Architecture firms get named in Google and AI answers by owning the pages a prospect's question actually matches, the practice-area, fee and process pages, and by earning mentions on the directories, associations and publications AI engines already trust. Posting more on social media will not do it. Neither will hoping referrals compound on their own.
This holds whether the question is typed into Google or asked of ChatGPT, Perplexity or Gemini, because both read the same pages and weigh the same third-party sources before naming a firm.
It will not name a firm with no indexed site, no professional-rules clearance to publish, or no page and no earned mention proving it has done the project type being asked about.
If prospects are already asking ChatGPT or Google's AI features which firm to call, see how AI search visibility works before you change anything else on the site.
Why Clients Ask AI Before They Ask You
AI engines answer a "which firm" question with a short list of firms, not a directory of every firm that could do the work, and they weigh independent evidence on pages and mentions the firm does not control more heavily than anything the firm says about itself on its own site.
A prospect with a renovation, an addition or a new build does not call three firms and compare proposals cold anymore. They ask Google, or they open ChatGPT, Perplexity or Gemini, and they ask which architecture firm handles projects like theirs in their area before they ever pick up the phone.
One industry survey found that 98% of US consumers verify unfamiliar AI-recommended brands before buying, so being named by the engine is the opening move, not the whole game. But it is the move that decides whether a firm gets considered at all. A firm with no page answering the prospect's actual question, and no mention anywhere the engine already trusts, does not get the follow-up search. It does not get the consultation.
This is the mechanism behind SEO for architects, building a firm's AI search visibility so it sits inside that short list instead of outside it. The rest of this guide breaks the mechanism into the two pieces a firm can actually build, the pages and the mentions, then gives a repeatable test for finding exactly where your firm's gaps sit today.
Partners who have run a firm for years feel this without naming it. A past client still forwards your name to a friend, but the friend does not call blind. They search the name first, and more of them now ask an AI assistant to confirm or compare before they pick up the phone. A firm with nothing for either check to find watches the referral stall before it reaches the phone.
The Pages Prospects And AI Engines Actually Read
A prospect's question maps to one of four page types, practice-area and issue pages, fees and process pages, professional bios, and office pages, and if the firm does not have the matching page, nothing else about its marketing matters for that question.
- Practice-area and issue pages answer who handles this kind of project, a renovation of a landmark building, a multi-family addition, a net-zero retrofit.
- Fees and process pages answer what this costs and how it works, the question every prospect has before they will book a call.
- Professional bios answer who specifically would run my project, the named architect or principal behind the practice area.
- Office pages answer do you work in my area, for a firm with more than one location.
| Page Type | What The Prospect Is Asking | What It Must Contain |
|---|---|---|
| Practice-area and issue page | Who handles this kind of project | The project type named plainly, past work described in specifics, the process for that project type |
| Fees and process page | What this costs and how it works | A plain description of fee structure and project stages, even without exact numbers |
| Professional bio | Who would actually run my project | The named architect, their specific project history, a direct way to reach them |
| Office page | Do you work in my area | The office address, the local projects, the licensing that applies in that location |
None of this requires special markup. Google's AI features guidance confirms that standard SEO applies and a page only needs to be indexed and eligible to appear in a snippet to be read and potentially cited. The barrier is not technical. It is simply having the page at all.
What differs from a classic search result is scope. Google's AI features run a query fan-out across the related subtopics around a question rather than matching one exact phrase, so one practice-area page that actually answers the follow-up questions, timeline, permitting, typical budget range, does better than five thin pages chasing five keyword variants.
The writing on the page matters as much as its existence. Content framed as a direct answer to a question outperforms content framed around a keyword when an AI engine is deciding what to cite, because the engine is assembling an answer, not ranking a list of blue links. A practice-area page titled around the project type, written to answer the actual question a client would ask about it, reads the way the engine wants its sources to read.
The same goes for how a firm describes itself across every page. A distinct, consistent point of view and category description helps an AI tool tell the same consistent story about a firm every time it is asked, rather than a different half-answer depending on which page the engine happened to pull from. A firm that describes itself differently on its homepage, its bio pages and its project pages gives the engine inconsistent material to choose from, and inconsistency is its own kind of invisibility.
Building this set of pages well enough to answer a prospect's actual question, not just rank for a keyword, is close to a content production problem on its own. A firm that would rather have content that brings enquiries built for it than assign the writing to a partner has a service built for exactly this set of pages. For the mechanics of how Google's newer AI surfaces actually source and cite pages, how Google AI Mode works goes a layer deeper than this guide needs to.
Run The Test: Which Firms Does AI Name For Your Project Type
You do not need a tool subscription to find out where your firm stands today. You need twenty minutes and the question a client would actually ask.
- Write out the real questions. Not "best architecture firm," but what a client would actually type or ask out loud, "who should I call about a historic renovation in my city" or "which firm handles net-zero multi-family projects near me."
- Ask the same question across Google and two AI assistants. The same buyer question gets different answers from different AI engines, so testing only one tells you about that one tool, not about your visibility overall. Run each question through Google's own AI features and through at least two of ChatGPT, Perplexity, Claude or Gemini.
- Record which firms get named and which pages get cited. Note every firm mentioned, not just the competitors you already track, and note which specific page on each firm's site the engine pointed to, if it names one.
A firm running this for the first time is often surprised by who shows up. It is rarely only the obvious competitors down the street. A firm three states away with a sharply written page on exactly the project type being asked about can out-rank a well-known local name that has never put the specifics in writing anywhere.
Once you have the list, map every gap to one page type. If no firm, including yours, showed up for a question, that is a page nobody has built yet, an opening. If a competitor showed up and you did not, go find the page the engine cited. Nine times out of ten it is a practice-area page, a bio, or a project write-up that answers the specific question better than anything on your own site.
It helps to know what you are testing against. An AI chatbot answers either from what it learned during training or from a live search it runs in the moment, and that live-retrieval path works much like classic search engine optimisation, rewarding the same indexed, well-structured pages that rank well in Google. A firm with strong classic SEO is already most of the way to showing up in the live-retrieval answers too.
Run this test monthly, not once. AI conversations are private and personalised to whoever is asking, so there is no dashboard anywhere that tells a firm how often it gets recommended in other people's conversations. This manual test, repeated, is the closest substitute a firm has, until it sets up the kind of ongoing tracking covered later in this guide. For a deeper look at the mechanics behind what gets cited and why, getting cited in AI answers covers the topic beyond what this test needs.
Earn The Mentions AI Engines Already Trust
The pages a firm builds are only half of what an AI engine weighs. The other half is what other people and publications say about the firm, on sites the firm does not control.
Call this the earned-mentions layer, the directories, rankings, association listings and trade-press mentions that exist outside the firm's own site. None of it is paid placement and none of it is the firm's own copy, which is exactly why an AI engine treats it differently. A quote from an executive, a trade press mention or an analyst reference carries more weight with an AI engine than anything a company says about itself, because none of those sources has a reason to flatter the firm.
Clients respond the same way these engines do. The same survey found that 69% of US consumers preferred a brand with trusted press coverage over an otherwise similar brand with none. A firm with a mention in a trade publication and a listing in a professional association's directory is not just more visible to an AI engine, it reads as more credible to the person on the other end of the recommendation too.
This is deliberately generic advice, and on purpose. A firm's own professional rules decide which specific directories, rankings and associations fit its practice, and that list looks different for a residential firm than for one doing institutional or civic work. The pattern holds across all of them even when the names do not. Earned mentions describes exactly this kind of ongoing work, pursuing mentions on professional directories, rankings, associations and trade publications relevant to a firm's own practice, for firms that would rather have someone pursuing it than let it happen by accident or not at all.
Start with whatever is already true and undocumented. Most firms have handled a project type, won an award, or had a principal quoted somewhere, and never linked it back to the firm's own site in a way an AI engine would find. Before chasing a new mention, make sure every existing one is actually connected to the firm's name online. A rankings listing that spells the firm's name differently from its own site, or a trade-press mention never linked from the firm's own project page, is a mention an engine has a harder time attaching to the right firm at all.
None of this needs to be dramatic. A short quote in a regional trade publication about a completed project, tied to the firm's name and linked back to its site, does more for AI visibility than a much larger budget spent entirely on the firm's own channels.
Engineering firms sit in an almost identical spot. The project types differ, but the mechanism, and the objections that stop a firm from starting, do not. How engineering firms earn the same mentions walks through the same earned-mentions pattern for a sibling profession.
If you would rather hand the page-building and earned-mentions work to a team that has done it for other professional firms, the GEO + SEO Growth Programme covers both. One of the founders runs the first call, and you leave knowing which page to build first.
Answer The Three Objections That Stop Partners Starting
Three objections come up in almost every conversation about this, and all three are reasonable. None of them actually blocks the work.
Our Partners Will Never Find Time To Write
No partner needs to write a page from scratch for this to work. TRM runs a 30-minute interview per topic, drafts the page from that conversation, and the partner reviews and approves it before anything publishes. The partner's contribution is thirty minutes of talking about a project type they already know cold, not a writing assignment on top of a full caseload.
Our Professional Rules Restrict What We Can Publish
This is true for every licensed profession, and it does not mean nothing can be published. The firm's own expert approves every page before it goes out, which means the content stays inside whatever boundaries the firm's own licensing board and professional conduct rules set, because the person who knows those rules best is the one signing off. That often means naming a project type and a process without naming a specific client, or describing a typical fee range without quoting an exact number. The firm decides where that line sits. The page gets built inside it, not around it.
Our Work Comes From Referrals
A referral is not the end of the decision anymore, it is the start of a search. A client who hears your name from a friend still opens a browser or asks an AI assistant to confirm the firm actually does this kind of work before calling. A firm with no page proving it has handled that project type gives the referral a reason to second-guess itself at exactly the moment it should be converting. This shows up most clearly with out-of-market referrals. A past client who moves cities and refers a friend to the old firm cannot vouch for it in person the way a local friend can, and the friend has only the firm's online presence to go on. Referrals and a strong online presence are not competing strategies. A referral that cannot be confirmed online is a referral that quietly dies before the first call.
CPA firms field the identical three objections, for the identical reason, licensed professionals with limited time and real publishing constraints. How CPA firms handle the same objections goes through the same three answers for a sibling profession, if it helps to see the pattern applied somewhere else first.
Track AI Visibility Against The Firms You're Actually Losing To
The manual test from earlier in this guide works, but running it by hand every month, across every project type a firm handles, against every competitor worth watching, is a lot to keep up alongside running a practice. AI conversations stay private and personalised to whoever is asking, so there is still no dashboard anywhere that shows a firm its actual recommendation rate directly. Ongoing tracking against named competitors is the closest workaround, run on a schedule instead of whenever someone remembers.
That is what monthly tracking against competitors is for. TRM tracks a firm's AI visibility every month against three named competitors, the same test from the previous section, run on a schedule and reported rather than left to whoever remembers to check. Technical fixes and a live dashboard sit alongside the tracking, so issues that quietly block an AI engine or a search crawler from reading a page get caught and corrected instead of sitting unnoticed for months.
Tracked this way, over time, a firm can expect to see movement on the outcomes that actually matter to a practice, reported month over month rather than promised upfront:
- Being the firm AI names when a prospect describes a project like the ones the firm actually wants.
- More consultations booked from search and AI, not just from referrals.
- Winning the matters and clients the firm most wants, not just whichever ones happen to call.
- A lower cost per consultation, as more of the pipeline comes from being found rather than from paid channels.
- Every office winning its own area, for a firm with more than one location, rather than one office carrying the rest.
- Knowing which pages actually bring in signed clients, so the next page built is an informed bet rather than a guess.
None of that is a guarantee attached to any single page or mention. It is what a programme run this way tracks and reports against, month by month.
For firms that want to go deeper on the Google side specifically, how to increase AI Overview visibility covers the mechanics in more detail. A firm that would rather run its own tracking than hand it to anyone can start with tools that track brand mentions in AI answers, which compares the options built for exactly this. Firms with more than one office can see TRM's Growth and Locations plans, with current pricing on the pricing page rather than repeated here. A firm not ready to commit to any of it can start with the free AI Visibility Report instead, a lower-commitment way to see where the gaps actually are before deciding what to build first.
What To Publish First, And In What Order
None of the above works without the basic machinery underneath it. A firm's website still has to function as an actual portfolio, built and technically sound enough to be indexed and found for high-intent searches, before any of the page types above can do their job. That groundwork is table stakes, not a differentiator, but skipping it undoes everything else.
With that in place, build in this order.
- The segment or industry page first. One page stating plainly what kind of firm this is and what kind of work it takes on, the page every other page on the site should eventually link back to.
- One practice-area page per service line. Each project type the firm actually handles gets its own page, answering the real questions a prospect would ask about that specific type of project.
- Fees and process pages next. The questions every prospect has before booking a call, answered plainly enough that the call starts past the basics instead of covering them again.
- Office pages last, for firms with more than one location. Each office gets its own page so a prospect searching locally finds the specific office, not a generic firm-wide page that could be anywhere.
A firm starting from literally nothing should not try to build all four layers in one sprint. Picking the single highest-volume project type and shipping its practice-area page first, then its fee page, then one earned mention to back it up, gets a usable test case live within weeks rather than waiting for the whole site to be finished before anything goes out.
Once that structure exists, the ongoing content work, the kind usually filed under content marketing or thought leadership, becomes maintenance rather than a cold start. A steady publishing cadence keeps the pages current and keeps new project types covered as the firm's work shifts, but it only compounds once the underlying pages exist to build on. Publishing commentary with no practice-area page behind it is building a second floor with no first floor underneath it.
This order is not specific to architecture. The order an MSP follows is nearly identical, segment page, then service pages, then the supporting pages, because the underlying mechanism, a prospect's question matching an indexed page, works the same way regardless of the profession behind it.
FAQs
How Much Partner Time Does This Actually Take Each Month?
Can We Publish Anything At All If Our Professional Rules Restrict What We Say?
Our Work Comes From Referrals. Do We Still Need This?
Which AI Assistants Should We Actually Test When Checking If Our Firm Gets Named?
Do Awards And Rankings Still Help Now That AI Is Doing The Recommending?
What Timeline Should We Expect Before We See More Consultations?
Do We Need A Separate Page For Every Practice Area, Or Can One Page Cover Several?
Won't Improving Our SEO Force Us To Compromise Our Site's Design?
How Is This Different From Local SEO Or A Google Business Profile Listing?
What's The First Page We Should Publish If We're Starting With Nothing?
Book a free 30-minute call and one of the founders will walk through where your firm stands today and which page to publish first.
