How does AI account research work in SalesOne?
Events first, then every rule, then a reason for every decision.
01Signals
Research starts from what changed.
S1 agents search for the events your profile names, such as a new executive, a new plant or a job post naming a project. Then they work out what the change means for the company: the problem it creates and why it matters now.
02Check
Every rule checked, every signal sourced.
Industry, headcount, revenue, ownership and exclusions are checked against named sources such as filings, the company website and job boards. Each signal is dated and counts only inside its window.
03Reject reasons
Every reject keeps its reason.
A rejected company keeps the rule it failed and the evidence, so you can always see why it is not on your list. Your team’s fit ratings, with their reasons, go into the next research brief, and when reject reasons show a pattern, SalesOne suggests a change to the profile for a person to confirm.
Why does research start from what changed?
A company’s need rarely appears out of nowhere. A new executive brings a new mandate, a new plant brings new costs, a job post names a project that has to land. Each change creates a problem that someone now owns.
So S1 agents date every change, count it only inside its window and write down the problem it points to, with the source beside it. Your team walks in with a reason to talk that is still true this month, and can show where it came from.
Research99%of B2B purchases are driven by organizational changes.B2B buying journey research, Gartner page as archived on September 25, 2026. Sample not published.Source (opens in a new tab)How is this different from a database or a quick AI lookup?
A contact database is the fastest, cheapest way to a phone number, but rarely shows where a field came from or when it was checked. A quick AI lookup answers one question per row. S1 deep research agents qualify every account against your profile, with dated sources, a reason for every reject and the people who own the problem.
See the full comparisonZoomInfo alternatives: cheaper data or deeper research
What makes an account qualified?
A dated signal inside its window
By default, an account qualifies only with a buying signal inside its window, such as 60 days for hiring in a function. A hiring signal from 213 days ago does not count. Exceptions are flagged for review.
Every rule passed
Industry, employees, revenue, ownership and location match your profile, and no exclusion applies, such as existing customers or competitors.
A source for every signal
Every buying signal links its source (98% dated to the day), and 95% of people carry a dated source, so anyone on your team can check a finding before using it.
A reason for every reject
The rule a company failed and the evidence, such as no buying signal in its window, revenue or employees outside your range, a subsidiary or pending sale, or an exclusion such as existing customers or competitors. On the latest runs each reason also carries a rule code, so a pattern shows up as a count.
A score out of 100
Accepted accounts are scored by default on fit up to 45, timing up to 40 and reach up to 15, then passed to the buying committee step. A profile can set its own weights.
United States only
SalesOne researches and qualifies US businesses. It does not offer international or GDPR coverage today.
SalesOne product specification, October 2026.
Who does what?
S1 agents do the homework. Your team builds the relationship.
S1 agents
Search sources, date signals, check every rule, record a reason for each reject, then re-check every signal in a second pass.
A person on your team
Sets the rules, reviews the accepted accounts and rates them, then chooses which to work.
Never
Drops a company without saying why, or presents a signal without its source and date.
What has the research shown so far?
31
searches and about 7 pages read, on average, for each account kept, with about 4 companies decided for each one kept.1
802
rejected companies, every one with its reason; 88% cite dated evidence.1
98%
of buying signals dated to the day; all 375 link their source.1
58 of 61
accounts usable in a blind test on one client’s profile, against 27 of 39 from a leading AI research API.2
- 1. SalesOne production research runs, October 2026: four runs, 1,051 companies decided, 249 kept; review-pass figures come from the three runs that had one (147 accounts). Client names withheld; method on the S1 agents method page.
- 2. Blind test on one client’s profile, October 2026: 61 accounts from S1 agents against 39 from a leading AI research API, checked against their sources by AI reviewers who did not know which side produced each. The API’s buying signals were more often valid (96% against 75%), and a second test on our own profile tied on usability. The AI checkers came from the same model family as S1’s agents. S1 had about an hour and 15 agents; the API ran at its default effort, about $1 a run.
Where this fits in the loop
01
ProfileWho you are, what you sell, your proof
02
TargetYour ideal customer, written as rules
03
Research
Every account, from dated sources
04
ScoreFit, timing, reach and who decides
05
SequenceA plan and a week of steps per account
06
EngageOutreach your team approves
07
CloseMeetings booked, the brief attached
08
RefineEach run builds on the last
Questions buyers ask
What is AI account research?
AI account research is using agents to gather and check what a company is doing before anyone contacts it. In SalesOne, S1 deep research agents find what changed at the company and the problem it creates, check the company against your ideal customer profile and record a source and date for every buying signal. For each account they keep, they run about 31 searches and read about 7 pages.
Is a general AI assistant good enough for account research?
A general AI assistant can describe a company when you ask, and a deep research tool can write a long report on one question. Neither holds your profile, checks hundreds of accounts against the same rules, or keeps a dated source and a reason for every decision. That consistency, across every account in a market, is the work SalesOne does.
Does SalesOne include contact data?
Yes, for the people your team picks. SalesOne is not a contact database: S1 agents research the account and find who owns the problem, and verified contact details are looked up from 20+ data providers only for the people your team chooses.
How long does a research run take?
A 50-account list typically takes about an hour of agent time, with research agents working in parallel. In four production runs in October 2026, a run took 44 to 82 minutes for 27 to 102 accounts kept.
Can I see why an account was rejected?
Yes. Every company turned down keeps the rule it failed and the evidence, such as no buying signal in its window, revenue outside your range or a pending sale: 802 of 802 in our October 2026 production runs, 88% citing dated evidence. When reasons show a pattern, SalesOne suggests a change to your profile for a person to confirm.
How accurate is SalesOne's research?
In a blind test on one client’s profile in October 2026, 58 of 61 accounts from S1 agents were usable, against 27 of 39 from a leading AI research API. The checkers were AI agents, and the API’s buying signals were more often valid (96% against 75%); a second test on our own profile tied on usability. A second S1 pass now re-checks every buying signal before a list is delivered.
Related reading
- Best AI tools to research accounts before outreach
Comparison
- ZoomInfo alternatives
Comparison
- What is account qualification?
Guide
- What is signal-based selling?
Guide
- Hiring buying signals
Signal library
- Technology buying signals
Signal library
- Account scoring matrix
Template
See which companies have a problem you solve.
Bring your offer and one customer you would like more of. We will show you the accounts that fit, what changed at each one, who owns the problem, and the point of view your team would walk in with.