If you’ve used ChatGPT, Perplexity, or Google’s AI search lately, you’ve probably noticed something: you’re asking questions differently than you used to.
Instead of typing “best accounting firm Boston,” you might ask “I need help with R&D tax credits for a manufacturing company in Massachusetts. What should I look for in an accountant?”
Your potential clients are doing the same thing. And that shift is changing how businesses that have spent years optimizing their websites for traditional Google search have to approach their content.
I recently attended a webinar called “Finding the Funnel in the LLM Era” that laid out how dramatically things are changing. In it, we explored available research on AI search, with some patterns emerging that are worth paying attention to now.
The Old Way vs. The New Way
For the past two decades, the customer journey looked pretty straightforward. Someone searched for a keyword on Google, clicked through to a website, spent some time learning about the company, maybe signed up for a newsletter, and eventually became a client. We could track all of that. We knew what worked.
Now, AI tools are inserting themselves into that journey in a way that changes everything. When someone asks ChatGPT or Perplexity for help, the AI doesn’t just send them to websites. It summarizes information, compares options, evaluates pros and cons, and makes recommendations. All before the person ever clicks on anything.
That means potential clients are forming opinions about your firm based on what an AI tells them, not just what’s on your website. And unlike Google, where you could see what keywords people searched for and track your rankings, AI search is mostly a black box.
What’s Actually Different?
The webinar introduced something they called “Lab Data vs. Field Data.” Here’s how they break down:
Lab data is when you test out prompts yourself to see what AI tools say about your business. You type in “best construction law firms in Denver” or “fractional CFO for tech startups” and see if your firm shows up, how you’re described, and whether the AI recommends you. That’s useful information, but it’s limited. It only tells you what’s possible, not what real people are actually doing.
Field data is tracking what happens when real users interact with AI and then click through to websites. That’s harder to get right now, but it’s the data that actually matters for understanding business impact.
The challenge is that most businesses don’t have good systems for tracking either one yet. We’re all learning as we go.
Four Layers of Your Brand Identity
Here’s something that really stood out from the research. The presenters talked about how AI tools pull information from four different “layers” when they talk about your business:
Your Known Brand is the official stuff:
- Your website content and About page
- Press releases and official announcements
- LinkedIn company page
- Any content you directly control and publish
- This is your cleanest, most polished representation
Your Latent Brand is everything other people say about you:
- Reviews on Google and Yelp
- Discussions on Reddit or industry forums
- Social media mentions
- Comments on LinkedIn posts
- This is often more influential than your official content because it feels more authentic
Your Shadow Brand is the uncomfortable stuff:
- Lawsuits and legal disputes
- Customer complaints
- Negative reviews
- Rumors or controversy
- Most businesses try to ignore this layer, but AI tools don’t. They pull from everything.
Your AI-Narrated Brand is how all of these layers get blended together by AI tools when someone asks about you. This is the reputation you don’t fully control, built by algorithms that weigh all those other layers in ways we can’t always predict.
Each layer influences what shows up in the AI-narrated version. If your Known Brand is strong but your Latent Brand is full of unanswered complaints, the AI will reflect that tension when someone asks about you.
Why Personas Matter More Than Keywords
In traditional SEO, you’d target specific keyword phrases. If you were a tax attorney, you’d optimize for “estate tax planning attorney” and track your ranking for that exact phrase. But people don’t talk to AI that way.
They ask things like “My father passed away and left property in three states. I’m the executor and I’m overwhelmed. What kind of attorney do I need and what should I ask them in the first meeting?” That’s not a keyword. That’s a whole scenario with context, emotion, and multiple decision factors baked in.
The webinar emphasized that businesses need to think in terms of personas and intent stages rather than keyword variations. Who is asking the question? What problem are they trying to solve? How ready are they to hire someone? What concerns are driving their search?
An accounting firm might need to think about “business owner worried about an IRS audit” as a distinct persona from “startup founder looking for first-time tax filing help.” Both need accounting services, but they’re asking different questions with different levels of urgency and sophistication.
What You Can Start Doing Now
The webinar gave some practical advice that doesn’t require waiting for new tools or analytics platforms. Start by collecting the actual questions your potential clients are asking:
- Look at your Google Search Console data to see what queries are bringing people to your site
- Read through contact form submissions and intake forms
- Listen to what prospects say on initial phone calls
- Check relevant Reddit threads and industry forums where your target clients hang out
- Review support tickets and common questions from existing clients
Those real questions become your prompt library. You’re looking for patterns in:
- How people describe their problems
- What alternatives they’re considering
- What concerns they express
- What outcomes they’re hoping for
Once you have 20 to 50 real prompts, start testing them. Type them into ChatGPT, Perplexity, and Google’s AI search. See what comes back:
- Does your firm get mentioned?
- How are you described?
- What competitors show up?
- What information is accurate or misleading?
This isn’t about gaming the system or trying to manipulate AI tools. It’s about understanding how you’re being represented when you’re not in the room. If an AI is telling potential clients things about your firm that are outdated or inaccurate, you need to know that.
The Sentiment Factor
Here’s something unexpected from the research. The way someone phrases their question influences the tone of the AI’s response. If someone asks with anxiety or fear, the AI’s answer tends to mirror that emotion. If they ask with excitement or optimism, the response leans more positive.
For professional services firms, this matters. Someone asking “how do I avoid getting sued by a contractor” is in a different emotional state than someone asking “how do I set up vendor contracts for my new business.” The first person is afraid. The second person is optimistic. And the AI will reflect that back in how it presents information, including how it talks about potential solutions and service providers.
Understanding the sentiment behind common questions in your industry helps you think about what emotional state potential clients are in when they’re learning about your firm through AI tools.
Where This Is All Heading
Nobody has this figured out yet. The webinar presenter was clear about that. We’re all watching this shift happen in real time and trying to understand what it means. Major brands with big marketing teams are experimenting with prompt tracking and AI visibility monitoring, but there aren’t established best practices yet.
What we do know is that optimizing for AI search isn’t the same as optimizing for traditional SEO. You can’t just stuff keywords into your content and expect to rank well when there are no rankings to track. The signals are different. The data is murkier. The whole concept of “visibility” has changed.
For professional services firms and B2B companies, the immediate opportunity is to make sure the information that AI tools can access about your business is accurate, comprehensive, and representative of your actual expertise. That means:
- Having authoritative content on your website that clearly explains who you help, what problems you solve, and what makes your approach different
- Actively managing your online reputation across all those layers (the Known Brand, the Latent Brand, and yes, even the Shadow Brand)
- Paying attention to the questions your prospects are actually asking
- Making sure your content addresses those specific scenarios with depth and nuance
Generic service descriptions aren’t enough anymore. People and AI tools are looking for answers to specific, contextual problems.
What We’re Doing at Scribendi
We’re not just monitoring this shift – we’re actively adapting our approach for clients right now. Here’s what that looks like in practice:
We’re using tools like AlsoAsked to map out the actual questions people are asking in each industry and practice area, then building comprehensive FAQ pages that address those specific questions. For a law firm specializing in construction law, that might mean creating an FAQ page that answers “What happens if a subcontractor files a lien on my property” or “How do I enforce a payment bond claim.” For an accounting firm, it might be “What tax credits am I missing as a manufacturer” or “How do I handle sales tax for multi-state operations.”
We’re combining data from Google Search Console with research from AlsoAsked, Reddit threads, and industry forums to write blog content that answers the real questions prospects are typing into search bars and AI tools. Instead of generic posts about “tax planning tips,” we’re creating content that addresses specific scenarios like “I just sold a rental property in another state – what are my tax obligations” or “My S-corp is growing faster than expected – when should I switch to C-corp status.”
We’re testing prompts for our own business and for clients to see how AI tools are describing their services, then updating website content to make sure those AI responses are accurate and comprehensive. When we find gaps or outdated information being pulled by AI platforms, we know exactly what content needs to be created or refreshed.
We’re also building content that’s specifically structured to be useful for AI tools – clear problem/solution frameworks, detailed explanations of processes, genuine expertise that goes beyond surface-level advice. The kind of content that AI tools can confidently cite because it’s authoritative and helpful.
This isn’t theoretical. These are adjustments we’re making to client websites this month. The businesses that will do well are the ones that stay curious, keep testing, and focus on being genuinely helpful rather than trying to game a system we don’t fully understand yet.
If you want to talk through what this means for your specific business, we’re happy to share what we’re learning. Nobody has a complete playbook yet, but we can work through it together.