A rep pulls up an account. Right industry, right size, right region. Everything on the company page checks out. He sends the first email anyway a little unsure, because somewhere in the back of his head he knows the company page is the least interesting thing about this account.
That uncertainty is the whole problem, and it shows up differently depending on where you sit.
At the rep level, it looks like manual detective work. One rep we talked to had already tried filtering his account list by something specific, which accounts run multiple shifts of a particular operational workflow, not just "companies of this size in this industry." A generic GTM intelligence tool couldn't get him there. His read on it was blunt: too broad, not specific enough to his industry to actually be useful. The data wasn't wrong, it just stopped at the company level when his job starts at the facility level.
He'd tried general-purpose AI tools too, asking them to sanity-check fit on accounts. They came back just as confident as the humans he trusted, and just as wrong. One flatly disqualified an account as not a fit around the same time he was on a discovery call with that exact company. Confidence isn't accuracy. At the facility level, that gap gets expensive fast, because the wrong confident answer costs a rep a week of pipeline he didn't need to lose.
At the manager level, the same gap shows up as a triage problem instead of a research problem. A manager we spoke with wasn't worried about her rep's work ethic, her rep is good. The issue was time: her rep spends too long just figuring out who to call. Not because names are hard to find, LinkedIn hands those out freely. The hard part is knowing whether that name sits at the corporate office, a regional office, or the facility itself. Her framing was exact: you need to find the people first, then figure out if they're corporate, if they're regional, are they over that facility.
That distinction isn't a nice-to-have. A corporate contact and a facility-level contact are having two entirely different conversations, even when the job titles look almost identical on paper. Her current stack, LinkedIn, Sales Navigator, a browser extension, gets her team names. It doesn't get them that layer. As she put it, it doesn't have all this, you're pulling up stuff my team would never see. She's not trying to get her reps calling everybody. She's trying to get them calling the right person. Most tooling is still built to solve the first job, not the second.
At the org level, the rep's uncertainty and the manager's triage problem compound into something a revenue leader feels every quarter without necessarily being able to name. When a team can only evaluate accounts at the company level, some of the pipeline logged as "qualified" is really a guess wearing a CRM stage. Nobody confirmed the specific facility actually runs the operation being sold into until three calls in, and by then it's already counted. That pipeline looks healthy on a dashboard and closes soft. Forecasts miss because a chunk of the deals inside them were never truly in-market, just close enough on paper to get logged.
It helps to think of the pipeline less like a headcount number and more like a portfolio. A portfolio full of positions nobody can actually verify isn't diversified. It's just unexamined. The fix at this level isn't more reps or more activity, it's resolving fit before an account ever gets touched, at the facility level rather than the company level. Fewer accounts moving through the pipeline, a higher hit rate on the ones that stay.
Build your pipeline with accounts that are a good fit your solution and show real signs of automation need
Three different jobs, three different symptoms, one underlying pattern: the company page tells you a business exists. It doesn't tell you what's actually happening inside a specific facility, and that's the only level at which fit is real. Everything above that level, industry, headcount, region, is a proxy. Sometimes a decent one. Never the thing itself.