SEO for Architects: What Google Can Actually Read

We pulled the real United States search results for "seo for architects" before writing a word of this. Seven pages hold the organic positions. Four of them show zero measured backlinks, the strongest one earns an estimated 319 visits a month, and two of the top six slots are forum threads where architects ask each other what works.

That last detail is the whole story. When a question sends people to Reddit and ResearchGate instead of to an article, the articles are not answering it.

We read the two best-ranked guides in full. Together they run about 6,300 words and name three concrete search practices between them: a landing page per service, images scaled down and renamed, and alt tags. Everything else is website advice. Fast loading, mobile layout, clear navigation, strong visual hierarchy, contact forms that qualify, social proof, a lead magnet.

None of it is wrong. It is also not SEO, and it leaves the half that Google documents in public completely vacant: what a crawler finds on a project page, which structured data type an architecture practice is supposed to use, how images are read, and which report tells you whether any of it worked.

This article covers that half. Every claim about Google's behaviour below is quoted from Google's own documentation, with the date we checked it, because this is a field where advice rots quietly.

What ranks today, and why that is good news

The search results for this query are weak in a specific, measurable way.

An AI Overview sits at the top with seven source links under it. Six of the seven are agency blog posts (Netvantage, Abstrakt Marketing Group, Elsner, ArchMarketing, KIJO), one is a YouTube video. Below that, the organic list is articles, not service pages: a guide for builders and architects, a lead-generation listicle, a Quora topic, a LinkedIn post, a page of marketing tools.

Three numbers matter more than the names:

Four pages ranking on a commercial query with no measured links pointing at them. There is no authority wall here. The pages rank because almost nobody has written a better answer, which is the only kind of search opportunity worth a practice's evening.

The AI Overview is the real competition, and it is a genuine risk: the searcher may read the summary and never click. You cannot opt out of that. You can be one of the sources it draws from, which is the subject of the last section.

Two honest caveats before the work. Every ranking estimate above comes from third-party tooling and moves week to week, so treat the shape of the result (weak pages, no links) as the signal rather than any single figure. And no one can guarantee a #1 ranking on Google. That is not our hedge, it is the sentence Google publishes on its own page about SEO services, last updated 2026-06-05.

Your project pages are the keywords, and most of them have no text

The queries that bring an architecture practice work are not "architect near me". They are a building type crossed with a place or a constraint: passive house architect in a named city, adaptive reuse of a warehouse, ADU plans for a sloped lot, historic district review, mass timber for a four-storey infill.

Those words belong on project pages. On most firm sites, the project page is a title, a one-line location, and thirty photographs.

Google's SEO starter guide (checked 2025-12-10) is explicit about what that costs you: "The text that's near images can help Google better understand what the image is about and what it means in context to your page." With no text near the images, there is nothing for that sentence to act on.

A project page that can rank describes the project the way you would describe it to a client who has not seen it:

• The program in plain words, with numbers. Single-family, 2,400 square feet, two-storey addition to a 1912 bungalow.
• The constraint that made it hard. A setback, a flood zone, a load-bearing party wall, a design review board.
• The materials and systems by name. Cross-laminated timber, mineral wool, a heat pump, triple glazing.
• The place and the process. The city, the permit path, how long the review took.
• The consultants, by discipline.

That is 250 to 400 words. It is not copywriting, it is the information you already have in the project file, written once.

To find out how much text a crawler currently finds on a project page, strip the tags and count the words:

curl -s https://example.com/projects/warehouse-loft \
  | sed -e 's/<[^>]*>/ /g' -e 's/&[a-z]*;/ /g' \
  | tr -s '[:space:]' '\n' | grep -c .

Under 150 means the page is a gallery with a caption. The same check, run across every project URL in your sitemap, usually finds the same number everywhere, because the template is the problem and not any one project.

One limit to keep in mind: that command reads the HTML your server sends, not the page after scripts run. If your gallery injects the description client-side, the count will be wrong in a way that matters. The section on rendering below deals with that case.

The same reasoning applies to any visual craft where the portfolio is the product. We wrote out the technical half of the same problem for a neighbouring design practice in SEO for interior designers.

There is no schema.org type for an architect

This is the single most repeated piece of bad advice in this niche, and you can check it in ten seconds:

for t in Architect ArchitectureFirm ProfessionalService LocalBusiness; do
  printf '%s %s\n' "$t" "$(curl -s -o /dev/null -w '%{http_code}' https://schema.org/$t)"
done

Run on the day this was published:

Architect 404
ArchitectureFirm 404
ProfessionalService 200
LocalBusiness 200

schema.org/Architect does not exist. Neither does ArchitectureFirm, ArchitecturalService or ArchitectureDesign. Any guide that tells you to mark up your practice as an Architect is telling you to publish a type no parser recognises. Some trades get the opposite answer: the vocabulary models a garage precisely, which is where SEO for auto repair shops starts.

ProfessionalService does exist, and carries this note on its own page: "The general ProfessionalService type for local businesses was deprecated due to confusion with Service." A deprecated type is not a crime, but it is not what you want in a fresh template.

Google's local business documentation says to "Use the most specific LocalBusiness sub-type possible; for example, Restaurant, DaySpa, HealthClub, and so on." For an architecture practice, the most specific sub-type that actually exists is LocalBusiness itself. That is the honest answer, and it is enough: the LocalBusiness rich result requires only name and address.

{
  "@context": "https://schema.org",
  "@type": "LocalBusiness",
  "name": "Fenwick & Ault Architects",
  "url": "https://example.com",
  "image": "https://example.com/img/studio.jpg",
  "telephone": "+1-503-555-0138",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "412 SE Ankeny St",
    "addressLocality": "Portland",
    "addressRegion": "OR",
    "postalCode": "97214",
    "addressCountry": "US"
  },
  "areaServed": ["Portland, OR", "Vancouver, WA"],
  "knowsAbout": [
    "passive house design",
    "adaptive reuse",
    "accessory dwelling units",
    "historic district review"
  ]
}

areaServed and knowsAbout both exist and both resolve. knowsAbout will not rank you for passive house design, but it states in machine-readable form what the practice does, next to a page that proves it.

additionalType also exists, and it is the property designed for exactly this gap: a URL from an outside vocabulary that names a type schema.org lacks. If you use it, point it at an identifier you have loaded in a browser yourself. No Google rich result depends on it, so an invented URL buys you nothing and costs you accuracy.

Two things not to do. Do not mark a photo gallery up as an Article, because it is not one. And do not expect markup to move you: Google's own words are that "Google does not guarantee that features that consume structured data will show up in search results." Structured data makes a page legible, it does not make it better. If your pages are not appearing at all, the cause is usually elsewhere, and we walked through that diagnosis in why your website is not showing up on Google.

Images are most of the site, so treat them like pages

An architecture site is 90 percent photography by weight. Google's image documentation is short and specific, and almost nothing in this niche follows it.

On filenames, Google says to "use filenames that are short, but descriptive. For example, my-new-black-kitten.jpg is better than IMG00023.JPG." Most firm sites publish the camera's output. Find yours:

curl -s https://example.com/projects/warehouse-loft \
  | grep -oE 'src="[^"]+\.(jpg|jpeg|png|webp)"' \
  | sed 's/src="//;s/"//' | sort -u

If the list reads DSC_4471.jpg, you are publishing a file name that describes nothing.

On alt text, Google asks you to "focus on creating useful, information-rich content that uses keywords appropriately and is in context of the content of the page" and warns against stuffing. For a project photograph that means the sentence a person would say:

On placement, "make sure images are placed near relevant text and on pages that are relevant to the image subject matter." This is the structured-data version of the previous section: the text you wrote about the project is what gives the photographs meaning.

On discovery, "provide the URL of images we might not have otherwise discovered by submitting an image sitemap." Firms with galleries built by scripts, or images served from a CDN, are the exact case this was written for:

<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"
        xmlns:image="http://www.google.com/schemas/sitemap-image/1.1">
  <url>
    <loc>https://example.com/projects/warehouse-loft</loc>
    <image:image>
      <image:loc>https://cdn.example.com/warehouse-loft-street-elevation.jpg</image:loc>
    </image:image>
    <image:image>
      <image:loc>https://cdn.example.com/warehouse-loft-clt-ceiling.jpg</image:loc>
    </image:image>
  </url>
</urlset>

There is a second reason to care about image metadata, and it is money rather than ranking. Architectural photography is commissioned, expensive, and copied constantly. Google's image license structured data (checked 2025-12-10) requires contentUrl plus at least one of creator, creditText, copyrightNotice or license, and recommends acquireLicensePage:

{
  "@context": "https://schema.org",
  "@type": "ImageObject",
  "contentUrl": "https://cdn.example.com/warehouse-loft-street-elevation.jpg",
  "creditText": "Photograph by Mireille Dax",
  "creator": { "@type": "Person", "name": "Mireille Dax" },
  "copyrightNotice": "Fenwick & Ault Architects",
  "license": "https://example.com/image-licence",
  "acquireLicensePage": "https://example.com/image-licence"
}

The same fields can be embedded in the file itself as IPTC metadata instead. Google's page says plainly: "Alternatively, you can embed IPTC photo metadata directly inside an image." If you want the licensable badge through that route, "You must include the Web Statement of Rights field for your image to be eligible to be shown with the licensable badge."

Weight is the last image question. The recommended Largest Contentful Paint target is "2.5 seconds or less" at "the 75th percentile of page loads, segmented across mobile and desktop devices". On an architecture site the LCP element is nearly always the hero photograph, since the qualifying elements include <img>, <video> poster frames, and elements with CSS background images. A 6 MB hero at full resolution is the one performance problem worth fixing before any of the rest.

The gallery that renders nothing

Many practice sites are built on a template whose gallery, and sometimes whose project text, arrives by JavaScript. Google handles that, with a caveat worth knowing exactly.

From Google's JavaScript SEO documentation (checked 2026-03-04): "Google processes JavaScript web apps in three main phases: 1. Crawling 2. Rendering 3. Indexing." And the failure case, in Google's words: "Some JavaScript sites may use the app shell model where the initial HTML does not contain the actual content and Google needs to execute JavaScript before being able to see the actual page content."

So the content is not invisible, it is conditional. Rendering is a separate step that happens on Google's schedule, not yours.

The check Google recommends is to "use the Rich Results Test or the URL Inspection Tool and look at the rendered HTML". Do that on one project page, and compare what you see against the word count from the curl command earlier. Three outcomes:

• Both counts are healthy. Nothing to do.
• The source is thin and the rendered HTML is full. It works, but your text depends on a render pass. Move the project description into the server-rendered HTML if your platform allows it.
• Both are thin. The text does not exist anywhere. Write it.

Two related traps. Lazy loading applied to every image, including the one above the fold, hides content and hurts LCP at the same time; Google asks that lazy loading follow its own guidelines. And a project brief that lives inside a PDF or inside the photograph itself is text that the page does not have.

If your site is a custom build rather than a template, the implementation details for the App Router are in Next.js SEO. If it is a hosted builder, the same principles apply through whatever code-injection panel the platform gives you.

Where the local half matters, and where it does not

Plenty of small practices work from a studio nobody visits, or from a spare room. Google's rules for that case are written down, so there is no need to guess.

A service-area business is "A business that visits or delivers to customers directly but doesn't serve customers at their business address", and the instruction is direct: "If you don't serve customers at your business address, remove your address from your Business Profile." If clients do come to your office for meetings, keep the address and the opening hours. If they never do, publishing a home address buys you nothing and cannot be undone easily.

City pages are the other local question, and the honest version is narrow. One page per city where you have actually built something, with the project named and linked, reads as a record. Forty pages generated from a city list with the place name swapped is a doorway set, and it is the single most common way a practice site gets worse rather than better.

The split between your Business Profile and your site is worth getting right once: the profile wins the map pack, the site wins the typology query. We set out the profile side in Google Business Profile optimization, and the construction-side version of the same local problem in SEO for contractors.

What to measure, including the report most firms have not opened

Search Console is free and answers the only question that matters: which queries already show your pages, and where.

Open the performance report, filter the page dimension to /projects/, and sort queries by impressions. What usually appears is a set of typology and place queries sitting at position 20 to 40 with a handful of impressions. Those are the pages where writing 300 words of description is worth more than any new page, because Google has already decided the page is relevant to the query and only ranks it low.

The second report is newer and almost nobody in this field has looked at it. The generative AI performance report covers "AI Overviews and AI Mode", and its single metric is defined as: "Impressions are how many times links to your site were shown to a user in a generative AI feature on Google Search."

Read the limits carefully, because they change how you use it. The dimensions are pages, countries, dates and devices. The metric is impressions. There are no clicks, no CTR, no position and no queries. It tells you presence, not traffic, and treating it as a traffic report will mislead you.

Getting into those answers needs no new file. Google's guide to generative AI features (checked 2026-07-10) says "Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add", and goes further: "You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search (including its generative AI capabilities), as Google Search itself doesn't use them."

What it does ask for is the one thing an architecture practice has in abundance and rarely publishes: "Don't just recycle what others on the internet have already said, or could easily be produced by a generative AI model." The permit timeline you actually experienced, the assembly that failed the blower door test, the cost per square foot on a real adaptive reuse. That material cannot be generated, which is precisely why it gets cited. The broader mechanics of being quoted by assistants are in answer engine optimization.

Finally, the timeline. Google's starter guide: "Some changes might take effect in a few hours, others could take several months. In general, you likely want to wait a few weeks to assess whether your work had beneficial effects in Google Search results." Anyone selling you a date is selling you something Google declines to promise.

An order of operations for a practice with no marketing hire

Four hours a week, in the order that pays first. The sequence matters more than the hours: the first two weeks are rewriting pages Google has already found, which is where the fastest movement tends to be, and the later weeks are additions.

Two things not on that list. There is no link building, because the pages that rank for this query do it on no measured links. And there is no blog in week one: on a practice site, project pages carry the typology queries, and a blog only earns its place once they are written. If you do add one, put it on your own domain in a subfolder, for the reasons set out in subdomain vs subfolder.

If you want the weekly loop itself rather than a one-off cleanup, we wrote it down in the SEO workflow for a one-person team.

FAQ

What are the best SEO keywords for architects?

Not a list, a pattern: building type plus place, or building type plus constraint. Passive house architect in your city, adaptive reuse architect, ADU architect, historic renovation architect, mass timber architect. These sit at low difficulty because national brands have no reason to compete for them and because they need a real project to answer credibly.

Find yours from Search Console rather than from a generic list: the queries already showing your project pages tell you which typologies Google associates with your site. The method we use for a visual portfolio is written up in SEO keywords for photographers, and it transfers directly.

How do architects get clients from Google search?

Through three different surfaces that need different work. Brand searches (your practice name) are won by having a site at all. Map pack searches are won on your Business Profile. Typology and place searches are won on project pages, which is where most firms have nothing indexable.

Search is also not the whole picture for this profession. Referral, repeat clients and local reputation move more work for most practices than organic traffic does. The useful framing is that search catches the fraction of demand that starts with a stranger typing a building type, and that fraction is worth owning precisely because competitors have left it empty. Our general argument about when the channel is worth the hours is in is SEO worth it.

Do architecture firms need a blog?

Not before the project pages are written. A blog on an architecture site usually produces posts about awards and office moves, which nobody searches for.

What does work is the project page treated as the article: the brief, the constraint, the numbers, the outcome. If after that you want a blog, write the pieces clients actually ask you about, like what a design review board wants or what an addition costs per square foot in your city.

Should an architecture practice hire an SEO agency?

It depends on which half of the work you mean, and the answer is more honest if the price is on the table. An agency is usually worth it for sustained content production and for link acquisition, which are the parts that need somebody else's hours every month.

The work in this article is not that. Rewriting five project pages, fixing image filenames and adding one JSON-LD block is a weekend of your own time, and no agency will know your projects well enough to write those descriptions better than you do. Pay for the recurring half if you want to, after the one-off half is done. The ranges actually billed in this market are broken down in how much SEO costs.

Does structured data make project pages rank higher?

No. Google's position is that "Google does not guarantee that features that consume structured data will show up in search results", and its generative AI guidance adds that no special markup is needed there either.

Structured data makes a page unambiguous to a machine: this is a business, here is its address, here is who holds copyright on this photograph. That is worth the twenty minutes. It is not a ranking tactic, and any guide that sells it as one is guessing. A full page-by-page review of what each page is missing is a different exercise, and we describe it in how to run an SEO content audit.

Can a small practice show up in AI Overviews?

It can appear as a source, and Google now reports whether it does. The generative AI performance report shows impressions for AI Overviews and AI Mode, by page, country, date and device, with no clicks or queries attached.

What gets a page used as a source is the same thing that earns a citation anywhere: material that is not available elsewhere. For an architecture practice that means the specifics of a built project, written out. A page that restates general advice about hiring an architect is competing with every other page that restates it, and Google's own guidance tells you not to write that page.