Key Takeaways
- B2B buyers are turning to AI assistants like ChatGPT and Claude to find solutions, forcing companies to shift from traditional SEO to Answer Engine Optimization (AEO) by publishing substantive answers to real customer questions.
- Traditional CMS processes create severe bottlenecks, but companies can modernize their websites incrementally by establishing an agreed-upon verbal identity and using AI-connected workflows.
Your buyer’s first move isn’t a Google search anymore. It’s a conversation as they open ChatGPT of Claude, describe their problem in plain language, and ask an AI assistant to help them figure out who can solve it.
The old front door, a website optimized for keywords and ranked by links, still matters. But a new one has opened, and most companies don’t yet know they’re standing behind it. The question that should keep you up at night: when a buyer asks their assistant about a company like yours, what story does it tell? PitchKitchen’s report, The Truth About Your Website went looking for evidence of the gap.
PitchKitchen founder Greg Rosner makes the case for answer engine optimization (AEO), publishing substantive answers to the real questions your customers ask so AI engines have something useful to cite, explains the Magnetic Messaging Framework and walks through the AI Brand Twin publishing workflow, a way for authorized people to update the website through conversation with AI instead of waiting weeks in an email queue.
Demand Gen Report (DGR): B2B buyers now open a chat window before a search box. For a marketing leader who’s spent a decade optimizing for Google, what’s the first thing they should stop doing, and the first thing they should start?
Greg Rosner: Stop assuming the buyer meets you on your website. They may meet you in a conversation you’re not in.
Think about what people are asking their AI assistants already. Help me fix the washing machine. Help me write a better proposal. Help me think through this decision. When those conversations keep being useful, my view is that the next question becomes easier to ask: “Who can help our company solve this?”
That’s a different front door.
An April 2026 survey by Clear Spark Digital found that 68% respondents trusted AI search results as much as or more than traditional Google results. It was an opt-in panel, not a representative sample of all Americans. But that’s meaningful trust for marketers to pay attention to.
I believe increasingly capable, personalized AI advisors will make this way of buying more common. The buyer can explain their situation, their constraints, what they’ve tried, and ask the assistant to help them find a fit. Give that conversation something useful to work with.
That’s where answer engine optimization (AEO) comes in. Publish answers to the questions your customers actually ask. Who is this for? Why choose this approach? What does implementation involve? What can you prove? Fifty good FAQ answers can cover a lot of ground. Fifty versions of “we deliver innovative solutions” give the buyer very little. The number isn’t the point. The substance is.
DGR: The report states that lost deals never show up in analytics. No bounce, no abandoned form, nothing. How would you coach a CMO to measure a loss that leaves no trace?
Rosner: You can’t put a lost deal in a spreadsheet if you never knew it existed. But you can look for the gaps. We use Peec AI and Searchable to see how a brand appears in answers to selected questions, which pages get cited and where competitors are supplying the information. That’s intelligence you can use.
Take the questions your sales team hears every week. The early questions, when a buyer is trying to understand the problem. The comparison questions. The “why should we choose you?” questions. Track those.
Then read the answers. If your company solves the problem but your website barely explains how, you’ve found work worth doing. Publish a useful answer, put evidence behind it and watch what changes.
The tools sample AI responses. They don’t let us listen to every private buyer conversation, and they don’t tell us we lost a million dollars because a page wasn’t cited. I wouldn’t pretend otherwise. Use them to make better publishing decisions. Then connect what you learn with qualified inquiries, sales conversations and a very ordinary question to the buyer: “How did you find us?” You won’t see everything. You’ll see more than you would by staring at website traffic alone.
What the Domain Scan Actually Measured, and What It Didn’t
DGR: You argue ChatGPT is the new word of mouth, and it repeats what it reads. What proof did you gather that the engines actually describe the stale version of a company, versus simply skipping it?
Rosner: We didn’t measure the engines’ answers in this study; we measured the websites and I want to be clear about that. We scanned 32,574 domains. Of the 20,496 homepages with a visible copyright year, 5,022 still said 2025 or earlier. We also looked at signals such as the platform and whether the homepage linked to a blog or insights section. That does not prove an AI engine repeated an outdated story. It does not prove the engine skipped the company. Those are separate questions, and they require a separate study.
The footer is the symptom that makes you stop and look. You can have an old footer and excellent content. You can also have a footer that updates itself every January while the story above it hasn’t moved in years. Changing the date takes a minute. Understanding what’s happening behind it takes a conversation.
In my client work, that conversation often reveals a business that has moved a long way and a website that hasn’t come with it. That’s the concern behind “the new word of mouth.” What information is available when a buyer asks an assistant about your company? Does it explain the business you run today?
The scan gives us a reason to investigate. It doesn’t give us permission to claim we measured something we didn’t.
DGR: You note the AI picture “updates slowly and it compounds.” How long does a company have before a frozen story becomes the story the engines default to permanently?
Rosner: I wouldn’t say permanent. We haven’t established a countdown.
The problem I see is happening right now. Companies are using AI to change how they work, how they deliver, even what they sell, and they’re making real progress inside the business while the website is still explaining the old offering. Then somebody adds “AI-powered” to the headline. Job done.
Except the buyer still has no idea what changed for them. I call it AI Parmesan: sprinkle AI over the old copy and expect everyone to taste the value. What can I do now that I couldn’t do before? How much faster? What costs less? What gets better? Tell me that.
If you’ve built a substantially better solution, leaving the old explanation in place makes it harder for buyers and their assistants to evaluate what you actually offer. You did the work. You made the investment. Now explain the difference. Our scan doesn’t tell us when an engine will pick up a change. The decision you control is when you publish an accurate story and how you keep it accurate.
Start there. The gap is already costing you clarity.
DGR: Smaller firms froze more often: 27.5% stale at 11 to 50 employees versus 22.2% at 51 to 200. Is this a budget problem, a bandwidth problem or a leadership problem, and does the fix change depending on the answer?
Rosner: The numbers don’t tell us the cause. My client work gives me examples of what can sit behind them.
One customer invested $3 million in an intelligent compliance system. The website hadn’t changed in a year. That capability was nowhere to be found. Three million dollars of progress. The buyer couldn’t see it.
Another customer, in freight management, had left the core website unchanged for two and a half years. There were seven blog posts, produced manually, while there was a whole world of questions freight buyers were asking about their problems, their options and who they should choose.
I hear the frustration from CEOs. They know the business has changed. They ask marketing to explain it. And now it’s another project on a list that was already too long.
That’s where we have to look: Is leadership clear on the story? Is somebody available to supply the facts? Does everything wait for one person’s approval? Or is the team spending its time manually producing work AI could help it create?
The fix depends on the answer. Give people a clear business context, access to the expertise inside the company and a sensible way to create, review and publish. There’s a lot more they can do. But the technology becoming available doesn’t mean the company changes its habits. People can keep working the old way for years.
That’s why I expect this problem to stick around. The capability moves fast. The decision to use it well can take much longer.
Does Switching Your CMS Fix Stale Content? Not Usually
DGR: WordPress carries the same share of stale sites as fresh ones, 53.1% versus 55.8%. Wix and Squarespace top the frozen list but represent a tiny slice of B2B. What should a leader take from that, and what shouldn’t they overread into it?
Rosner: Don’t read this report and immediately go shopping for another CMS. WordPress shows up heavily on both sides. These numbers don’t establish that a platform change will fix your publishing habits, and the Wix and Squarespace results come from a much smaller part of the sample.
Ask a more useful question: What happens here when we need to change a page? Follow the work. Does someone send an email, wait for a draft, send it back, approve another draft and then hand it to somebody else to build? That process deserves your attention.
A different architecture may remove some of those handoffs. Headless can separate the content system from how the website is presented, but it can still use a CMS. The word “headless” doesn’t make a team fast. The test is whether an authorized person can get accurate, useful information onto the site without an unnecessary production queue.
New platform. Same vague story. Same waiting. You haven’t solved much.
DGR: You call hand-typing pages into a CMS “the biggest bottleneck a business website has in 2027.” For teams juggling seventeen agencies and five sales stories, what breaks that bottleneck without adding headcount?
Rosner: First, get the people telling five different stories to agree on the business. Otherwise, you’re about to automate the confusion. AI has driven the cost of producing beautiful pages down dramatically. The difficult part is still giving the right customer a reason to care, a reason to choose you and a useful next step.
At PitchKitchen, we extract that truth with leadership and capture it in the Magnetic Messaging Framework. Then we make the context available through an AI Brand Twin and connect it to how the company creates and publishes.
Here’s what that should feel like. A CEO with publishing authority opens their AI assistant and says, “On the public relations page, change the explanation to reflect the service we now offer.” The system works from the approved story. The CEO reviews the change. It goes live through that same conversation.
Why should that require a week of emails?
You still need factual checks, clear permissions and a way to reverse a change. Build those into the workflow. Then let authorized people do the work. The CEO shouldn’t have to beg the company’s own website to catch up with the company. That’s the shift. Your website should move at the pace of your business.
DGR: The Magnetic Messaging Framework runs 40 to 50 pages, and you call it the training data behind every AI in the building. What’s the biggest mistake leaders make when they try to write that document themselves?
Rosner: They start writing before they’ve agreed on what’s true. I can ask five leaders in the same company what the business does and get five different answers. Put all five in a document and you haven’t created clarity. You’ve recorded the disagreement.
This is why outside facilitation matters. Who is the customer? What are they struggling with? Why should they change? Why now? Why should they choose this solution, from this company? Stay with those questions until the answers mean something.
Every business needs a verbal identity alongside its visual identity. The Magnetic Messaging Framework captures that verbal identity and the evidence behind it. It’s the spine of the story, from who this is for to why they should care. Forty pages or fifty pages isn’t the achievement. Agreement is. Clarity is.
When I call it “training data,” I’m talking about the approved context your AI tools work from, rather than necessarily retraining a model. We believe that context should be available through an MCP server, using a connection standard that lets compatible AI tools access resources and tools. Then an authorized person creating a proposal, customer email, presentation or website page can work from the same current understanding of the business.
Call it a blueprint. Call it the business bible. Work with us or build it yourselves. But create it. Keep it current. Check what AI produces from it. And remember that permission to read the story is separate from permission to publish on the company’s behalf.
How the “Strangler Fig” Migration Modernizes a Site Page by Page
DGR: Your “strangler fig” migration grows the new site around the old one, page by page, with the engines reading fresh work within weeks instead of six months. What kind of company should not attempt this in-house, and where does it most often go wrong?
Rosner: The name sounds more alarming than the approach. You look at the website, page by page, and decide what should move into an AI-connected publishing workflow and what needs to stay in the CMS or another system.
You don’t have to rip everything out. The first conversation is about authority. Who can create? Who can approve? Who can publish, and where?
A company might currently give three people access to its CMS. Keep that authority if it works. Or expand it deliberately. Let the public relations person own the newsroom. Let a salesperson publish useful answers in an approved part of the site. Salespeople publish LinkedIn articles already. Why couldn’t the right people contribute to the company’s website too?
Connect their AI tools to the approved brand framework through MCP and give designated publishers access to a publishing system that enforces their permissions. The connector to the story doesn’t, by itself, grant publishing authority. Within that authority, routine content shouldn’t need to climb several approval ladders. Pricing changes, sensitive claims and pages tied to critical business systems can keep additional review. Make the rules clear. Then let people move.
We migrate suitable pages incrementally, check the changes and keep version history so we can correct or withdraw them promptly. Something already cached or quoted elsewhere may remain there, so review still matters. And publishing today doesn’t guarantee an engine will index or cite the page tomorrow. If nobody in-house can own the permissions, integrations and recovery process, get implementation help.
This should feel methodical. One page, checked and working. Then the next. You’re building the ability to keep improving the website. You can start without turning the whole business upside down.
DGR: What closing message would you leave CEOs with about the gap between how their business has evolved and how the market actually sees it?
Rosner: You can build a better business and still look like the old one.
A better product. A service you’ve transformed. A problem you can now solve in a fraction of the time. You know what changed because you’re inside the company every day, but the buyer looking at your website sees none of it, and the assistant helping that buyer has an incomplete story to work from.
That should bother a CEO; It bothers me.
The stale footer is where we started looking. What I want leaders to examine is the distance between the business they’re running and the business the market can understand. The familiar response is another redesign. Better photographs, a fresh template, the same vague promises underneath. AI has made beautiful website production dramatically faster and cheaper. The difficult work is still giving the right customer a compelling reason to choose you.
You have two jobs. Get the story clear. Modernize how you publish it. Start with either. Both have to get done.
A brilliant story sitting in a document won’t help a buyer who never encounters it. A fast-publishing system can put a lot of meaningless content into the world. At PitchKitchen, we get leadership aligned on the truth, capture it and make it available through an AI Brand Twin. We grow the new website around the existing one and give authorized people a way to publish through conversation with AI.
Now the knowledge inside the business has somewhere to go. The question sales answered yesterday can become a useful page. The capability you invested in can be explained. Your proposals, emails and presentations can draw from the same agreed story.
That’s a capability a company can use every day. I believe the competitive gap will widen between companies that build it and companies that keep every change waiting in a queue. While you’re passing the draft around, someone else can already be helping the buyer understand what to do next.
You don’t need to wait for a six-month rebuild to begin. Your business has moved. Let the market see it.
