TAR

I’ve Narrowed It to Two Architects. How Do I Compare Them?

By Tina Modrusan

Regulatory information reviewed July 27, 2026

Where the shortlist runs out

Portfolios, interviews, and fees got the search down to two finalists. They can't get it from two to one — not because they don't matter, but because they're built to look similar across good firms, and both of these firms cleared them. What's left to settle is two separate questions that look like one: whether the design is genuinely good, and whether it will get built the way it looks. Different evidence answers each. The portfolio answers the first well and the second hardly at all.

Two firms remain. Both licensed, both with credible work, and nothing in the interviews or references knocked either one out. This is the hardest position in the whole search, and most advice abandons the search exactly here — it explains how to build a shortlist, then goes quiet once there is one.

The reason it goes quiet is that the two firms are no longer different in any way a checklist can catch. One shows a finished project that looks like the one wanted. The other seems to have done the harder version. One sells the principal; the other sells a team. Both proposals say design, documentation, approvals, construction. The words match. What sits under the words usually doesn't, and that gap is the whole comparison. When it closes, what's left is taste — not as the small call allowed after the serious work, but as the last live question between two architects who have both proven they can deliver.

Illustration 1:
Matched on the Surface

Two finalists line up on every visible category — portfolio, experience, team, references, schedule. What sits under the headings, and decides the project, is the part the shortlist never fills in.

Why two good architects are genuinely hard to tell apart

Because the words firms use to describe themselves are more standardized than the work those words describe. Three words do most of the damage.

Experience hides three different claims. It can mean the firm is old, or the principal across the table is seasoned, or the person who will draw the project every day has done this before. A firm founded in the nineties tells you little about whether the project architect assigned to the job has ever put a wet room over a neighbor's ceiling and made it hold. Which of the three is being sold is the thing to establish.

Communication hides two. In the interview it means the conversation was easy and the architect was sharp. On the project it means something operational and unglamorous: who writes decisions down, who chases the structural engineer, who answers the contractor at 4pm on a Friday, and who covers the three weeks the principal is unreachable in August. A great first meeting is evidence of the first. It is little evidence of the second.

Schedule hides where it ends. One firm's timeline runs to the day drawings are filed. Another's runs through the end of construction. Neither is lying. They are measuring different distances and calling both "the schedule," and the shorter one wins for being shorter when it's only stopping earlier.

None of this comes out on its own. The interview is built to make firms look comparable; the work is to make them legible instead, and legible means opening those three words up against one project.

What actually makes a past project relevant to yours

A past project is relevant when its hard parts match your hard parts — not when its photographs match a Pinterest board. The photograph proves the design is real: that question is answered, and it should be. What the photograph can't show is whether the hard part of your project is a hard part this firm has actually done, because the photograph is the deliverable a firm controls most tightly, and it survives largely intact no matter how the job went.

Picture two prewar renovations. The first is the apartment you'd have saved: beautiful, calm, exactly the right taste. The second is uglier in photos — but it combined two units, cut a structural opening, cleared the building's alteration review, and hit a fixed closing date without slipping. Both are genuinely well designed. For a combination on a deadline, the second architect has also done the actual project, and the first has done a beautiful room.

The hard part of a co-op or condo is that a private renovation runs through a building's own government, and that government is now part of the public record at the moment of filing. As of February 2, 2026, a DOB NOW: Build filing by a condo unit-owner or co-op shareholder cannot proceed until the board's representative attests, inside the system, that the board authorized it — the board itself now signs the filing, not just the shareholder. ^[1] That's the rule. Here's what the rule doesn't tell you, and what an architect who's lived it knows: the formal attestation is only one part of the review. Much of the project-specific work happens earlier, in the building's alteration-agreement review, where a self-managed

board with a part-time managing agent moves at the speed of whoever happens to read the submission first — and the difference between a firm that files into that rhythm and one that files against it is the difference between a clean approval and a slow one. An architect who has run a board approval talks about the managing agent by the second sentence. One who hasn't talks about the drawings.

The hard part of a brownstone is that the exterior isn't entirely yours to design, and the review that decides what you can do to it is invisible in any finished photo. In a New York City historic district, most changes to the front and rear façade require Landmarks Preservation Commission review — and so does broad in-kind replacement, like swapping a whole window or a façade element, even when the new one matches the old exactly. Only narrow repairs are exempt: caulk, matching paint, a replaced pane of broken glass.^[2] An architect whose brownstone went through LPC has sat across from the commission — and has learned the thing that matters most, which is that the schedule is won or lost on the second submission, after staff comments come back, not the first. A firm whose brownstone work never touched Landmarks can be perfectly good and still have never been tested on the part of the job that will hurt.

So the finished apartment proves the firm can design one — that question is settled, and settled well. Whether this team can deliver this one, on this building's terms and by the date the closing is set and the old place is already sold, is the other question, and the photograph says little about it.

Does "upstate experience" mean experience with your project?

Partly, and not in the way the phrase suggests. The code does vary town to town — a firm that knows Dutchess County hasn't, by that fact, filed in Ulster or the next town over, and local zoning, preservation, and environmental review sit on top of the state code and don't transfer.^[3] But any competent firm confirms local code before it draws. That's not where upstate projects go wrong, and treating it as the differentiator insults the reader and the architect both.

Two things are actually hard. The first is distance, and it tends not to show up until the project is well into construction, a few months in. A firm two and a half hours from the site, working with a contractor it has never built with, is fine through design and fine through filing — and then the project reaches the point where a question that takes an hour to answer in the city takes a week upstate, and the decisions that quietly assumed the architect would be in the room start getting made without one. A firm that has run a distant site has a cadence to it — a schedule of visits, a named local contact, a rule for what the contractor decides alone. A firm that hasn't calls the drive manageable and books the first visit for framing.

The second is who the architect can actually get to show up. Ask a firm that has built upstate how it lines up trades, and the experienced one answers in specifics — the people it has worked with, how far out they book, what it does when someone it was counting on goes quiet. A firm that has only designed upstate, drawn the house and handed it off, answers in generalities, because it has never had to get a crew it doesn't employ to a site two and a half hours away on the week the schedule needed them. The firm that has done it talks about the people; the firm that hasn't talks about the plan.

Who inside the firm will
actually run your project?

The firm and the team are two different pieces of evidence, and they have to be held apart, because on the day a wall opens up and the condition behind it isn't what the drawings assumed, it's the team standing in the room — not the firm's reputation.

The big firm's risk is the hidden team: the principal who runs the interview may hand the job to a project architect the moment the contract is signed and reappear only for the meetings that matter, and the drawings may come from people never met. The boutique or solo firm carries the mirror risk. Its real strength is that the principal genuinely is the person and genuinely is involved — but that same person can be stretched across too many jobs at once, and involvement means less when it's divided six ways. Both versions interview well. The question that exposes both is the same: who is actually on this project, and what do their next six months look like. If the team proposed for the job didn't work on the precedent shown, that precedent is the firm's, not theirs.

The team that shows up in writing — names, roles, daily contact — is the team being bought; the team introduced only in the interview is not. What separates them is not who is on the team but who answers when the principal is unreachable, and whether they have done it before.

Illustration 2:
Two Questions, One Project

The photograph proves the firm can design it, and proves it well. Whether this team can build this one — on this building's terms, by the date that can't move — is the question the photograph stays quiet on.

Are the two proposals even describing the same job?

Often, no — the same phase names cover different deliverables, different exclusions, and different stopping points, so two proposals can look parallel and describe genuinely different amounts of work.

Some vocabulary helps here. The AIA's standard owner-architect agreement, B101, sorts an architect's services into Basic, Supplemental, and Additional^[4] — and for a custom single-family house the AIA uses a different contract entirely, B110, paired with its own general conditions, A110, not the B101/A201 pair most people have heard of.^[5] Useful to know the map exists. But the map is not the territory, and that's the point: "construction documents" doesn't tell you what will actually be drawn, specified, or coordinated. "Approvals" doesn't name which submissions and consultants the architect handles versus punts back. "Contractor selection" can mean emailing out the drawings, or fielding bidders' questions, or comparing bids line by line, or sitting beside the client in the negotiation — any of those, under one word.

Read as headings, the two proposals rhyme. Read as responsibilities, they may not — and a shorter proposal might cover less, or might just say less. The only way to tell is to get both firms on a call and make them walk the words through: what's actually in "construction documents," where "approvals" ends, whether the schedule runs to filing or to final inspection. The short proposal that quits early and the short proposal that's merely terse read identically on paper and answer completely differently on the phone.

The most important place a proposal stops is the start of construction. "We stay involved" and a defined construction-phase role are not the same thing, and the difference is worth real money. Under the standard AIA framework, an architect's construction-phase role can include reviewing submittals, visiting the site on a schedule, reviewing the contractor's payment applications, interpreting the drawings when they're ambiguous, and formally rejecting work that doesn't conform.

The same framework draws a hard line the other way: the architect does not control the contractor's means, methods, or site safety — that's the builder's, always.^[7] A different contract can move these lines around, which is the entire reason "we'll be there for you" settles nothing until there's a clause that says what "there" means.

Construction is where every unknown in the drawings turns specific and expensive. A wall opens and the condition behind it isn't what anyone assumed. A substitution changes the look or the cost or the timeline. A consultant's number contradicts an architectural detail. Each of those is a decision, and each decision has an owner. The comparison that matters isn't which architect says they'll stay — both will. It's which one, in writing, will review, respond, and document, who inside their office does it, and what's left sitting with the client or the contractor when they don't. The broadest construction role isn't automatically the right one. The right one is the one understood before the first surprise came out of the wall — not the first time the contract gets read closely because something already went wrong.

One thing that can be verified independently, for free. On any NYC filing, the Registered Design Professional named as Applicant of Record is the one responsible for submitting that application and coordinating its documents — and a big job carries several, because construction, plumbing, and mechanical each file separately.^[6] That record is public and retrievable. It confirms who filed. It will not tell you who designed the thing, who drew it, or who was still answering the phone when the job hit trouble — so it checks a claim, it doesn't settle the choice.

And the budget: a construction number is only as comparable as the assumptions under it. Two firms can quote the same figure and mean different scopes, different contingencies, different levels of finish, different things excluded. A number stated with its assumptions is a price; a number stated without them is a placeholder that gets renegotiated later, usually once the work is already moving.

How much should the interview and the references actually count?

They're real evidence and they're both compromised, in opposite directions — the interview is a performance staged for the client, and the reference is a witness the architect chose. Both are worth using; neither is worth trusting on its own.

A good interview shows clarity, taste, and the ability to make a hard thing sound simple. But charisma is the wrong thing to read for. Architects are a particular kind of person — detail-obsessed and big-picture in the same breath, and often not natural salespeople — so a flat performance in the room is not evidence of a weak architect. The thing to read for is how they think, not how they present: whether the answer to a hard question is specific, and whether it survives a follow-up. The interview can't stage the things that actually decide a project — the email that goes unanswered for a week, the decision that sits open for three, the disagreement, the coordination that falls apart at 5pm, the question nobody thought to ask until the demo was already open — so a smooth room is worth very little and a sharp mind is worth a great deal.

References feel weak because the architect picked them — but a former client knows exactly what no photo or proposal can show: whether the principal actually stayed, how bad news traveled, what happened the day the project changed shape. The difference is in the questions. Whether the project actually resembled this one, because a glowing reference for a different kind of job proves little. What the client was afraid of going in, and how the architect handled it. And the one that does the most work: walk through the worst moment of the project — what went wrong, who moved first, how the decision got back to them. A reference asked whether they were happy reports satisfaction; a reference asked those three reports what happened. The architects who came through hard projects well have the story ready; the ones who didn't get vague.

And then, yes — taste, rapport, the sense of being able to stand this person for a year. That's a legitimate part of the choice, and residential work is a long chain of judgment calls that no matrix survives. The only rule worth holding: let taste decide between two architects whose evidence is already clear. Don't let it replace the evidence that never got checked.

Which difference should actually decide it?

The one that matches the hardest part of the specific project — because the same strength counts for a lot on one job and almost nothing on another.

For a combined apartment on a closing date, the deciding differences are structural and consultant coordination, the building's review, who owns the revisions, and whether the team that pitched is the team at the end. For an approval-heavy townhouse, it's the exact conditions of the precedent, who files, façade review where it applies, and how they behave once the contractor's on site. For a second home hours away, it's site attendance, a local contact, contractor communication, and how cleanly responsibility divides between the people in the city and the people on the ground.

Lined up against one project, usually one architect is simply more right for it than the other — not better in the abstract, better for this. When it's genuinely a tie on the evidence, that's when taste is allowed to break the tie, and it's a good tie to have, because it means the part almost nobody does got done.

The architect worth having isn't the most technical one or the most gifted one. It's the one who is both, and right for this project. Design judgment and delivery aren't a trade-off to pick between — they're two things to confirm, and the portfolio only confirms one of them. So the whole comparison comes down to three questions, asked against one project, that together confirm the other: who has actually done this specific hard thing, who will be in the room when a wall opens, and where does each proposal quietly stop. Portfolios, interviews, and fees can't answer those. They got the search to two. These three get it to one.

That last judgment — general competence versus fit to one specific project — is the whole premise of what The Architect Registry does: it evaluates architects against a single named project and recommends one, or none. The distinction in this article is the distinction it's built on. A shortlist proves competence. It hasn't touched fit. That work starts after the shortlist, and it's the work that decides how the year of construction actually goes.

Sources

^[1]NYC Department of Buildings, "Condo/Co-Op Board Added as a Stakeholder in DOB NOW: Build," service notice, effective February 2, 2026.

^[2]NYC Landmarks Preservation Commission, "Permits and Making Alterations." nyc.gov/site/lpc/about/permits-and-making-alterations.page. Verified live July 27, 2026.

^[3]NYS Department of State, James A. Coon Local Government Technical Series, "Administration and Enforcement of the Uniform Fire Prevention and Building Code" (revised 2026); 19 NYCRR Parts 1202/1203. dos.ny.gov/administration-and-enforcement-uniform-code. Retrieved July 27, 2026.

^[4] AIA Contract Documents, "B101–2017." aiacontracts.com. Verified current July 27, 2026.

^[5] AIA Contract Documents, "B110–2021" and "A110–2021." aiacontracts.com. Verified current July 27, 2026.

^[6] NYC Department of Buildings, "Plan Examination / Design Professionals / DOB NOW: Build FAQ" (Applicant of Record). nyc.gov/site/buildings. Verified live July 27, 2026.

^[7]AIA Contract Documents, "A201–2017." aiacontracts.com. Verified current July 27, 2026.

Regulatory and professional-source information was reviewed July 27, 2026. Requirements and standard documents change. Obligations depend on the property, jurisdiction, participants, and agreements involved.

THE ARCHITECT REGISTRY

Independent guidance for selecting a residential architect
in New York City and Upstate New York.

About · How We Evaluate · Articles · Project Inquiry