Your next visitor might be a person, a search crawler, or an AI assistant acting on someone’s behalf. Here is the technical and content checklist for building a website that works for all three in 2026.
For most of the web’s history, a website had two audiences: people and search engine crawlers. In 2026 there is a third. AI assistants read your pages to answer questions, compare providers and, increasingly, take actions such as filling in a form or adding a product to a cart on a person’s behalf.
Google added an experimental “agentic browsing” category to its Lighthouse auditing tool in 2026, a clear sign that how well software agents can use a site is becoming a measurable quality, alongside speed and accessibility.
The good news is that you do not need a separate “AI website”. The sites that work best for AI are, almost without exception, the sites that are clear, fast, well-structured and accessible for people. This checklist sets out 15 requirements, grouped into structure, content, technical performance, trust and action.
Key takeaways
- Build one website that serves people, crawlers and AI agents; the same fundamentals help all three.
- Structure: semantic HTML, one H1, logical headings, server-rendered content, clean URLs.
- Content: answer-first copy, specific facts (pricing, timelines, service areas), real FAQs and dates.
- Technical: Core Web Vitals in the “good” range, accurate structured data, open access for search and AI crawlers.
- Trust and action: named people, consistent business details, accessible forms and clear calls to action that an agent can understand.
Part 1: Structure
1. Use semantic HTML, not just visual design
Headings, lists, tables, buttons and links should be real HTML elements, not styled boxes. A card title styled to look like a heading should be marked as one. A button should be a button, not a div with a click handler. Semantic structure is how screen readers, crawlers and AI agents understand what is on the page and what each element does.
2. One H1, then a logical heading outline
Each page should have one H1 that states its topic plainly (“Mobile App Development for iOS and Android”, not just “Apps”). Sections use H2, items within sections use H3. Do not skip levels for styling reasons. Many page builders let you set the HTML tag independently of the visual style; use that.
3. Render important content on the server
If your headline, service descriptions or prices only appear after JavaScript runs, some crawlers and agents may see an empty page. Make sure core content is present in the initial HTML. Sliders, tabs and accordions are fine, as long as their content is in the markup.
4. Clean URLs and a clear information architecture
Readable, stable URLs (/services/website-development/) and a logical hierarchy (Home, Services, Service) help everyone navigate. Avoid duplicate pages for the same topic, placeholder links that point to “#”, and leftover demo pages from your theme.
Part 2: Content
5. Lead with the answer
Open every page, and every major section, with a direct answer in 40 to 60 words: what this is, who it is for, and what happens next. That is the part an AI assistant is most likely to quote, and the part a busy visitor reads before deciding to stay.
6. Be specific
AI systems and buyers both reward specifics:
- Price ranges or “starting from” figures, in your local currency.
- Typical timelines.
- Service areas (“Toronto and the Greater Toronto Area, with remote projects across Canada”).
- What is included and what is not.
- Platforms and tools you use.
Vague claims such as “tailored solutions for every need” cannot be quoted usefully.
7. Add real FAQs to commercial pages
Write the questions your sales team hears every week, and answer them plainly. Mark questions as headings and keep answers concise. FAQ content is among the most frequently extracted material in AI answers.
8. Show dates and keep content current
Display “last updated” dates on service pages and articles, and update them when facts change. Assistants answering time-sensitive questions prefer recent, clearly dated sources.
Part 3: Technical performance
9. Meet Core Web Vitals
Aim for the “good” thresholds on mobile:
| Metric | What it measures | Good threshold |
|---|---|---|
| Largest Contentful Paint (LCP) | How quickly the main content appears | 2.5 seconds or less |
| Interaction to Next Paint (INP) | How quickly the page responds to input | 200 milliseconds or less |
| Cumulative Layout Shift (CLS) | How much the layout jumps while loading | 0.1 or less |
Common fixes: resize and compress images (WebP or AVIF), set image dimensions, avoid lazy-loading the main hero image, remove unused plugins and sliders, use caching and a CDN, and defer non-critical scripts.
10. Implement accurate structured data
Structured data (schema.org markup in JSON-LD) describes your business in a machine-readable way. For most business sites the priorities are:
| Schema type | Where | What it tells machines |
|---|---|---|
| Organization (or LocalBusiness / ProfessionalService) | Homepage, About | Legal name, logo, address, contact details, founder, official profiles |
| Person | About, author pages | Who your experts are and their credentials |
| Service | Each service page | What you offer, where, and who provides it |
| Product and Offer | Product pages | Price, availability, reviews |
| Article / BlogPosting | Posts | Author, publish and update dates |
| BreadcrumbList | All pages | Where the page sits in your site |
| FAQPage | Pages with visible FAQs | Questions and answers |
Rules: markup must match what is visible on the page, avoid duplicate Organization nodes from multiple plugins, and validate with Google’s Rich Results Test and the Schema.org validator.
11. Keep the door open for the right crawlers
Check that robots.txt, your security plugin and your CDN do not accidentally block the crawlers that power search and AI answers. Relevant user agents include Googlebot, Bingbot, OAI-SearchBot and ChatGPT-User (OpenAI), PerplexityBot and ClaudeBot (Anthropic). Make deliberate decisions about training crawlers separately. Submit sitemaps to both Google Search Console and Bing Webmaster Tools.
An llms.txt file, a short Markdown list of your key pages, is optional. Google has said it does not use it for search. It is cheap to add, but it is not a substitute for any item on this list.
Part 4: Trust
12. Show who is behind the business
Name your founder and key people, with photos, credentials and links to their professional profiles. Add author bios to articles. Display a real address, phone number and email. Remove anything that looks fake or unfinished: template testimonials, “0 projects” counters, lorem ipsum, links to your theme vendor’s demo site.
13. Keep business details consistent everywhere
Your business name, address, phone number and description should match exactly across your website, structured data, Google Business Profile, LinkedIn, directories and partner sites. Inconsistency makes it harder for search engines and AI systems to be confident they are describing the right business.
Part 5: Action
14. Make forms and calls to action accessible and unambiguous
AI agents, like people using assistive technology, rely on clear labels. Every form field needs a visible label tied to it in the code, clear error messages, and a sensible tab order. Buttons should say what they do (“Request a free audit”, not “Submit”). Avoid CAPTCHAs that block legitimate use where lighter-weight spam protection would do. Follow WCAG 2.2 AA; in Ontario, organisations with 50 or more employees also have accessibility obligations for their websites under the AODA.
15. Prepare for agent-driven transactions (if you sell online)
Shopping and booking are starting to happen through AI agents. In 2026, Google launched the Universal Commerce Protocol with retailers including Shopify, Etsy, Wayfair, Target and Walmart, letting agents in AI Mode and Gemini check out on a shopper’s behalf, and OpenAI had already introduced its Agentic Commerce Protocol, developed with Stripe, in 2025 for purchases inside ChatGPT. If you sell online:
- Keep product data, prices, shipping and return policies identical across your site, your structured data and your product feeds.
- Maintain a complete Google Merchant Center feed.
- Follow your e-commerce platform’s guidance on agentic checkout programmes as they roll out in Canada.
- Make sure analytics can attribute orders that arrive without a normal website session.
The AI-ready website checklist
| # | Requirement | Done? |
|---|---|---|
| 1 | Semantic HTML for headings, lists, tables, buttons, links | |
| 2 | One descriptive H1; logical H2/H3 outline | |
| 3 | Core content present in server-rendered HTML | |
| 4 | Clean URLs; no duplicate, placeholder or demo pages | |
| 5 | Answer-first introductions on key pages | |
| 6 | Specific pricing, timelines, service areas, inclusions | |
| 7 | Real FAQs on commercial pages | |
| 8 | Visible publish and update dates | |
| 9 | Core Web Vitals in the “good” range on mobile | |
| 10 | Accurate, validated structured data | |
| 11 | Search and AI crawlers allowed; sitemaps in Google and Bing | |
| 12 | Named people, credentials, real contact details; no template leftovers | |
| 13 | Consistent business details across the web | |
| 14 | Accessible, clearly labelled forms and CTAs (WCAG 2.2 AA) | |
| 15 | Consistent product data and feeds for agentic commerce (if e-commerce) |
The most common problems we find on business websites
When we audit small and mid-sized business websites, the same issues come up again and again. Most are quick to fix, and most hurt people, search engines and AI assistants at the same time.
| Problem | Why it hurts | Fix |
|---|---|---|
| Theme demo content left on the live site (testimonials, counters, sample projects) | Destroys credibility; signals an unfinished site | Search the site and database for demo text, images and links before launch |
| One-word H1s (“Services”, “About”) | Gives no topic signal | Write descriptive H1s that match the page’s purpose |
| Headings chosen for size, not structure | Breaks the outline that machines read | Set heading tags separately from visual styles |
| Images with empty alt text and 2,500px originals | Slow pages; invisible content | Resize, compress and describe images |
| 300-word service pages | Not enough to rank, convert or be cited | Expand with deliverables, process, pricing, proof and FAQs |
| No named people or credentials | Weak trust signals | Add leadership bios and author boxes |
| Menu items linking to “#” | Dead ends for users and crawlers | Link to real hub pages |
| Social icons linking to the platform’s homepage | Missed entity signals | Link to your actual company profiles |
| Blog posts in “Uncategorized” with no dates | Poor organisation and freshness signals | Add categories, authors and visible dates |
| Contact forms with vague buttons and typos | Lost leads | Clear labels, clear buttons, proofread copy |
A pre-launch QA routine for developers
Before any new site or major redesign goes live, run this routine:
- Crawl the staging site with an SEO crawler to find broken links, missing titles, duplicate pages and redirect chains.
- Search the database for theme vendor names, “lorem ipsum”, “demo” and placeholder URLs.
- Run Lighthouse on the homepage, a service page, a blog post and the contact page, on mobile.
- Validate structured data on each page template.
- Test every form, including error states and confirmation messages, with a keyboard only.
- Check robots.txt, the XML sitemap and that the staging “discourage search engines” setting is switched off at launch.
- Confirm that http and www versions redirect to the canonical URL in a single step.
- Ask two or three AI assistants about your business after launch, and record the answers as a baseline.
Rebuilding or retrofitting?
Most businesses do not need to start again. A focused retrofit, covering heading structure, schema, performance, content rewrites for key pages and template clean-up, often delivers most of the benefit in a few weeks. A rebuild makes sense when the theme is slow and heavily customised, the content management experience is painful, or the information architecture no longer matches what you sell.
If you are rebuilding, write the requirements above into the brief, and ask your developer to show Lighthouse and Rich Results Test results before launch.
Frequently asked questions
No. Search engines and AI assistants read the same pages people do. Building a clear, fast, well-structured, accessible site serves all of them.
No. Structured data helps machines understand your content, but search engines and assistants decide what to show. It is a foundation, not a guarantee.
It depends on your business model. Blocking the crawlers that power AI search reduces your chances of being cited or recommended. Many businesses allow search and retrieval crawlers and make a separate decision about training crawlers.
Run Lighthouse and PageSpeed Insights, validate structured data, check your robots.txt and CDN settings, review pages with a screen reader, and ask several AI assistants questions your buyers would ask to see how they describe your business.
Most mainstream platforms, including WordPress, can meet every requirement here when built carefully. Implementation quality matters more than the platform.
The bottom line
The websites that win in 2026 are not the flashiest. They are the clearest: easy for a person to understand in five seconds, easy for a crawler to parse, and easy for an AI assistant to quote or act on. Build for all three audiences, and you build a site that keeps working as search keeps changing.
Netxeno designs and builds fast, AI-ready websites and helps businesses get found in AI answers through Generative Engine Optimization. See how your current site scores with a free audit.
