- Websites are not going away, but their role is changing as more customers get answers before they click.
- Generic content is losing value. Original, specific and experience-based answers are becoming more important.
- Most businesses do not have a shortage of expertise. They have a knowledge capture problem.
- Real customer conversations reveal questions, objections and content gaps that prompt tracking alone cannot reliably uncover.
- The businesses best prepared for AI search and AI agents will capture, organise and publish better answers.
For the last 15 years, most businesses have treated their website as the centre of their digital marketing universe.
Google Ads sent traffic to it. SEO was designed to rank it. Social posts linked to it. Email campaigns drove people back to it. Forms, pop-ups, call-to-action buttons and sticky headers were all designed around one simple idea:
Get people to the website, then convert them into leads.
That model has worked well. And to be clear, websites are not going away. But their role is changing fast.
We are moving from a search-first internet to an answer-first internet. We are also beginning to move into an agent-first internet, where customers do not just ask AI for information. They ask it to research, compare, shortlist, monitor, book, buy and take action on their behalf.
That shift has massive implications for every service business, especially local businesses and those that rely on being found when a customer has a problem, a question or a need.
In the old model, your website needed to rank. In the new model, your business also needs to be clearly understood, supported by credible information and suitable for AI to include in an answer or recommendation.
That requires a different kind of content. Not generic blog posts. Not fluffy "top tips" articles. Not keyword-stuffed service pages. Not a flood of AI-generated filler.
The businesses that win in this next phase will be those that answer real customer questions better than anyone else and make it easy for people and AI agents to take the next step.
The old website model was built around traffic
Until recently, the customer journey was predictable. Someone needed a service, so they went to Google and searched for something like "plumber near me", "heat pump repair Auckland" or "commercial cleaning Christchurch". Google returned a list. The customer clicked a few links, read reviews, compared options, and then either called, filled in a form, or left.
That is why everyone became obsessed with website traffic and conversion rates. If 1,000 people visited your website and 5% converted, you got 50 leads. So the focus became: can we rank higher, get more clicks, improve the landing page, shorten the form and lift conversion from 5% to 7%?
That thinking still matters. A good website still builds trust. A good landing page still converts. Fast page load and mobile usability still matter. But traffic is no longer the whole game because more customers are getting useful answers before they ever reach your website.
Search is becoming a conversation
Traditional search was largely based on keywords: short, clipped phrases such as "roof repair cost" or "blocked drain plumber". AI search is different. People no longer need to reduce their problem to two or three keywords. They can ask full questions in plain English and continue with follow-up questions that retain the context of the conversation. Google is already building this conversational back-and-forth directly into Search.
Instead of searching "heat pump repair near me", they ask:
"My heat pump is running, but the room is not getting warm. What could be wrong, and do I need a technician?"
Instead of "gutter cleaning price", they ask:
"How much should gutter cleaning cost for a two-storey house, and why do some companies charge so much more than others?"
Instead of "mortgage adviser first home buyer", they ask:
"I'm self-employed and buying my first home in Tauranga. What should I know before applying for a mortgage?"
This is a completely different type of customer behaviour.
Customers are not always looking for a business first. They are looking for an answer.
They want context. They want to understand their options, what something might cost, whether they can solve the problem themselves, when they need professional help and who they can trust.
In this environment, your content is not simply competing for a position on a Google results page. It is competing to become useful source material for an answer.
The real opportunity is not more content
Many businesses will respond to this shift the wrong way. They will hear that AI needs context and assume the answer is to publish more blog posts.
So they produce another "10 tips for maintaining your home this winter" article or "5 reasons to choose a professional" post.
The problem is that this content is generic. It could belong to almost any business. It says what everyone else says. It rarely answers the specific, nuanced questions customers are actually asking, and it contains little original expertise.
The opportunity is not simply to create more content.
The opportunity is to create better source material.
The useful knowledge is already inside the business. It is with the owner, the sales team, the technicians and the customer service staff. It appears in quoting conversations, email replies, support tickets, form submissions and phone calls.
But most of that knowledge disappears. A customer asks a great question, someone answers it, the conversation ends and the insight is gone.
Most businesses do not have a shortage of expertise. They have a knowledge capture problem.
AEO still starts with strong website fundamentals
Answer Engine Optimisation does not replace SEO, and it does not require a secret technical trick. AI systems still need to discover, access and understand useful information.
Your pages should be crawlable, fast, mobile-friendly and clearly structured. Use descriptive headings. Answer the question early, then add the detail needed to support it. Make important information available as text rather than hiding it inside images, videos or downloadable files. Keep facts, prices, policies and service details current.
FAQs, comparison tables, process explanations and clearly labelled sections can all help people find the information they need. Appropriate structured data can also help search engines understand the page and qualify it for existing rich results.
But there is no special AEO schema that guarantees inclusion in AI answers. Google itself says there is no special structured data required for its generative AI search features. The fundamentals still matter because they make good content easier to find and interpret, not because schema is a magic AI visibility switch.
The problem with measuring AI visibility
In this new AI and AEO world, everyone is racing to measure "share of voice", "mention rate" and "AI visibility".
These metrics can be useful. If you test a consistent panel of prompts, platforms and locations over time, they may show whether your visibility is improving or declining.
But they are directional indicators, not an objective ranking position.
Change the wording of a prompt, add a location, describe a different type of customer or run the same test on another model, and you can receive a completely different set of recommendations. Research into commercial AI recommendations has found that even natural paraphrases of the same basic buying question can produce materially different brand results.
This does not make visibility tracking worthless. It means the methodology matters, and the result should not be mistaken for a complete picture of what your market is asking or how every AI system perceives your brand.
A prompt tracker measures the prompts you chose to test. It does not automatically reveal the questions your customers actually ask.
Stop guessing and start capturing
Most businesses think they know what their target audience wants to know. Sometimes they are right. Often they are only partly right.
We see this every day when reviewing Cleva.Bot transcripts. A business expects visitors to ask about one feature, but the repeated question is about eligibility. The website focuses on benefits, while customers keep asking what happens next. The service page explains what the company does, but visitors want to know whether it applies to their particular situation.
The best way to uncover these gaps is to capture questions wherever real customer conversations are already happening:
- Website chatbot transcripts
- Sales calls and consultation notes
- Support emails and helpdesk tickets
- Contact forms and enquiry forms
- On-site search queries
- Social media comments and direct messages
- Customer interviews and feedback
- Search Console query data
No single source gives you the complete picture. But together, they provide strong first-party evidence of what customers want to know, how they describe their problems and what prevents them from taking the next step.
Real customer questions are optimisation gold
Every day, your customers are telling you what content you should be creating. They ask questions before they buy, before they book and before they act. They tell you when they are confused, when they are comparing options, when they are worried about cost and when they are not even sure if they need help.
- A customer might ask: "Why does my drain keep blocking even after it has been cleared?"
- A customer might ask: "Can I repair one section of my roof, or do I need to replace the whole thing?"
- A customer might ask: "Do I need income protection insurance if I already have ACC?"
- A customer might ask: "Can my partner work while my visa application is being processed?"
- A customer might ask: "Is this tree dangerous, or does it just need pruning?"
These are not generic keywords. They reveal intent, uncertainty, context and objections. They show what customers need to understand before they can make a decision.
That is the kind of granular, useful content businesses should be capturing and publishing.
Why website chatbots are strategically important
You probably view chatbots as customer service tools that answer common questions, capture leads and provide after-hours support. All true.
But in the AI search era, a well-designed chatbot has another major advantage:
It creates a continuous, searchable record of the questions website visitors naturally ask in a conversational AI interface.
A chatbot does not reveal every question your entire market asks ChatGPT, Gemini, Perplexity or Google. Its users are a particular group who have already reached your website.
But that does not make the data less valuable. These visitors are often close to making a decision, and their questions expose service confusion, pricing concerns, objections, missing information and the language real customers use to describe their needs.
Over time, patterns emerge. Ten visitors ask about pricing. Five ask whether you service a particular area. Several ask what happens after they submit the form. Others ask questions your website does not answer at all.
That is not just support data. It is a content strategy hiding in plain sight.
Turn conversations into useful source material
Capturing questions is only the first step. The real value comes from turning them into clear, accurate and publicly accessible answers.
A practical workflow looks like this:
- Capture. Collect real questions from chatbots, sales, support, forms, search data and customer conversations.
- Identify patterns. Group repeated questions, objections and points of confusion.
- Verify the answer. Have the right person inside the business confirm what is accurate, current and safe to publish.
- Publish. Turn the answer into an FAQ, service page section, pricing guide, comparison article, troubleshooting guide or process explanation.
- Reuse. Add the approved answer to your chatbot, internal knowledge base, sales resources and future AI agent workflows.
- Measure. Track whether the new content is being found, cited, used and followed by better enquiries.
If visitors keep asking "how much does it cost?", create useful pricing content. If they keep asking "do you service my area?", improve your service area information. If they keep asking "what happens after I submit the form?", explain the process.
This is practical Answer Engine Optimisation. Not theory. Not keyword stuffing. Real questions, real answers and real business value.
Pricing is one of the clearest content gaps
One of the strongest patterns in customer conversations is pricing. People want to know what things cost.
Businesses often avoid publishing pricing because competitors might see it or because "it depends on the job". But avoiding the subject does not stop customers asking. It simply leaves them to find guidance somewhere else.
You do not need to publish a fixed price for work that genuinely varies. You also do not need to expose your entire price book.
But most businesses can explain more than they currently do:
- Typical price ranges
- Starting prices or minimum charges
- What affects the final cost
- What is included
- Why quotes can vary significantly
- When a cheaper option may cost more later
- Examples based on common situations
If customers repeatedly ask about price, that is not an annoyance. It is evidence of unmet information demand.
Publishing useful pricing guidance may not be the biggest AEO opportunity for every business, but it is one of the first content gaps worth investigating.
Why this matters for AI agents
AI search is only part of the change. AI agents take the next step.
Today, people ask AI for answers. Increasingly, they are also asking it to research options, compare providers, monitor availability and complete tasks on their behalf.
"Find and schedule a reputable local electrician with competitive pricing who can install my new Evnex E2 EV charger on Wednesday or Thursday afternoon next week."
To complete that task, an AI agent needs reliable and actionable information. It needs to know what services you offer, where you operate, how your pricing works, whether you are qualified, whether appointments can be booked online, what availability exists and what conditions apply.
Marketing language alone will not be enough. The information must be explicit, current, accessible and supported by the systems needed to take action.
Your knowledge library is the source-of-truth layer
A knowledge library is a structured collection of what your business knows and wants people and AI systems to understand. It may include services, service areas, FAQs, pricing guidance, expert answers, policies, processes, credentials, testimonials, booking rules, troubleshooting advice and approved answers from past conversations.
But an internal knowledge library is not automatically visible to AI search systems. Its value comes from what you do with it.
The website remains the public publishing and engagement layer. It is where people can read, verify, trust and act on the information. The knowledge library sits behind it as the organised source of truth that feeds your website, chatbot, sales material, support systems and future AI agents.
Your website is still the shopfront. Your knowledge library is the system that keeps the answers behind it complete, consistent and current.
What businesses should do right now
- Capture real customer questions. Use chatbot transcripts, sales conversations, support messages, forms and search data.
- Review them regularly. Look for repeated questions, pricing concerns, service confusion, objections, booking friction and content gaps.
- Turn the strongest questions into public answers. Create FAQs, service page sections, pricing guides, comparisons, process explanations and troubleshooting content.
- Organise approved answers into a knowledge library. Build a structured source of truth for the business.
- Use that knowledge everywhere. Improve your website, chatbot, sales process, support resources, SEO, AEO and future AI agent readiness.
The future belongs to businesses with better answers
The next phase of digital marketing will not be won by the business with the most generic blog posts, the business that blindly publishes AI content every week, or even the business with the prettiest website.
It will be won by businesses that understand what customers need to know and provide clear, specific, current and well-supported answers.
Customer conversations are not just support transcripts or sales notes. They are a roadmap. They show you what people want to know, where your content is failing, what your business needs to explain and what AI systems may need to retrieve before they can confidently include you in an answer or recommendation.
Because in the AI agent era, the winner will not simply be the business with the most traffic.
It will be the business with the answers people need and the systems that make those answers easy to find, understand and act on.