practical AI workflow that actually works Historically, meetings generate tasks and work. But what if you approached meetings differently, knowing the AI note-taker is transcribing everything? > "What if you could really specify what are the criteria, what are the restrictions, what are the goals that we have, and come out with all of those things that you would have otherwise given to a large language model prompt?" Fred said. At the end of the meeting, you give the transcript to your LLM and say, "Go and do the things that needed to be done." Because you were intentional about loading it with the right questions and answers, the LLM can execute better. Fred does this specifically for content ideas. After internal meetings, he gives the transcript to an LLM and asks: "What could I blog about this? What are some ideas? Even if this was an internal meeting, what are some of the bigger ideas that could benefit the industry as a whole?" > "Meetings have never been seen as a marketing and messaging vehicle, but now you can add that on," Fred said. If you're navigating automation in high-stakes verticals, this conversation is worth your time. --- ## Episode Transcript **Frederick Vallaeys:** Hello and welcome to another episode of PPC Town Hall. My name is Fred Vallaeys. I'm your host. I'm also CEO and co-founder at Optmyzr, a PPC management tool. So for today's episode, we have Anthony Higman, who is the founder and CEO of Adsquire. He operates in one of the most expensive verticals that exist in PPC advertising. So if he has been testing something, chances are pretty good he's put a good amount of money towards it, and his findings are going to be meaningful for all of us listening today. Now, Anthony also has very strong opinions when it comes to AI and automation and many of the things that Google is doing. So I think he's going to be a great person to learn from as far as what he has seen work in AI, but also what to be cautious about and where he sees people often making mistakes. So super excited to have Anthony on as a first-time guest on PPC Town Hall. So thanks, everyone, for watching, and let's get rolling with this episode. Anthony, thanks for joining us. **Anthony Higman:** Thanks for having me. **Frederick Vallaeys:** All right. So Anthony at Adsquire—tell us what you do, but also how you got there because you had kind of an interesting story, right? **Anthony Higman:** Yeah. So we manage basically anything Google advertising for law firms. A lot of our clients tend to be in the personal injury space. And yes, interesting story how I got here. I actually started in the mailroom of a large personal injury law firm after college and worked my way up to advertising, all of their Google advertising, and really elevated all of that and got super interested in it. And this was in 2011, so everything was keywords and manual. And I've kind of seen the evolution up until now, which has been interesting to say the least. **Frederick Vallaeys:** Yeah, for sure. So someone who's been doing this now 14, 15 years, going from having ultimate control over every keyword to being asked to run AI Max campaigns—let's talk about that. I mean, how do you feel about that change in where the control lies, and is that a good thing or a bad thing? **Anthony Higman:** Yeah. So personally, I hate it. I do understand aspects of it. I think I agree with you more than you would think that I do in a lot of the things that are happening. Why I dislike it is because it seems like every advancement in automation takes away more control from advertisers. And I think it's really an important distinction that more gets hidden in that bucket of automation. The really valuable insights seem to be getting hidden more and more in the name of automation. And I think just a lot is lost there in actionable insights that you can get from the data that used to be there that's now very black-box and hidden because of automation. With AI Max specifically, that came out. It was being pushed really hard by Google. We obviously always test everything that comes out and are on the new things that are emerging. So we jumped on it as an agency. And especially when they were saying, "Hey, it's one of the only available things for AI ad placements," we got into it. We said, "Okay, maybe that's the lesser of the three evils basically for what we do specifically." And we tested it for all of our clients. And the results were pretty bad, to be honest with you. And I think it's another consequence of automation, especially in what we do. So I yell a lot, and what I yell about is mainly specific to me because it's about law firms. And I think that's another thing—it's a sensitive ad category. Personal injury law, pharmaceutical advertising—I think these things have kind of been hit the hardest with the automation push because there's just data gaps that they can't—we can't use specific things. Google has tried to come up with different solutions to that, but none of them have really worked in my opinion. Again, back to Performance Max—what it tends to do is turn every specific nuanced vertical of law into "lawyer" or "law firm." And lawyers don't like that. They're all specific law firms—medical malpractice, personal injury, truck accident. There's variations of personal injury law that certain law firms specialize in. And PMax, all the automation types for the most part, tend to turn law firms into the lowest common denominator, which is just "law firm." And then they get landlord-tenant dispute calls and just all sorts of wild stuff, and they hate it. So it's been a struggle. There's still lots of things that do work for what we do. But that's kind of my main beef with it—for specifically sensitive ad categories, even with all the solutions that Google's coming up with. And I do applaud them trying to come up with solutions for what I'm talking about, but it's not working. **Frederick Vallaeys:** Yeah, it's tricky. And especially in these regulated verticals, it's also tricky because you can't say whatever you want in the ad where you want to be very specific. And so if Google is automating that portion of it, there's a loss of control which can lead to—I mean, especially in pharmaceuticals and financial instruments—you can get into pretty big trouble for that, right? So how do you look at that and sort of controlling what the AI is able to say on your behalf that doesn't get your clients into hot water? **Anthony Higman:** Yeah. I've actually brought something like this to the internet's attention recently, and again it goes back to the whole automation thing. There's local services ads, and they just did something probably three or four months ago where they're dynamically pulling things from the URL that you put into local services ads. So you put your website's URL into the platform, and it's a very black-box platform, and it's just going to pull different USPs and put them onto local services ads on your behalf. You can't control it. I assume unless you see that and then maybe change anything that says that on your website. But there was an instance of this where on local services ads it said "everything is confidential," and the whole premise of local services ads is Google's recording and listening to all the conversations, and nothing's attorney-client privilege. And so again, just another thing where it just doesn't line up. I get it. Over time, given enough bandwidth and runway, I'm sure these things will work themselves out. But I am just exceedingly angry at the pain that we have to endure while it gets worked out, to be honest with you. **Frederick Vallaeys:** Yeah. And I suppose that pain and the anger comes from the fact that your client—the average client—kind of feels miffed about getting a click on something that's not highly relevant. When you're paying a couple hundred dollars for that click, that being miffed turns into real anger, right? And then that probably— **Anthony Higman:** Exactly. Yeah. And honestly, I don't know if it's probably like this for every vertical to be honest with you, but law firms don't hang long when things are going south. They're very quick to be like, "Okay, what else can we do if Google is not working anymore?" And so that's kind of my internal struggle right now. I get the shift. I get where Google wants to go with it. I just don't agree with it all, to be honest with you. And even take away the sensitive ad category stuff that I always yell about—because you're right, there are things that you can't say. And I think a lot of the solutions and the push is AI Max—let it make up ad copy for you. And we just can't really do that because it sometimes turns out negatively. I kind of just disagree with all of it because I get the data push and I get what Google's doing. They want it to be predictable. They want everybody to just be able to go in there, select a goal, press a button, and type in their inputs. I just think it doesn't really work that way in the real world. And when you factor in all this algorithm learning time and all the data—I think there's a thing with too much data that it reaches this weird level of complexity that makes it weird in a lot of aspects. I'm a Google nerd. I am on Google all the time. When I'm not working, I'm looking and trying to find stuff and understand from a user's perspective what's happening. I just think it's going to get to this level of creepiness with what they're doing that it's going to turn a lot of people off and lead to more data gaps and more people trying to find workarounds from this constant surveillance. It's going to be in two camps. Some people have given up on the whole "there's any privacy in the world." I think people that are under 30 don't really care anymore. And then there's older people—I'm getting old—who are kind of angry about it. "You can't track me like this." I think there's just going to be unintended consequences of everything that they're doing with all the data and that there's a thing to me at least with too much data and how it reaches this level of complexity. On that point too, I feel like it's a scam in that they're really just trading keywords for first-party data. But what does that really change, honestly? I get what they're saying. "Oh, it's more conversational queries." But it's the same thing. They're just trading this one thing for this other thing. And everybody's first-party data, if you're of a certain business type, is going to be the same. We're all going after the same conversions. We're all going to have the same audience data. So what is it really doing besides trading this thing that had more control for this thing that has less control? **Frederick Vallaeys:** Yeah, that's a really interesting point—how the first-party data is really the same for every lawyer who's looking for a specific type of case. **Anthony Higman:** Yeah. In any business, in my opinion. If you're selling widgets or whatever you're selling, I feel like it's the same audiences that you're going after. And then your differentiating factor maybe becomes how efficient you are at converting those prospects into customers, and then that becomes a first-party data point which can be fed into the bidding mechanism so that you come out with a profitable endpoint. So maybe there's different treatments of going after the same audience, but how exactly you pursue it and which keywords you can bid on versus can't bid on because maybe you're not efficient enough to convert it at that rate—that's where the difference lies. This is kind of something I brought up to Google too. Okay, I get it. It's shifting to this. It's not going to stop. I can yell as much as I want, but they have their internal plan. Okay, then give me more control on the audience data. Give me more levers to pull there—what I can target, when, why. Make first-party data more like keywords, if that makes sense. Because I think there's going to be a thing with automation that it's just going to go after the same thing for multiple business types. And I'm not sure how it accomplishes a return on ad spend or whatever the new plan thing they're talking about—investment plan. I'm not sure how it gets everybody there if we all have the same kind of data that we're going after for the same thing. **Frederick Vallaeys:** Yeah. Now about this data and talking about wanting maybe more insight into the search terms—but let's talk about AI and AI mode, AI overviews, ChatGPT, people using Claude to find stuff. You said this is more prompt-based. This is conversational. This is no longer just individual keywords. So it kind of complicates that scenario, right? Because in the past, you might have wanted a keyword like "personal injury attorney," and that's what you want the ad to show. But now that we're in this whole conversational landscape, how do you know what is the trigger point when eventually the conversation someone had with a chatbot becomes that clear signal that your law firm would have been a good ad to present in that scenario? And the last prompt that they did at that point may kind of miss the whole thing that came before it that actually pointed to the fact that this would eventually be the right outcome. So obviously this data is going to become more plentiful like you're saying, but now there's too much data, so maybe you can't pick the signal out of it. How are you thinking about that? And how are you talking to your clients? Is that even a conversation point at this stage? **Anthony Higman:** It's becoming more of one. I think that lawyers are a little—they're slightly more tech-savvy than the average bear, and they are always on the new thing. They love the shiny object. So we are having these conversations. ChatGPT referrals are novel, and so they stick out like a sore thumb, and people get excited about them. "Oh, that was a really great lead that we got from ChatGPT. How do we lean into this more?" And we're on that. We're trying to get access to ChatGPT ads as soon as possible. We filled out a thousand forms. I get what you're saying—that they're more conversational. To me, it's like bring back broad match modified and let me say these words need to be included in the conversational path. That solves the problem to me. But I get it—it's going to advance and get further. So these conversations are beginning to happen. "How do we get into more of these AI ad formats?" It's not happening for our vertical, though. And so this is another weird spot to be in because they're not going to do it for sensitive ad categories—finance, injury, pharmaceutical. And I think the reason is because LLMs still hallucinate. They can't have a weird query relate to something way far out there and then show an ad, because that would be a media nightmare for them. So they're treading very lightly, all of the platforms that are pushing the AI content, on what they're going to do with sensitive ad categories. Somebody recently pointed out they might start working them more into AI mode, and that AI mode is kind of more grounded in the Google knowledge graph. That could be a way. I know Google is full of very, very smart people, and they're working these things out. But to me, it just kind of sucks because we're stuck in this middle ground of "AI, AI, AI, AI," and then everybody kind of getting hurt. And we've seen it over the last three years—the seesaw of results when they're testing a new thing. And I think they've really ramped up testing as they've kind of locked everybody into "you got to be logged in." I don't know—you worked at Google. I feel like the testing ground used to be not-logged-in users, and now it's just the world. And so we have seen weird anomalies and performance dips. It seems like it's geo by geo, but there's not a lot of info being passed to us in this weird testing ground that is now just real life. And this is also why I yell a lot about AI—because it just seems unfair in how they're doing things. And in the terms and conditions of Google Ads, which is a brilliant little clause they have in there, it's like "we can do whatever we want at any time, and there's no recourse for you." So why I'm angry about AI for the most part is it just seems like a wild testing ground that really nobody knows what the direction is yet, or all these bugs that are still trying to be worked out. And we're just guinea pigs for it all. And that hurts sometimes. So that's why I yell and I'm angry. **Frederick Vallaeys:** Yeah, it hurts for sure. But I think you're right. It's nobody knows, and it's not that they're not telling us. It's just such a new experience, and nobody's figured out how to transition from this world that we've been building for 20 years where everything revolved around a singular keyword. And now we have to figure out how to place ads when it's these massive prompts and conversations. And so this is being figured out on the fly. But at the same time, they can't take a really slow and methodical approach because—I mean, you saw that your clients are getting these referrals from ChatGPT. People are going on there to find stuff. And it's not like Google is dying—obviously they have Gemini, so they've got amazing technology. But people are shifting how they find stuff and get answers. And that is a monetization risk to Google and a monetization opportunity for ChatGPT. So that's why they're going to do things that maybe make us unhappy, because they're trying to figure it out as quickly as they can too. **Anthony Higman:** Yeah, you're not wrong. I mean, it is all driven by the code red from ChatGPT coming out. So I always try to see the Google angle. I know they have more data and more awareness of these things before anybody else does. So I don't disagree with you. It just kind of sucks half the time. **Frederick Vallaeys:** Well, it sucks for us, and you're angry. But let's talk about lawyers, right? You work with a lot of lawyers. So recent news is that Claude Code now has these plugins and connectors specifically for contract review and doing a lot of legal work. So a lot of big legal stock entities had a huge drop in valuation because of that. That's one end of the spectrum. But then in that same week, I read in the Wall Street Journal that some lawyers now charge $3,500 bucks an hour, and clients don't even flinch at that anymore. So it's the highest rates they've ever charged. So you got this one place where lawyers are becoming unnecessary, and this other place where lawyers are just commanding higher fees than ever. Is that—how do you see that shift in the industry? And besides that shift, how do you think that's going to come back to impact people marketing on behalf of these law firms? **Anthony Higman:** Interesting questions. Yeah. So I see this in lots of industries right now, actually. I think that we're at a very interesting place in society right now. It's driven by a lot of different things, in my opinion. But I think a lot of the world right now and markets are driven by little headline bursts, to be honest with you. I don't know if you saw the thing—it was about autonomous trucking. There was this karaoke machine company that turned into an autonomous vehicle company. They put this white paper out that can't be found on the internet anymore, and that literally moved this market and all these stocks severely tanked. I think we're at such a weird spot in society in that social media is a really big part of everybody's lives right now, and not nobody—lots of people still read things and are intentional and take that time. But I think AI is kind of accelerating this—"Give me the AI summary of what happened"—just these little short bursts where people don't dig deeper into the details, and that kind of drives these crazy stories to just go further. And then the next social media person picks it up and keeps pushing that. So back to your point about the lawyer thing—I don't disagree that there's something to it and that these tools are in ways democratizing a lot of information and making things easier. I think it has weird effects—just weird effects. Markets are driven a lot more now by short little blurbs and stuff. And then there are people that then dive into that because of the blurb. "Okay, how do I use this thing now for my law firm? How do I use AI to get more work done or whatever the case may be?" I think that's kind of what all the AI companies want—"Yeah, dive in and figure out how to vibe code this thing for your business." I think it doesn't go anywhere. It's a weird revolving circle where people spend lots of AI credits to half vibe code the solution. And they get it to 60%, and it's this really cool concept, but where do you go from there? So I think it's a lot of weird social media stuff that drives these things. People dive in, they get it to 60%, they can't really get it past that, and it kind of just fades back. But the AI companies win because everybody's diving in to spend their credits. That's my view on it at current. It's sort of a mirage. But I do think that bigger companies are working on these things, and they are gaining traction. I did not know about the lawyers charging $3,500 an hour thing. That's fascinating to me. It might be kind of what happens to us in the same regard. All of these companies that we work with every day are putting all the AI stuff right into the platform, which I also hate. But it's lowering the barrier of entry for people. But I think that it makes the good people more valuable, if that makes sense. **Frederick Vallaeys:** Yeah, exactly. And that's what I think we're seeing in our company too. And I'm curious what it looks like in an agency landscape like yours. But the idea is that your most valuable people become even more valuable, and so you can start charging these higher rates, and they bring more utility to your company. Whereas the more junior people—what they do for your company can now be done with a fairly high rate of accuracy by some AI system, and that AI system can do it 10 times as fast or 100 times as fast. And if you assume that whatever work the junior people did had to be reviewed by senior people, well, why not just have the senior people review what Claude or GPT or Gemini was doing? So that kind of makes a challenge for entry-level folks. They may not be wanted anymore. So what do they do? But I'm curious, Anthony—how do you think about hiring in your agency? Has it shifted the types of people that you look for given what is now possible with AI? **Anthony Higman:** Maybe unintentionally, actually. So I used to be of the school of thought—and we're a startup, so it's slightly different. Bootstrapped up until this point. I started it with literally nothing. And so because of that, because we were bootstrapped, I had to hire less experienced people in PPC and then train them up. So my initial kind of hiring was "let me find really smart young people who just got out of college who kind of just have the right mindset, are curious, want to know why things are the way that they are," and hire them and teach them everything I know. And I have recently shifted towards I want more experienced people. So maybe unintentionally I've made that shift based on everything that's going on. So we are looking to hire again. I want that kind of person—someone that wants to peel the onion basically. What's under this? Why is it like that? This doesn't make sense. That's kind of the main skill I'm looking for. But I am hiring more advanced people. But to your point—well, what do all those people do then? This is also why AI just gets me super heated. It's like, okay, so we're just going to put all of these people out of business—for what? I guess it's for advancement. It just—there's so many bugs that I don't see the advancement to where I think it needs to be. Yeah, I agree with you. It does maybe do the 80/20, but I'm more a details person. So I care about that other 20% and why it's messed up more so than "the 80% is fine." **Frederick Vallaeys:** Yeah. I mean, my view on that is some of the things that we asked AI to do even a year ago—it'd be sort of buggy or it'd be slow or it'd be leaving huge gaps in its answers. Those same challenges you give it now, and it's orders of magnitude better. Is it perfect? No. It's still doing some things wrong. But just the pace of acceleration—and me living here in Silicon Valley, I talk fairly frequently to engineers at these companies. And obviously they have an incentive to hype up the company. But at the same time, you do hear that they are using these systems themselves to build the next version of the LLM. And so now we're talking about AGI but also super-intelligent AI. And so these things are perhaps just a couple of months away of that next iteration where it just blows our mind again at something we didn't think it could do before. And now it can. So that's why I'm scared and excited at the same time. And I'm an optimist. Hey, but Anthony, you said vibe coding, right? So let's maybe go into how you've experienced using some AI. Vibe coding you say kind of gets to 60% then it gets stuck. What have you found? What have you tried building, and where has the frustration been that it hasn't kind of gotten you to the place where you needed it to be? **Anthony Higman:** So personally, I have not really vibe-coded anything. Now, I do use all these things because I'm a curious person by nature, and I'm not naive. I know it's the goal and the push. So I do want to be knowledgeable on the subject. Every morning, I used to—for the last 14 years—do a search every morning that was "Google Google Ads" and just read every article on the first 10, 30 pages. Now I've added AI to that. So I do an AI search every morning, and I read all the articles—just know what's happening in the landscape. There's definitely cool things I do see. I'm not—while I'm very anti-AI, there are definitely things that I've seen value in. When we need to change something on our website or something and I can't figure it out, if I go to AI and say what the thing is, what the problem is that I'm experiencing, "How do I solve this?" I think that's kind of cool. It does make things easier that would have taken days and developers and stuff. But that also makes me angry because I'm like, what about the developer? But we are using them. I do find certain things about it cool. I'm extremely interested in it for the things that matter to me—how do we get into it for "best personal injury lawyer Philadelphia"? Where's it going and why? I forget exactly what your exact question was, but what I don't fully grasp yet is where it's going and why for search specifically. To me, I get the value in it. ChatGPT—you can type in a problem that you could not figure out before, and it'll walk you through the steps or give you a code that you can put in there. For search, why I'm interested in it is how do you get into "best personal injury lawyer near me"? But what I kind of see is an interesting bend in the road here. There's these incumbents like Google, and then there's a new guy, ChatGPT. And I see it emerging—Google kind of being more local, and ChatGPT sort of being the go-to for everybody's other stuff. Or Claude. I've not used Claude. I'll be honest. But I know that's the hot one on the street right now, especially for vibe coding stuff. So yeah, that's kind of my experience. What about you? I know you're an AI optimist. I know you do Optmyzr. Are you using this to vibe code stuff? Are you making progress? **Frederick Vallaeys:** Oh my god. Oh my god. Way more time than you can imagine, Anthony. No, it's so exciting. It's—listen, but I think you brought up an interesting point, and then I'll answer your question to me. So I go and give a lot of presentations at PPC conferences, and one thing that I've sort of been challenged by is the hot thing and the thing that's really changing how business operates and how society will eventually operate is AI. And vibe coding is one part of that. But how do you bring that back to a marketing story? Because oftentimes the audiences want to talk about, "Well, I've been managing keywords and I've been managing bids. So tell me how I do these things with this AI." But maybe that's the wrong question. Do we still need to manage bids? Do we still need to worry about keywords? There's this whole shift that's happening. And so then it becomes much more to me about: "Well, what I've been spending all this time managing keywords and managing bids, but if I didn't have to do those things and I could actually be a marketer, what would I want to do?" And so around the holidays, I was like, "Hey, well, these Advent calendars seem to be gaining momentum in the United States." And for people who don't know what an Advent calendar is, basically you open one panel on a little paper box every day, 24 days leading into Christmas, and there's a little treat in there. And I was like, "Well, why don't we make something like that for PPC advice, and we'll give away some books and we'll give away some conference passes?" And that would have been a big project. But now I vibe-coded it, and so it took about two weeks, and I vibe-coded something that worked pretty well. And it was a different way to connect with my audience. And then the other thing on the back end—now maybe you get new leads and you get new prospects. Well, how do you engage them better? So one thing we struggle with sometimes at Optmyzr is "how do you use the system?" It's this massive toolbox with so many capabilities. Where do you start? And so one thing we've always had was a training course on Udemy—a bunch of videos. Now, you have to sit through two hours of videos to become educated. Now, Udemy got acquired, and we had to shift our course. So I was like, "Well, why should I invest in a place where I have to pay? I'm just going to vibe code a lightweight clone of Udemy." And by the way, I'm not saying I rebuilt Udemy. But I don't need Udemy. If you really think about what is it you need in your business, it's these two features from Udemy, and those are pretty easy to replicate. And now that I've got those two features, now I can start thinking about: if Udemy did exactly what I needed it to do, which is "don't force people to watch two hours of content, but let's figure out who is Anthony, who is Fred, what do they know?" And out of these two hours of content, what would they actually benefit from enhancing their knowledge? And now maybe I can give you 17 minutes of content that actually matters to you, the new stuff. And that's where I'm able to push things. And so again, but it's weird because this is not Google Ads PPC. But this is beneficial to people trying to give new solutions to people. **Anthony Higman:** Yeah, it's a really good point, actually. Thank you. Because another thing that I continue to get angry and rail on is the thing that you just brought up—the messaging, which I often phrase as propaganda in some regards, has always been, "Okay, well, we're going to save you more time to go focus on the other things that you care about." And my whole thing is, "What things?" But you just gave two very insightful, actionable things that you did with more time. So I appreciate that because I keep asking people this question: "What are you going to do with your more time?" But I think I discount—I'm always kind of thinking like a marketer. Yes, we manually bid, and yes, we still look at search terms every day and do old-school stuff. But I think it is important to have things that are real on that messaging point. And I haven't really been able to find a ton when I ask people that. So I appreciate it. **Frederick Vallaeys:** Well, and it's—I am very disappointed usually in people when I give them a vibe coding tool or they start vibe coding, but they kind of stop at having rebuilt the previous thing and then they just can't imagine what is next. And it's the whole Steve Jobs thing—his customers couldn't envision what they really wanted. And so it takes that creativity. And I think too many of us have just not grown up being asked to be creative because it becomes about, "Well, hey, you want to work in digital marketing, so you need to know what buttons to push in Google Ads to do the thing that's necessary." But those buttons have just been the interface layer that connected you to the underlying system. And that underlying system is the one you want to manipulate. And by the way, if you had access to the underlying system and not just the five buttons—and there's more than five buttons, but the five buttons that Google decides to give you—imagine what you could do. And this even comes out in: if you have access to the API for Google Ads, you can generate some amazing reports that go far deeper than what they put in the UI because you're able to piece it together. You could literally be like, "For each keyword, a specific ad combination, how is that performing?" So let's not worry about ad group performance. Let's literally go down to the keyword-plus-ad level. You can pull that report, but Google doesn't expose it because it's too much data. It's going to confuse people. But if you're creative in this field and you ask that question, you can now build it very easily. And that's what excites me. But I also need people to really think beyond what they've historically done. **Anthony Higman:** Yeah. Yeah. I mean, I do appreciate the perspective. And I think that maybe it does go that way with people thinking up new and creative things. I have a question for you, though, actually. Back to AI—and I'm very much enjoying this conversation because you're giving me arguments to my points that are things that I always get angry about. But in my opinion, a lot of stuff that gets automated, you pay less and less attention to it over time. Everybody's got an AI note-taker now. And I think that's cool because when something comes up, you can go back and reference it. But I think that the amount of time that people actually go back to their AI note-taker dwindles down. I think what gets automated, less attention gets paid to it. And so smart bidding came out. That was the push by Google. "We're going to save you more time. You don't have to individually bid." And it's a great message—"Spend less time looking at bids." Back to the point: I think what gets automated, people stop paying attention to it. And I'm not saying necessarily that all big tech companies are Machiavellian or have the wrong intentions. But they're a business, and their goal is to make money. And I think there's more BS that gets pushed in when things stop being paid attention to. So I think more crap searches get put in, and maybe you could call it manipulation happens when you're not looking at the thing that's really important anymore because you're off coming up with new ideas. So I always kind of have this in the back of my head: "What's their motivation, and why am I seeing certain things that I'm seeing now that seem to be way off base more so than they ever have been in terms of what's matching?" And you could argue it's because of AI and conversational matches and different moves by Google. To be clear, I don't think Google needs to have Google Ads not work, or they don't have customers. I'm just saying these little tiny things at scale globally can make them a lot of money when you're not paying attention. **Frederick Vallaeys:** Yeah. I mean, shaking the cushions, right? **Anthony Higman:** There you go. Yes. **Frederick Vallaeys:** So yeah, I don't disagree with that, and I wish I could counter it. But I think you're right. So there's two things you said. One of them is people stop paying attention as things become more automated, and that's kind of on people then. You could have still paid attention, but you consciously decided that it was not worth your time to do that. So you made some sort of economic value proposition in your head that was like, "Well, okay, maybe they are shaking the cushions, but it's not worth my time that I could spend on doing something else to actually go after that." That's one. Now, the other is the obfuscation of the data. So even if I wanted to go and validate what they were doing, I just can't. Because with prompt-based advertising, it's not tying back to a keyword. And like you said, it's privacy. So they're never going to give me the full history of why that ad was shown. So I just have to at some level trust the system. And that is hard because things are going to happen. And yes, you're right. Google is—not just Google, but all of these companies—they're in the business of making money, making profits. So yes, they may do things that are not ideal for an individual customer. And generally, they will have to do the right thing because if people become highly displeased with Google Ads, they might find alternatives—assuming alternatives exist. And that's why I'm happy that ChatGPT, for example, is a viable alternative in how people find things, and soon we'll also be able to place ads there. But yeah, that's sort of the other half of that. So it's the 80/20. As long as the 80% keeps growing, the 20% is not that important—if people choose it's not. **Anthony Higman:** Yeah. You're not wrong. **Frederick Vallaeys:** Hey, and you brought up the note-takers, right? So everybody's got these note-takers. So but nobody goes back to the notes, you said. So but one thing I've started thinking about is my meetings as prompts. So historically, the way people—you go into a meeting, and everyone comes out with more tasks and more work. But what if you think about those meetings as "the AI is in the room because it's all being transcribed"? And what if you could really specify "what are the criteria, what are the restrictions, what are the goals that we have" and come out with all of those things that you would have otherwise given to a large language model prompt? And then at the end of that meeting, you give that whole transcript to your LLM, and you say, "Go and do the things that needed to be done." And because you've been intentional about loading it with those questions and answering those questions, the LLM can actually do a better job. And so one thing that I specifically do in this regard is I often come out of meetings, and I give the transcript to the LLM, and I say, "What could I blog about this? What are some ideas? And even if this was an internal meeting, what are some of the bigger ideas that could maybe benefit the industry as a whole, knowing how we think about these things?" And so, but it's these new things. Meetings have never been seen as a marketing and a messaging vehicle, but now you can add that on. And then that's just one of the many. So again, and that's where I think people need to be creative and think of that next thing that they've always wanted to do and maybe just didn't have the time to do. **Anthony Higman:** That's cool. You know what I do? I'll tell you a funny thing. So I don't know if you have this experience, but the AI note-taker always shows up before the person does. And so I go, "Hey, AI note-taker, fuck you." That's in the beginning of the transcription. But again, this is another thing with weird data points and what that does to different things. I think this is going to be this bug that happens for the next 10 years until they figure out how to make certain data points important and certain data points not important. **Frederick Vallaeys:** So when the AI gains consciousness and manifests itself in the form of robots, you know who the first person is that it's going to go after? **Anthony Higman:** It's coming after me for sure. **Frederick Vallaeys:** Okay. Well, Anthony, this has been great—fun conversation. Thanks for sharing kind of how you think about all of these things and the frustration points and how to—yeah, we operate in a world that can be frustrating, but it's the sandbox we play in. So thanks for sharing all of those insights. **Anthony Higman:** Absolutely. **Frederick Vallaeys:** If people want to know more about you, what's your favorite place to hang out, and how can people get in touch with Adsquire? **Anthony Higman:** Yeah. So I'm active on Twitter and LinkedIn are kind of my main platforms. My name is just Anthony Higman on any social channels. They can go to adsquire.com. **Frederick Vallaeys:** Cool. Well, hey, thanks for sharing. And thanks, everyone, for watching today's episode. Please give it a thumbs up, subscribe, and that way you'll find out when the next episode comes out. We also do read the comments. So if there's any counterpoints or additional points to what Anthony or myself said, please put those in, and we'll do our best to respond to all of them. But with that, Anthony, thank you. Everyone, thanks for watching, and we'll see you for the next one. --- *Source: [AI in PPC: Strong Case… But Some Weak Matches](https://www.optmyzr.com/ppctownhall/ai-in-ppc-legal-automation-risks/)* *© Optmyzr. All rights reserved.*