If you sell integration services, cobots, machine vision, or controls into plants, you already know the pitch isn't the hard part. Finding the plant with a line still run by hand, and enough headcount on it to make the payback obvious, is.
The short version
- Company-level tools can't show you which plants still run a process manually. Headcount and NAICS code describe a company, not a floor.
- Filtering on headcount alone doesn't fix this: most manufacturers are too small for a flat cutoff to mean anything, so you get everyone or no one.
- The real signal is headcount weighed against building square footage, paired with a search for the actual process (palletizing, deburring, hand assembly), in a territory you can work, with a contact at that specific plant.
- The IFR says the automatable work moved out of automotive into food, metal, and electronics. Most integrator pipelines haven't followed.
Automotive isn't where the growth is anymore
The IFR's World Robotics 2026 release found US industrial robot installations rose 11% year-on-year to 38,000 units in 2025. Automotive, the traditional backbone of integrator work, fell 1%, while food processing surged 30% and metal/machinery and electronics each grew alongside it. If your rep team still calls automotive tier-1 and tier-2 accounts by habit, you're not wrong that automotive buys. You're just missing the part of the market growing three times faster, and there's no field in ZoomInfo or Apollo that tells you where it is.
The bigger picture backs this up: Precedence Research puts the global industrial automation market at roughly $256 billion in 2025, growing to $613 billion by 2035. That growth isn't spread evenly across a NAICS code, it's showing up plant by plant, in specific facilities a company-level database has no way to flag.
How to actually build the list
- Weigh headcount against square footage. A high ratio of people to building size in a production facility is a reasonable proxy for a line still run by hand, not automated, worth confirming on a site visit rather than treating as gospel.
- Search by process, not industry code. A plain-English query like "precision aluminum machine shops with CNC mills" should match on what a facility actually does, ranked by meaning rather than keyword overlap, not just a NAICS code that happens to correlate.
- Filter to the industries actually growing (food, metal fabrication, electronics) instead of defaulting to the automotive accounts every other integrator already has.
- Reach the plant directly. A domain-matched contact gets you the operations engineer or plant manager who owns the decision locally, not a procurement contact at HQ who's never seen the line.
This only works if the underlying data is at the plant level, not the company level. Getting there means pulling from a lot more than one source: satellite imagery for building footprint, the facility's own website for what it actually makes or processes, EPA and state environmental filings, business registries and public records, and professional networks for who works where, all cross-checked and resolved onto one record. That's what it takes to get to 20+ structured fields per facility instead of the handful a standard B2B database offers, and to tie an employee to the specific plant they work at rather than just a company that might have thirty locations.
The scale matters too. There are more than 600,000 mapped plants, warehouses, and branches across the US, geocoded and linked to their parent company, with 25 million employees attached to the specific facility, not just the corporate record. For an integrator, the relevant slice of that universe breaks down roughly like this: 182,000 manufacturing facilities, 168,000 industrial services facilities, and 98,000 warehouse and distribution facilities. Filtering the type, square footage, headcount, and territory should all still work as stackable filters on top of a process search, not instead of it.
"I drive past facilities every day that could be customers, but I have no way to know they exist until I literally see them from the road," says Mike, an industrial sales rep. That's the plant-level gap in one sentence, and it's exactly what headcount-only filtering can't close.
If you're deciding whether you're selling to the integrator or the end-user plant in the first place, that split, and how each buyer's economics differ, is covered here. For why NAICS-code targeting misses plant-level detail generally, see Why "Manufacturing" as an Industry Filter Is Useless for Industrial Sellers.