An AI assistant can now find a business that fits a customer’s needs and book the job without the customer ever visiting that business’s homepage. In that case, the website’s main job is to give AI systems accurate information they can act on, and attracting visitors becomes secondary.
Sachin Puri, CEO of Newfold Digital, Bluehost’s parent company, joined me on the SEJ podcast to discuss that shift. Puri explained that a website remains the authoritative source of business information, even when the customer journey happens somewhere else.
“AI is actually not making websites irrelevant; it’s just changing its job.”
Rankings and clicks no longer show the full picture. The work now includes making sure an AI system understands a business well enough to describe it accurately and help a customer book or buy.
A Website Can Support A Decision Without Receiving The Visit
Puri illustrated the change with a hypothetical customer who asks an agent to find an electrician available tomorrow afternoon who charges less than $300 and has strong reviews, and then to book the appointment.
The customer might never visit the electrician’s website, but the agent still has to answer the same questions a person would. It has to find out where the business operates, whether it is licensed, what it charges, how customers rate it, and whether it has an opening tomorrow.
To find those answers, the agent may read the electrician’s website directly, including the structured data in its schema markup. It may also check the business’s Google Business Profile, its Yelp listing, and reviews on other sites. Puri refers to all of that information together as the business’s digital presence. The website is the one part of it the business fully owns and controls, so it serves as the reference point everything else should match.
“The page views or visits may disappear, but the need for authoritative digital presence continues,” Puri said. As he describes it, authority comes from information that is accurate and consistent across sources, and from the reputation customers build through their reviews.
What 83.4 Million Daily AI Crawls Measure
Bluehost’s position as a host gives Puri a view of machine activity across millions of sites. He cited a 90-day Bluehost analysis that found 83.4 million verified AI crawler requests per day across Bluehost-hosted sites.
He cautioned against reading that volume as customer discovery. By Bluehost’s classification, roughly 90% of those requests were for model training, about 9% for search indexing, and roughly 1% for live assistant fetching.
The figures describe crawler activity. They don’t show purchases, referrals, or instances in which an AI system recommended a business. Puri described crawling as the first step in a longer sequence. After a site is crawled, its information may be retrieved, cited in an answer, and used in a recommendation that eventually brings the owner business.
Keep Crawler Activity Separate From Referrals
For ChatGPT and OpenAI specifically, Puri said Bluehost’s report estimated one observable referral visit for every 5,350 verified HTML crawl requests.
Puri said the ratio isn’t a conversion rate. The crawl count includes training activity, and Bluehost can’t observe every customer journey that AI influences.
His point was that machine consumption and human referral traffic are separate measurements, and businesses should track them on their own. He urged businesses to keep measuring clicks and optimizing their existing funnels. Just recognize that some customers now choose a business based on what an AI assistant told them, and those decisions may never show up in website analytics.
Measure Visibility, Accuracy, And Action
Puri proposed a measurement framework with three parts:
- Visibility: When a prospective customer asks a relevant question, does the business appear? Is it mentioned or cited?
- Accuracy: When it appears, are its prices, services, locations, and operating hours represented correctly?
- Action: Does that exposure lead to an observable referral, qualified inquiry, booking, purchase, or revenue?
A business needs all three. Showing up in an AI answer doesn’t help much if the answer lists the wrong price. Even an accurate answer may never turn into a phone call or a booking.
Accuracy is a problem I run into with clients. Outdated pricing can linger on a business’s own site, in its schema markup, or in third-party discussions such as Reddit threads. AI engines increasingly treat the website as a hub for cross-checking what they find elsewhere, so those inconsistencies carry real weight.
Puri said he wouldn’t treat crawler volume as an objective at all. The goal is for the business to be understood, recommended, and ultimately chosen.
Good AI Optimization Is Still Good Web Engineering
Puri sees the shift as an opportunity for the people who already build and optimize websites. Clean site structure, accessible content, accurate information, consistent business details, and schema markup are as important as ever.
The goal of that work has expanded. A strong ranking still helps, but the business also needs a machine to understand what it offers well enough to match it with a customer’s specific request.
Puri said many local service pages were built to rank, which left them heavy on keywords and light on facts. A page promising “reliable electrical services” gives an agent almost nothing to work with. A page that lists the services offered, the service area, hours, licensing, and recent customer reviews gives the agent enough to match the business to a customer’s request.
He gave a restaurant example. A customer could ask for a family-owned, family-friendly restaurant with vegetarian and vegan options, wheelchair access, and an accessible restroom that can seat six people that evening.
An agent can’t fill that request from a general description. It needs to find each attribute and confirm availability, and as I pointed out during our conversation, it would also need to book the table before someone else did.
For an agency, the work is helping a business present those facts in a form both people and agents can use. Puri said reviews and customer feedback carry more weight now because agents rely on signs of authority.
“If you have a great foundation, it’s going to amplify.”
Puri said AI amplifies a shaky foundation as well. Producing more content won’t fix outdated information or a poor customer experience.
Audit AI Answers Without Using Your Brand Name
Puri said Bluehost’s small business AI confidence study found that only 13% of respondents said they were optimizing for search assistant visibility.
His suggested starting point is to spend 10 minutes a day acting like a prospective customer.
Ask ChatGPT, Gemini, or another assistant the questions customers would ask before choosing the business. Leave the company’s name out of the prompt, because a branded query gives an owner a narrow view of what customers see before they’ve made a choice.
Then check whether the business appears, whether its information is accurate, which competitors show up, and what sources the assistant relies on. Pay attention to whether the assistant recommends the business or only mentions it.
I’d extend the exercise beyond prompt wording, since customers also research by voice, in mobile apps, in the car, and with images. For one of my clients, which sells smart kitchen appliances, Google Search Console’s new multimodal reporting showed Google Lens searches leading to product pages. My working theory is that people photograph an appliance they see in a friend’s kitchen and search for it.
Puri recommended repeating the exercise of auditing AI answers regularly and using the visibility, accuracy, and action framework to decide what needs attention.
Use The Time Agents Save To Build Human Trust
Puri distinguished chatbots, which answer questions, from agents, which perform tasks.
For a business with limited staff, agents can handle inquiries, book appointments, follow up, send reminders, and support marketing campaigns. An AI receptionist, for example, can answer calls and book appointments around the clock so the owner can focus on delivering the service.
He said Bluehost’s research found that 79% of respondents were aware of AI agents, but only 16% had deployed one.
In his view, the main benefit is the time agents free up for work AI can’t do for a business, which is building trust and customer relationships. That trust comes from doing good work over time and asking satisfied customers for reviews and referrals.
Agents can help with the asking. An agent that confirms an appointment and texts the customer when the technician is on the way can also check in after the job and ask for a review. Those reviews give AI systems more evidence that customers trust the business.
3 Ways To Start This Week
Puri closed with three recommendations a business can act on now:
- Be your customer. Test relevant, unbranded questions in AI assistants. Check visibility, accuracy, recommendations, and the sources behind the answers, then fix missing or outdated information.
- Automate one recurring task. Find work that eats time every day and look for a purpose-built tool or agent to handle it. Reinvest that time in service quality and customer relationships.
- Get the fundamentals right before chasing answer engine optimization (AEO) tricks. Prioritize accessibility, structure, freshness, consistency, authority, and technically sound infrastructure. Keep measuring traffic, and also track recommendations and meaningful customer actions.
Puri said the traditional funnel still works. More of the customer’s decision now happens inside an AI assistant, though, and there a business is judged by the information it has published and the reputation its customers helped build with reviews.
That gives SEOs a clear assignment. Keep a client’s pages and schema markup accurate, because AI agents use that information when they talk to customers who are ready to buy. Helping clients earn more reviews gives agents another reason to recommend the business.
The best place to start is the first recommendation. Open an AI assistant, ask it what your customers would ask, and see what it says.
More Resources:
Featured Image: Ball SivaPhoto/Shutterstock