THE AI-READY SITE ARCHITECTURE — PART 2 OF 5

The AI-Ready Site Architecture, Part 2: Projects

This is Part 2 of a five-part series on the minimum viable site architecture for AI citation — pages such as: About, Projects, Service Areas, Pricing, and FAQ. In Part 1 we covered About, the entity anchor. Now we get to the page that does more AI-citation work than any other page on your site: Projects.

What This Page Is For, From AI’s Perspective

AI answer engines don’t rank brand sentiment — they rank retrievable proof. A homepage that says “we build custom glass countertops” is invisible to AI. A page that says 130 sq ft, 1.5″ thick, low-iron glass, L-shaped, 687 lbs, St. Petersburg FL, 19-day turnaround, LED with 6 million color options, backed by a verified 5-star review — that’s a citation machine.

That’s the entire mechanism: specificity plus verifiability plus structured data is exactly what a retrieval system is built to surface.

• Specificity = extractable facts. Numbers, dimensions, locations, and timelines are the atomic units these systems pull into an answer.
• Named entities = trust signals. A specific city, a specific spec, a specific date give AI something real to ground the claim in.
• Structured data = machine-readable proof. Project/CreativeWork schema plus Review/AggregateRating schema turns your story into a record a model can parse with certainty.
• Third-party validation = authority and trust. A verified review attached to the specific job confirms the claim independently — it’s not you saying it, it’s the customer.

The Great Version

Strip away the schema and formatting and every great project page is telling the same small story: customer had a need, customer wanted a specific outcome, customer bought, customer received it, customer was happy, customer’s result is documented in photos and their own words. The “clinical” specs aren’t decoration on top of that story — they ARE the proof inside it.

Architecture to hit every time:

• Spec Block — exact dimensions, materials, weight, finish, technology used
• Location/Timeline Block — city/state, start date, delivery date, total turnaround
• Proof Block — photos or drawings of the actual work, named customer, verified review

This architecture holds across industries. A law firm’s version: case type, jurisdiction, methodology, and a measurable outcome, backed by an attributed client testimonial. A contractor’s version: square footage, materials, crew size, and a homeowner review. Same three blocks, different trade.

The Bad Version

The pattern we see constantly: a single “our work” gallery page — a grid of photos with no specs, no dates, no named outcomes, and no schema. It looks fine to a human scrolling by. It gives AI nothing to extract, because there’s no fact in the page, just an image.

The Interlink Mechanic

• Links out to Service Areas — every project should be tagged to the zip code or city it happened in, so a library of projects becomes a geographic density signal, not just a list
• Links out to About — ties the work back to the credentialed entity that did it
• Links in from Gallery, if one exists — photos live inside the project first, gallery second

The Untapped Asset: Your Back Catalog

You don’t need new projects to start. Most businesses with any real history have 20-30 completed jobs sitting in old folders, invoices, and camera rolls — undiscoverable to AI until they’re published in this format. Pulling the spec, pulling the photo, pulling the quote, and dropping it into a reusable template is retrieval and assembly work, not creative work. With organized records, ten pages in a day is a real number.

What’s Next

One project proves a job got done. Many projects, tagged by location, prove you own a territory. That’s Service Areas, which we cover in Part 3.