FIELD GUIDE · RESEARCH & INTELLIGENCE
AI Lists vs. Verified Research
An AI can generate four hundred names tonight. It cannot tell you who is movable, what they cost, or whether the market you’re searching even exists. Scraping produces names. Research produces evidence - verified, sourced, current. The difference shows up at the intake, when a list either becomes a strategy or a do-over.
01 - WHAT AI LISTS GET RIGHT
Volume and speed. Conceded honestly.
AI-generated name lists are fast. For a role with a clear title pattern and a wide market, an AI tool can return hundreds of names in minutes - names that would take a human researcher days to compile by hand. That is a real advantage, and pretending otherwise wastes your time.
They are also reasonably good at identifying who holds a given title at a given company, surfacing LinkedIn profiles with matching keywords, and building a raw universe to start from. For industries with consistent titling and large candidate pools, they narrow the starting point quickly.
The question is not whether AI lists have value. The question is whether a list is what you actually need - or whether you need a strategy.
02 - WHAT THEY CANNOT VERIFY
The four things a list can never tell you.
Availability
Whether any of these people are movable - actively looking, passively open, or locked in by equity, a new role, or a non-compete - is not in a profile. It requires a conversation. AI cannot have that conversation.
Compensation reality
Salary surveys are averages. What a specific candidate in a specific market in a specific company expects to make is a negotiated data point. AI tools do not have access to that number. You will discover it when you make the offer - or you will research it before you open the search.
Org context
Whether the candidate runs a function independently or shares it, whether they have a team or inherited a vacancy, whether their role is expanding or being restructured - none of that appears in a title. Context determines whether someone is a realistic target or a name on a spreadsheet.
Interest
The gap between 'could theoretically be approached' and 'is actually interested in this opportunity' is everything. A list tells you who exists. Research tells you who is worth approaching - and why.
03 - WHAT VERIFIED RESEARCH ADDS AT EACH STEP
The difference at each step of the engine.
Step 1 - Winning
Verified market research confirms whether the candidate standard your intake defines actually exists. If the market for your definition of winning is three people in the country, you learn that before spending a retainer - not after a search fails.
Step 3 - Behavior
Research surfaces org context - what the candidate has actually run, how independently, with what resources. That context is what allows evidence-based evaluation prompts to be calibrated correctly. Without it, you are asking behavioral questions into a vacuum.
Step 6 - Performance
Compensation reality gathered before the offer is made protects the close. A hire who accepts an offer misaligned with their market rate is a retention risk from day one. Verified data prevents that.
04 - WHEN TO USE WHICH
Not a competition. A sequencing question.
AI-generated lists serve one purpose well: building a raw universe to start from. For a mature market with consistent titling - VP of Finance at mid-market manufacturing companies in the Midwest, for example - a list gives researchers a starting point, not a strategy.
Verified research takes over where the list ends: confirming availability, gathering compensation context, mapping org structure, identifying who is actually a realistic target versus who appears on every list and has been approached by every firm this quarter.
The mistake is treating the list as the output. Names in days is not the same as a verified, targeted candidate pool. One of them survives an intake. The other produces a do-over.
“Names in days. Verified availability in weeks. Only one of them survives an intake.”
Need names fast? Need verified intelligence? We run both - starting with which one the search actually requires.
