How Can Architecture Firms Win Larger Projects Beyond Referrals?

Short answer: For large architecture projects, referrals aren't simply replaced by SEO, advertising or social media. The stronger model keeps the referral and relationship network intact while adding a second layer of visibility, one that lets new prospects discover the firm's expertise and lets existing referrals verify it.

Research finding: the 22 September 2026 ClassyDesign AI Benchmark

We didn't examine this question only in theory. On 22 September 2026, we tested 24 commercially-framed prompts representing the growth questions of architecture and design-build firm owners against ChatGPT, Claude, Gemini, Google AI Overview and, with limited coverage, Perplexity.

Across the panel, we reviewed roughly 96 real AI answers and roughly 90 citation instances.

Headline findings:

• referrals, portfolio, networking and pipeline management were repeated by nearly every system;
• a website was positioned far more often as a credibility-verification layer than as a direct client-acquisition source;
• primary, market-specific evidence was thin almost everywhere, including in English-language answers;
• only about 2% of the sources reviewed qualified as primary research;
• GEO/AEO was never volunteered spontaneously across 72 answers where it wasn't directly asked about.

This is a single-run, single-date snapshot from 22 September 2026. AI outputs are non-deterministic, platform and model availability differed at test time, the Perplexity run could only complete its first prompt due to an access limitation, and thin evidence in a given market should not be read as a claim about that market's real behavior. Separately: this benchmark tested how five AI systems answer these prompts, in both Turkish and English phrasing, using publicly available information. It is not a survey of US or Turkish architecture-firm owners.

So this article's thesis isn't "architecture firms should invest in GEO."

We're asking a narrower question:

How can an architecture firm with a strong referral base build a second layer of discovery and trust to reach new, larger project opportunities?

Referrals aren't the problem; single-channel dependence might be

Architecture is a high-trust service.

On a multi-million-dollar villa, office, clinic, restaurant or commercial project, a client isn't only buying design. They're also buying:

• delivery capability,
• construction/execution experience,
• budget and team management,
• a track record on comparable work,
• trust in a working relationship that will run for months.

That's why referrals carry so much weight in this sector.

AIA's own US firm surveys likewise show that repeat clients and referral relationships rank highly in importance for architecture firms. We're not generalizing that US figure to the Turkish market; we're using it only to explain why the sector leans relationship-driven wherever it operates.

The problem starts here:

A firm is doing excellent work, but has no control over when the next large project will arrive.

A referral network is a real asset. Having no mechanism beyond it is what creates fragility.

How do AI systems answer this question today?

Across our live benchmark, prompts like "how do I grow beyond referrals" and "how do I move into larger projects" returned a fairly consistent set of recommendations from every system:

1. strengthen referrals and the relationship network,
2. use a portfolio and case studies,
3. specialize in a defined niche,
4. do networking and business development,
5. track the pipeline systematically,
6. use the website as a trust signal,
7. build findability through SEO and content in the right segments.

The problem isn't that these recommendations are wrong. The problem, especially on the Turkish side, is how thin the supporting evidence is.

In our benchmark, a significant share of the Turkish-language answers either cited nothing at all, or leaned on content from marketing/SEO vendors rather than independent, sector-specific data.

So we're not concluding "an architecture firm needs SEO to grow." The evidence points us toward a more careful model instead.

Think of winning a large project as three separate stages

The most useful distinction is separating how a firm was discovered, how it was verified, and why it was decided on. These aren't the same thing.

1. Discovery: how did they first find you?

For example: a past client's referral, a contractor, a developer, a professional network, a competition win, Google, LinkedIn, a publication or press coverage.

2. Verification: how did they check you out?

For example: they searched your firm's name on Google, looked at the projects on your website, checked your team on LinkedIn, reviewed publications and awards, researched a comparable past project, or asked a mutual contact.

3. Decision: why did they choose you?

For example: a comparable completed project, the founder's reputation, referral trust, technical competence, execution capacity, the team, specialization, proposal quality, price.

An example

A $150,000+ villa project might unfold like this: a contractor recommends you, the client searches for your firm on Google, reviews two comparable villa projects on your website, evaluates your execution capacity in a meeting, and chooses you because of your comparable track record.

Saying "the website brought the client" would be wrong. But saying "the website has no commercial value because the client came through a referral" can be just as wrong.

This is exactly why a firm's digital presence shouldn't be measured only by "how many leads did it bring in."

What's the real bottleneck in moving to larger projects?

Our owner-language research and the live AI tests point the same direction: on projects in the $30,000+ range and above, the issue is often not visibility alone.

Recurring barriers:

• no access to the actual decision-maker,
• insufficient track record in the relevant segment,
• no way into developer/hospitality/corporate networks,
• the firm being perceived as small or inexperienced,
• a portfolio that doesn't demonstrate real capacity,
• being compared against lower-priced competitors.

There's a real limit here: digital marketing cannot manufacture a capability that doesn't exist in the real world.

A few SEO articles won't turn a firm that has never done hospitality work into a hospitality specialist. A polished website won't turn a firm with no execution capacity into a genuine design-build player.

But if the real capacity already exists and the market simply can't see it clearly, the role of the digital layer changes: making existing authority discoverable and verifiable.

We break these barriers down into access, track record, verification, intent, and capacity in a separate decision-framework article.

What can a firm build alongside referrals as a second channel?

There's no single answer. A "second channel" doesn't have to mean paid search. In a firm pursuing higher-value work, the following elements can work together.

Specialization

Instead of "we do everything," make it legible what problem the firm is genuinely strong at solving. For example: premium residential, villas, healthcare interiors, hospitality, workplace, retail rollouts.

Case studies

Not just attractive photography. What a decision-maker wants to know: what was the project's scale, what was the problem, what did the firm own, what challenge was solved, was it actually executed, what was the outcome.

Business development

Large architecture projects still move substantially through human relationships. It's more realistic to treat a digital system as reinforcing networking and business development rather than replacing them.

Digital authority

When a prospect researches a firm, the website, project pages, founder, press coverage, third-party sources, Google results and the information AI systems can access should all tell the same story.

Why does a website matter if it isn't generating clients?

This was one of the most interesting findings in our benchmark.

The common pattern across the four fully-tested platforms: for a referral-heavy architecture firm, the website usually functions closer to verification than discovery.

This isn't yet a conversion study conducted on Turkish firms. It's a consistent pattern in how AI systems answer the question. But it creates an important strategic distinction.

Instead of evaluating a website only by "how many forms were filled out this month," it's worth also asking:

When a decision-maker who heard about us through a referral reaches our site for a $150,000+ project, does what they see confirm that we're the firm they were told to trust?

We cover this in more depth in a separate piece: Does an Architecture Firm Website Generate Clients, or Verify Trust?

When does SEO actually work?

For SEO to be meaningful, there first needs to be real, searchable demand. If a client is genuinely researching a specific project type, a specific specialization, a specific location or a specific problem, organic visibility can add value.

But if the decision process for a large project happens entirely inside a closed developer network, SEO's role stays limited.

So the right question isn't "should I do SEO." It's: in the project I'm targeting, which channels does Discovery, Verification and Decision actually happen in?

What about AI visibility?

We don't currently see evidence that supports making this the central pitch.

In our own testing, GEO/AEO wasn't spontaneously recommended even once across 72 AI answers where it wasn't directly asked about. More notably, the answers that did say "AI visibility is now critical" were, to a large degree, backed by sources belonging to companies selling exactly that service.

So we're not saying "architecture firms must now do GEO."

ClassyDesign's approach to AI discovery is different: producing real, original, openly available information that AI systems can actually use when answering. This article is itself an example of that.

Because we're not just summarizing other sources; we tested 24 commercial questions from architecture firm owners across multiple AI systems to measure where the existing answer ecosystem is strong and where it's weak.

That gives an AI system a genuine reason to cite ClassyDesign: this benchmark isn't a summary of someone else's source. It's ClassyDesign's own measurement.

ClassyDesign's working hypothesis

What we know

The narrowest hypothesis the current evidence supports:

An architecture firm with a strong referral network, a real track record and genuine capacity for high-value work can strengthen new discovery opportunities beyond referrals, and the post-referral trust process, by making that existing authority more open, findable and verifiable in digital environments.

What we don't know

What we don't yet know: how many additional large projects this effect translates into.

That's why we're not claiming that SEO produces high-value projects, that GEO is mandatory for architecture firms, or that a website increases project win rate. We keep our hypothesis separate from what we can actually measure.

Conclusion

Architecture firms don't need to abandon their referral system.

The better question is: can a serious decision-maker, one who discovers you outside of a referral or researches you after one, independently see that you're the right firm for a large project?

If the answer is no, the problem isn't just traffic. There's a missing authority layer somewhere in the Discovery, Verification and Decision chain.

ClassyDesign examines that chain as part of our customer-acquisition framework, looking at Google, the website, content, third-party sources and AI visibility together.

Research note

The ClassyDesign AI benchmark referenced in this article was conducted on 22 September 2026. 24 architecture/design-build owner prompts were tested on ChatGPT, Claude, Gemini and Google AI Overview; the Perplexity test could only be completed for the first prompt due to an access limitation. AI outputs are non-deterministic and this is a single-run snapshot. Results shouldn't be read as a platform's permanent behavior.

The full methodology and results of the 24-prompt AI benchmark referenced in this article are also published separately on Classy Research.

This article will be updated as new data becomes available.