Does Your Small Business Website Need to Be AI-Ready?

Every few months there is a new reason to rebuild the website. First it needed to be headless. Then AI-powered. Now it needs to be ready for the agentic web. Medium-sized service businesses hear the same pitch, often with a larger invoice attached.

I recently joined a LinkedIn discussion started by Imran Siddiq (opens in a new tab) about WP Engine’s article on preparing WordPress for AI agents. His concern was familiar: developers get excited about new technology, but what problem does it solve for the person paying for the website?

That is the question I would start with too.

Your website needs to help people find you, understand your offer and become customers. Additional technology should earn its place by improving that process or reducing the work behind it.

What AI-ready actually means

The phrase gets used for several different things, which makes the conversation unnecessarily confusing for a small business owner trying to decide where to spend money.

It might mean that an AI search service can find your business and use information from your website. It might mean an assistant can check availability and make a booking. Or it might mean you use AI inside WordPress to help write and manage content.

Those are different jobs.

A carpenter who wants to appear when someone searches for fitted wardrobes needs clear service information and evidence of their work. A booking business that wants an assistant to reserve appointments needs reliable availability and a way to create those reservations. Someone who wants help drafting project descriptions needs an editing tool.

Buying one does not automatically solve the others.

WP Engine’s article makes a useful observation here. In the testing it describes, agents often used existing HTML rather than the experimental features added specifically for them. The article also cautions against treating those experiments as established best practice. That supports a measured approach, although it does not prove that every agent can use every ordinary website successfully. Read WP Engine’s article.

What I would fix before buying an AI upgrade

For a typical small business website, I would first check whether a potential customer can answer a few straightforward questions:

• What exactly do you do?
• Where do you work or deliver?
• Is your service suitable for their particular job?
• What evidence shows that you can do it well?
• What happens when they contact you?
• How do they call, enquire, book or buy?

These sound obvious. They still deserve more attention than another badge in the website footer. The same gaps show up on many medium-sized company sites, only with more pages and more people waiting for the enquiry to arrive.

An electrician might have a beautiful homepage that never explains whether they take small residential jobs. A bar or hotel might still show last season’s menu as a photograph. A photographer might show excellent work without explaining packages, turnaround or how to request a booking.

Those gaps create extra work for customers and staff. I would spend money fixing them before paying for a custom AI integration. If the public pages are already unclear, making the business easier to mention in AI search only sends more people to the same missing answers. See how ChatGPT visibility actually works.

The technical foundations matter too: a usable mobile layout, readable text, working forms, reasonable loading speed, maintained software and backups that can be restored. Someone also needs to own the routine work of updating the information. That is part of a normal small business website budget, not a separate AI project.

If the contact form fails, being mentioned by an AI assistant only sends more people to a broken contact form.

HTML, schema and AI search: what is worth doing?

HTML gives a web page its structure. Proper headings, descriptive links, labelled fields and useful text help make that structure understandable.

You do not need to know how to write it. You should expect whoever builds your WordPress website to handle it competently.

Schema markup adds structured descriptions of things on a page, such as a business or a product. It can be useful when it accurately represents the content. It should stay consistent with what customers actually see.

But “schema for everything” is not a strategy. Adding more markup does not compensate for missing information or an unclear offer.

Google currently says that its generative AI search features do not require special schema or files such as llms.txt. Its guidance continues to point website owners toward established SEO practices. That statement concerns Google Search; other services can have different integrations and requirements. Google’s guidance on AI search.

My practical recommendation is to make your business information clear, keep it current and implement relevant structured data properly. If somebody proposes additional markup, ask what consumes it and what result it enables.

“AI prefers it” needs a more specific explanation before it becomes an invoice.

Reading your website and booking a job are different problems

An assistant finding your opening hours is relatively straightforward. An assistant booking one of your engineers requires more information and more control.

It needs to know which services can be booked, which areas you cover, how long each job takes and whether the appointment is still available. The system must also prevent duplicate bookings and explain what happens if the request fails.

For some trades, automatic booking may be the wrong process altogether. A bespoke staircase or an uncertain repair can require photographs, measurements or a conversation before anyone can promise a price or a date.

In that case, a well-designed enquiry form might be the better investment. Connecting that form to the system you already use for follow-up is often more useful than a chatbot. See when a website CRM connection is worth it.

An API allows another application to request information or perform actions without using the normal website interface. WordPress already has a REST API for working with site content, though specialist operations such as bookings depend on the systems involved. WordPress REST API documentation.

MCP, or Model Context Protocol, provides a standard way to connect AI applications to data and tools. It can be useful when you have a defined workflow that an assistant should perform. Official MCP introduction.

Neither term tells you whether the integration is commercially useful for your business. You still need to identify the task, the system responsible for it and the people who will use it.

For an agent that changes bookings or customer records, I would also expect limited permissions, a record of its actions and a practical way to handle mistakes. That work belongs in the proposal alongside the impressive demonstration.

Which technologies would I prioritise?

These are my starting recommendations for a typical small or medium-sized business website. The priorities change when the actual workflow changes.

GEO means generative engine optimisation. AEO means answer engine optimisation. You may also see AIO used with different meanings, so ask the supplier to spell it out.

The names matter less than the work. A proposal should tell you which pages, information, systems or customer journeys will improve.

Accessibility belongs in the foundations

I mentioned WCAG alongside several other technical terms in the LinkedIn comments. In a longer discussion, I would separate it out.

WCAG is a set of accessibility guidelines for making digital content usable by people with disabilities. It addresses real barriers to using websites. W3C’s WCAG overview.

For a business owner, that includes customers being able to read the content, navigate with a keyboard, understand form labels and recover from input errors.

I would treat accessibility as part of the quality of the build. A proposal mentioning “WCAG 2.2 AA” should explain the scope and how conformance will be evaluated. A plugin badge alone is insufficient evidence.

You can question an unclear accessibility proposal while still taking accessibility seriously.

What this looks like for different businesses

Consider a few hypothetical examples. The useful question is not “are we AI-ready?” It is which customer or staff problem is actually costing you money.

Hospitality

For restaurants, bars and hotels, current menus, opening hours, directions and a reliable booking process come first. The same applies to events, rooms and table availability. Reminders or better cancellation handling may beat a conversational assistant. Agent booking is more interesting when the existing provider supports it.

Trades

For plumbers, electricians and similar trades: accurate service areas, clear job types, a prominent phone number, useful project evidence and a short enquiry form. Connect the form if enquiries regularly get missed. A custom agent endpoint needs a much more specific reason.

Beauty services

Hair, nails, makeup, massage and similar appointments need a current price list, who is available, opening hours and a booking process people can finish on a phone. Reminders and no-show handling often save more time than a chatbot. An agent is only useful if the booking tool you already use can actually take the appointment.

Designers and photographers

The portfolio needs packages, turnaround, usage and how quotations work. A form that collects the brief, dates and reference files may save more time than a chatbot trying to price work that needs human judgement.

Small online shop

Product details, stock accuracy, delivery costs, returns information and checkout deserve attention first. Further integrations become worthwhile as the catalogue and order volume grow, especially when staff copy information between systems.

Gyms, trainers and health services

Class times, membership options, who is available and a booking process people can finish on a phone. Trial sessions and cancellation rules should be obvious. If the timetable or membership software already handles bookings, connect that before adding a conversational assistant.

Headcount alone does not determine technical needs. A small company can run a complex operation. A larger company can still have a straightforward public website.

AI may help behind the scenes first

One comment in the discussion pointed toward something more immediate: website owners want help doing the repetitive work inside their existing systems.

That makes sense to me.

Preparing a draft from job notes, suggesting a reply to an enquiry or organising incoming requests could save time without changing how customers use the website.

I would still start with one task. Define what a correct result looks like, decide what needs human review and see whether the time saved survives that review.

Some jobs do not need AI at all. If every enquiry should create a customer record and send an acknowledgement, a normal integration may handle the entire process.

AI becomes more useful when the input varies and interpretation helps. The business still needs a way to catch incorrect interpretations.

Ask for the full cost and a small test

A feature can work technically and still be a poor purchase.

The cost includes setup, subscriptions, usage charges, maintenance, staff training and the time spent checking or correcting results. It also includes whatever else you delay to pay for it. That is the same thinking I use when someone is deciding whether to update, redesign or rebuild.

Before approving an extra feature, I would ask:

1. What problem does this solve in our current business?
2. What evidence shows that we have that problem?
3. Can our existing software already handle it?
4. What will it cost to operate and maintain?
5. How will we tell whether it worked?
6. Can we remove it without disrupting the rest of the business?

For automation, measure staff time and errors before and after. For enquiries, look at their quality and whether they turn into work. For bookings, look at completed reservations and the administration required to manage them.

AI visibility is harder to attribute perfectly. A mention or a crawler visit is a useful signal to investigate, but it does not by itself demonstrate a paying customer.

A limited experiment can still be sensible before substantial demand appears. It should have a budget, a defined purpose and a review point.

What I would invest in today

For most small businesses and trades, I would begin with a website that clearly explains the business, works reliably and makes the next step easy. Then I would improve the process behind the enquiries, bookings or orders it generates.

After that, I would assess additional AI features against a specific opportunity. The same order applies to a growing medium-sized company, with more attention on the systems that already sit behind the public site.

My own comments in the discussion included the possibility that structured JSON could be cheaper for AI systems to process than entire pages. That can be true for a focused response, but the format alone does not establish the saving. Payload size and how the system processes it matter too. It certainly does not establish a business case for every website owner.

Spend on the next problem you can explain and measure. Keep enough flexibility to adopt useful technology when the reason becomes clear.

If you already have a proposal full of unfamiliar terms, ask the supplier to connect each one to something your business needs. You should be able to understand why you are paying for it.

What an AI proposal should not hide

Final takeaway

Most small and medium-sized businesses do not need an AI-ready rebuild. They need a website that explains the work, stays current and makes it easy to call, enquire, book or buy. AI can help later, either in search, in booking or behind the scenes, once you can name the problem and measure the result.

Spend on the next problem you can explain and measure. Keep enough flexibility to adopt useful technology when the reason becomes clear.

If a proposal is full of unfamiliar terms, ask the supplier to connect each one to a page, a system or a customer journey in your business. You should still understand the job after the acronyms are removed.

Frequently asked questions