All posts
Ray Iyer
Ray Iyer
Co-founder & CEO, Anglera

HVAC/R has a product-data problem — and 2026 is when it starts costing deals

HVAC/R catalogs are drowning in A2L SKUs and thin feeds while buyers shift to AI search. Here's what's breaking, what it costs, and how to fix it.

HVAC/R has a product-data problem — and 2026 is when it starts costing deals

The HVAC/R channel just absorbed the biggest catalog event in two decades, the A2L refrigerant transition, at the exact moment buyers stopped clicking through websites and started asking AI tools to shortlist vendors for them. Most distributor and manufacturer product data was not built for either shift. Here's what's actually broken, what it's costing, and why 2026 is the year it starts showing up in lost deals instead of just messy spreadsheets.

What's broken: feeds built for ERPs, not readers

Walk into most HVAC/R product data today and you'll find the same pattern repeating across manufacturers and distributors: catalogs that were built to move a SKU through an ERP system, not to answer a contractor's question. Attributes live in inconsistent units, refrigerant charge shows up as free text in a description field, and compatibility data between an air handler and a condenser lives in a PDF spec sheet nobody re-keyed into the PIM.

That's not an isolated complaint. Distributor-data vendors serving the space describe teams that "spend hours managing spreadsheets, updating product attributes, and reconciling manufacturer updates" because "product data from manufacturers rarely arrives in consistent formats" (Distributor Data Solutions). Multiply that across a catalog with tens of thousands of SKUs and the arithmetic stops working. Manual enrichment of a single SKU (finding the source spec, mapping it to your attribute schema, quality-checking it) commonly runs 30-45 minutes of skilled labor. Nobody has enough analysts to do that at HVAC/R scale, so the gaps just persist.

Then 2025 happened. On January 1, 2025, the EPA's technology transitions rule effectively ended new-equipment manufacturing with high-GWP refrigerants like R-410A, pushing the industry to A2L refrigerants such as R-32 and R-454B. Industry coverage of the shift doesn't mince words: "the shift from R-410A to A2L is the largest catalog event the HVAC channel has seen in two decades" (Flxpoint). Every SKU touched by the transition now needs six new regulated attributes attached correctly: refrigerant type, GWP, manufacture/import date, AHRI certification number, hazmat classification, and Section 608 buyer-certification status. Get one wrong and a contractor orders an incompatible cylinder, or a shipment gets flagged at hazmat. That guide is blunt about the failure mode: manual spreadsheet tracking "fails beyond approximately 50 SKUs," and unversioned records turn an ordinary EPA rule update into a forensic cleanup project.

What it costs: thin PDPs, wrong-warehouse returns, invisible SKUs

Incomplete product data doesn't fail loudly. It fails as small, compounding leaks:

Failure modeWhat it looks likeDownstream cost
Missing/inconsistent specsRefrigerant type or GWP buried in a PDF, not the PDPBuyer can't confirm compatibility, abandons cart or calls support
Free-text attributes"3/4 in NPT" vs "0.75in npt" vs "3/4-inch" across feedsSearch and filters return incomplete results, SKU looks out of stock when it isn't
Stale AHRI/cert dataCertification number not updated post-transitionOrder gets cancelled or flagged at hazmat/customs
No structured compatibility dataAir handler-condenser matching lives in a spec PDF onlyWrong-unit orders, restocking fees, install callbacks

Two forces are turning these leaks into lost revenue instead of background noise.

First, the channel itself is bifurcating. The median HVAC/R distributor in HARDI's Annual Benchmarking Survey generates only about 4 percent of sales online, while contractors already source roughly 36 percent of their purchases digitally, and that gap has been widening for several years (RS Web Solutions summary of HARDI/LBMX e-commerce report). The distributors closing that gap aren't doing it with more headcount, they're doing it with data infrastructure. Watsco's centralized product information system, cited by Digital Commerce 360, covers roughly 900,000 SKUs with up to 40 attributes each, and its digital channel now runs as high as 80 percent of sales in some regions (Digital Commerce 360). That's what a catalog looks like when it's actually maintained at scale rather than patched SKU by SKU.

Second, the buyer researching that catalog is changing. Gartner's most recent sales survey found 67 percent of B2B buyers now prefer a rep-free purchasing experience, up from 61 percent the year before, with buyers increasingly using AI tools during the research phase of a purchase (Gartner). Layer HVAC/R's own workforce shift on top: a large share of the technician and counter-buyer base is aging out, and the replacements coming in are digital-native by default. They're not calling a distributor rep to ask what refrigerant a unit takes. They're searching, and increasingly, prompting.

The AI-search problem is a data-structure problem

Ask an answer engine which condenser units under 5 ton are compatible with an R-32 air handler and shipping same week and watch what happens. The model isn't crawling your site live, it's reasoning over whatever structured, citable content it can parse in a few hundred milliseconds. A raw feed description like:

"24000 BTU condenser unit R-32 SEER2 up to 16 various voltage options see spec sheet"

gives it almost nothing to cite. An enriched attribute table gives it everything:

AttributeValue
Capacity24,000 BTU / 2 ton
RefrigerantR-32
SEER2 rating16.0
Voltage208/230V, single phase
AHRI certified reference number2025xxxxxx
Compatible air handlersModel AH-24, AH-24V

One of these gets cited in an answer. The other gets skipped for the competitor's SKU that parses cleanly, even if the underlying unit is identical.

Why this is a mechanism problem, not a marketing problem

None of this requires new supplier relationships or a multi-year systems overhaul. The values needed to fix most of the table above (refrigerant type, AHRI numbers, GWP, capacity, compatible models) already exist in supplier spec sheets, EPA documentation, and AHRI's own certified-product directory. The gap is that nobody has enrichment capacity to extract it, score it for completeness, and keep it current as the transition rules keep shifting through 2026.

That's the layer Anglera sits in. It plugs into whatever a distributor or manufacturer is already running, Akeneo, Salsify, inriver, Stibo, Syndigo, Pimcore, Informatica, or a flat file if there's no PIM at all, and continuously scores, gap-fills, and enriches product data by extracting values from the actual source documents rather than guessing at them. It doesn't replace the PIM or touch the CRM. It does the enrichment work that's currently eating 30-45 minutes per SKU manually, at the volume the A2L transition and the AI-search shift both demand, so the catalog stays complete for the buyer who's still calling and readable for the one who isn't.

Ray Iyer

About the author

Ray IyerCo-founder & CEO, Anglera

Ray is the co-founder and CEO of Anglera, building the product-data infrastructure for agentic commerce — turning messy catalogs into structured, AI-readable data that buyers and answer engines can find. Previously product at Uber; Stanford CS.

See it on your own SKUs.

A 30-minute walkthrough on your categories and your supplier data.

Book a demo