How architects research building products: the 7-step workflow and what decision-ready data looks like (2026)

83% of architects want more time to research building products, according to the AIA 2023 Journey to Specification report. Only 6% of architects currently use AI in their specification workflow. More than 80% are personally responsible for finding products for their projects — it is a core job function, not a secondary task. The bottleneck is not motivation. The bottleneck is that the 7-step manual research workflow is fragmented, rep-dependent, and missing pricing at every stage.

What is the standard 7-step workflow architects use to research building products?

The standard architect product research workflow is a 7-step process that begins with passive inbound information from manufacturer representatives and ends with a specification section in MasterSpec or SpecLink. Each step involves a different tool, a different friction point, and a different risk of stale or incomplete data.

1. Manufacturer rep visits and AIA CE lunches: 92% of architects aged 43 and younger prefer in-person lunches as their preferred method for learning about new products. Reps provide samples, cut sheets, pricing estimates, and project references. The limitation: each rep represents one brand. There is no neutral comparison.

2. AIA MasterSpec / Deltek Specpoint: MasterSpec contains 50,000+ vetted product listings. Most architects use the platform from schematic design through CDs. Cost: $1,200–$5,000/year.

3. ARCAT: Free building product database with CAD details, BIM objects, and CSI-formatted master specifications. Widely used for product discovery. The limitation: product data currency depends on manufacturer update frequency.

4. SpecLink (RIB Software): A subscription specification editing platform with a pre-formatted CSI master library. Used primarily for CD-phase spec editing.

5. CSI MasterFormat discipline: Architects organize product research by CSI division number. Understanding MasterFormat structure is a prerequisite — a specifier who does not know that acoustic ceiling tiles live in Division 09 cannot efficiently search any product database.

6. Manufacturer websites and literature: Over half of architects rely on technical data, product specs, design guides, CAD/BIM files, and pricing from manufacturer websites. Only 65% are satisfied with manufacturer website usability.

7. AIA continuing education (AIA CES): 83% of architects use CE courses and webinars to stay current on new products. CE is the most common formal learning channel — but it is manufacturer-funded, which means it carries the same rep bias as step 1.

What data does an architect need before a product can be specified?

Decision-ready product data is a 10-point checklist. Products that cannot supply all 10 items quickly are unlikely to be specified by firms without a prior manufacturer relationship.

Only 66% of architects are satisfied with how well manufacturers keep product information current. Only 65% are satisfied with manufacturer website usability. These two satisfaction scores explain why 77% of architects want manufacturers to proactively keep them informed — the pull research workflow is too slow and too incomplete to rely on.

How does rep bias shape product selection?

Rep bias is the structural distortion in architect product knowledge caused by relying on manufacturer-funded information channels. Manufacturer representatives are the most accessible source of product data — and they represent one brand. There is no neutral third-party service that can compare, for example, five acoustic ceiling tile brands across NRC, CAC, cleanability, cost, and FGI compliance in a single query. Architects must synthesize across multiple vendor conversations, which takes time most firms do not have.

The AIA 2023 data shows: 83% say they need more time to research products. Fee compression and lean staffing mean product research is under-resourced. Junior staff do initial shortlisting; senior architects do final review. The specification that reaches CDs is often a senior architect’s previous standard spec, updated based on what the most recent manufacturer rep presented.

Specification conservatism is the rational response to this environment. Every product an architect specifies carries professional liability if it fails. Specifiers prefer proven products with long track records over newer alternatives, even if the new product performs better on paper. Breaking into a standard spec requires a rep relationship, a peer recommendation, or an owner mandate — not just superior performance data.

What does AI currently do in the product research workflow?

AI adoption in specification is nascent but accelerating. As of the 2023 AIA survey, only 6% of architects regularly use AI in their workflow. 20% expect to use AI for product research in the near future. 25% of principals and 33% of project managers expect AI adoption — indicating that adoption will be driven top-down, not by junior staff.

Current AI applications in product research:

• AI-assisted specification drafting: NBS Chorus and emerging tools generate first-draft specification sections from natural language descriptions. Input: “write a spec section for a fiberglass acoustic ceiling tile meeting FGI 2022 for semi-restricted areas.” Output: a CSI-formatted draft ready for editing.
• AI search and comparison: Tools like datadrivenaec.com build AI agents that cross-reference NRC, CAC, cleanability, cost, and FGI compliance across multiple brands in a single query — the neutral comparison that manufacturer reps cannot provide.
• EPD analysis: AI tools can parse EPD PDF documents and extract GWP, AP, EP, and PENRT values for direct comparison. A task that takes a human 30–60 minutes per document takes an AI agent seconds.

AI cannot reliably verify current pricing or lead times — data goes stale too fast. AI cannot substitute for the relationship value of a trusted rep who solves a problem at 5pm on a Friday. AI cannot provide project-specific legal and liability guidance.

What will structured product data mean for specification?

Manufacturers who provide structured, machine-readable product data — JSON or schema-tagged — will appear in AI-powered search results. Manufacturers who bury specifications in non-searchable PDFs will be invisible. This is not a prediction — it describes the current state of how AI agents query product databases. Structured product data aligned with buildingSMART and IFC standards is becoming a specification prerequisite in owner-driven sustainability programs and large-firm standard processes.

The 2026 AIA Journey to Specification report examines how U.S. manufacturing trends, sustainability priorities, and AI adoption are reshaping product research. The report is expected to show AI adoption moving from 6% to materially higher use among project managers and principals who have the authority to change standard specifications.

Common mistakes in building product research

1. Stopping research when the first acceptable product is found. Consequence: The first-found product is usually the one the most recent rep presented. No performance comparison happens. Specifiers miss products that may outperform on CAC, cleanability, or lifecycle cost.

2. Accepting manufacturer-reported performance data without test documentation. Consequence: Self-reported NRC, DCOF, and fire ratings have no enforceability without third-party test documentation. Specification liability attaches to the specifier, not the manufacturer, when performance claims fail.

3. Treating “contact rep for pricing” as a normal part of the workflow. Consequence: No budget number at schematic design means cost is unknown until DD or CD. Change orders follow. Every product shortlist should include a budget range estimate before the design advances.

4. Downloading BIM objects without checking Revit version compatibility. Consequence: BIM objects from manufacturer databases are often outdated. An incompatible Revit family wastes time and may introduce geometry errors in the model. Check Revit version before downloading, not after insertion.

5. Using CE course knowledge as the primary product update mechanism. Consequence: AIA CE courses are manufacturer-funded. The course content reflects the manufacturer’s product line, not the competitive landscape. CE is a starting point, not a comparison tool.

How DataDrivenAEC automates product research

Comparing products across specs, cost, and compliance variables manually takes 10+ hours per section. DataDrivenAEC builds custom agents that run this research across your spec section and deliver the comparison as a structured report. See all agents →