Kind: Measured report
By Asif Hussain, Founder and CEO, SpecIndex
In a Utah project manual for a technical college heat pump job, the approved manufacturer list runs to 184 names, and the section opens by inviting products from any of them. A rep who finds his brand on that list has won almost nothing, and that is the plain message of this count: in higher education, the project pipeline barely shows the campus work at all, and where the documents do name products, a place on the list is not a position.
Facts at a glance
- In a Utah project manual for a technical college heat pump job, the approved manufacturer list runs to 184 names, and the section opens by inviting products from any of them. (From: the opening paragraph.)
- We ran search_projects with project_type "education", which returned 6,229 projects in total. (From: Method note.)
- We then ran the same query with announced_within_days set to 660, which returned 4,171 projects, and paged the full window at ten rows per call. (From: Method note.)
- Documents were counted from the documents_held field on each row: the three higher education projects hold zero documents, and the full education window holds 81 documents across 13 projects. (From: Method note.)
Three campus names show up in a window of more than four thousand education projects
We counted every education project in the index announced in the last 660 days. The window holds more than four thousand projects. Three of them name a college or a university in the project name: a chiller plan for Queensborough Community College in New York, the Dental College of Georgia campus in Savannah, and a site development record on University Drive in Savannah that matches on a street name rather than an institution.
None of the three holds a document. Across the whole education window, thirteen projects hold documents of any kind. A rep watching this feed for campus work is watching a list of project names with no manuals behind them.
The records read like permits, and permits do not say what a building is
About five in six projects in the education window sit in New Jersey, and the names read the way permit records read: alteration, addition, remodel, and a town name. A permit for a college building and a permit for an elementary school can carry the same shape of name, and the record does not separate them.
That is why the name count above is a floor, not a census. Campus work is in this window. It is filed under building names, street addresses, and permit verbs, and a name search cannot pull it out.
The campus decisions sit in the documents, and the two exhibits here point in opposite directions
The document index is a different population from the project index, and it does hold campus jobs with named products. The two exhibits below came from that index.
At Texas Tech University, an amendment to the manual for a new cotton classing facility on the Lubbock campus names one product in Division 09, Finishes. The specified product reads "WOODWORKS Grille, Item # 7266DO as manufactured by Armstrong World Industries" (project "Construction of New Cotton Classing Facility on Texas Tech University Campus Lubbock, Texas", document "Amendment 4.pdf", pages 25 to 28).
The row is marked basis of design and proprietary, with two approved manufacturers on the section. That is a position: one named product, a short list, and a closed door behind it.
At Uintah Basin Technical College in Utah, a Division 23 section on a heat pump job carries Zurn on its approved list. The substitution language reads "Subject to compliance with requirements, provide products by one of the following:" (project "DFCM Construction - Prequalified Mechanical Stage II - Uintah Basin Technical College - Roosevelt Campus - Roosevelt Heat Pumps Phase 2 - Project #25093250", document "10978173", pages 29 to 288).
The approved manufacturer count on that row is 184. Zurn is named there the way everyone is named there, on a list long enough to decide nothing by itself.
A place on the list is not a position
Our judgment: campus specification work is won or lost on the role a brand holds in the section, not on whether its name appears. A basis of design line on a proprietary section, like the Texas Tech exhibit, is a decision already made.
A name on a 184 manufacturer list, like the Utah exhibit, is an invitation to compete at bid time against the whole market. Treating the two as the same kind of win is the error this count exposes.
The second error is trusting the announcement feed to show the work. In the current window the feed shows three campus names and no campus documents, while the document index holds campus sections that were decided on paper months or years earlier.
What to do with this
If you sell to colleges and universities, work the document index rather than the announcement feed, and grade every campus hit before you log it: read the role, read the openness mark, and read the approved manufacturer count on the row. A basis of design hit is pipeline. A list hit is a lead at best.
If you write campus specs, decide what the list is for. A list that runs past a hundred names hands the product choice to the low bidder.
If the intent is a real choice among equals, shorten the list. If the intent is one product, name it as basis of design and say so.
Method note
Population and queries, as of October 16, 2026. We ran search_projects with project_type "education", which returned 6,229 projects in total. We then ran the same query with announced_within_days set to 660, which returned 4,171 projects, and paged the full window at ten rows per call.
We classified higher education client side by matching project names against the terms universit, college, polytechnic, seminary, and institute of technology. Three projects matched: two in Georgia and one in New York. Their share of the window is 3 of 4,171, about seven in ten thousand.
Documents were counted from the documents_held field on each row: the three higher education projects hold zero documents, and the full education window holds 81 documents across 13 projects. The top states in the full window were New Jersey with 3,455 projects and Texas with 489.
The exhibits came from find_brand_positions, a separate document index and a different population from the project index above. Returned position totals on October 16, 2026 were 937 for Trane, 1,503 for Eaton, 1,995 for Armstrong, and 1,076 for Zurn. We paged the returns at five rows per call and filtered them client side to the same name terms.
The first 500 returned positions for Trane and for Eaton held no match. The Armstrong exhibit sits at offset 345 of its returns and was announced on January 17, 2020, outside the 660 day window.
The Zurn exhibit sits at offset 995 of its returns and was announced on September 10, 2026. We quote the specified_product and substitution_language fields exactly as returned, with the project name, document_title, and page span exactly as returned.
Limits, stated plainly. A name based classifier misses campus projects whose records carry only a building name, an abbreviation, or a street address, and one of our three matches is a street name, so the true count of named institutions in the window is two to three, not a firm three.
Project names in the window are permit style records, which is the main reason the campus count reads so low. The project index and the document index are different populations with different coverage, and neither is a census of campus construction.