Company search finds businesses that fit an audience. It combines company filters, ranking by relevance, and research questions. It can support account selection, partnerships, market research, or the company-first stage of an outreach campaign.
The output is a set of companies. People search can then find relevant contacts at those companies. Targeting groups organize the connected searches and lists.
Where a search looks#
The Starting Set defines the pool being searched. Company search can use:
| Starting point | What it covers |
|---|---|
| All of Distill | Companies available in Distill's searchable data, which covers part of the wider market. |
| A company list | Companies already supplied, imported, or saved in an accessible list. |
| Available network | Companies associated with contacts in the selected, accessible network. |
| Additional discovery sources | Supported web or LinkedIn discovery on search paths that use those sources. It is not included in every search automatically. |
A company criterion can also find the current employers of people in a selected list. This differs from starting with a list of companies.
Natural-language search turns an audience description into visible criteria and a starting pool. Required filters, ranking preferences, and research questions have different effects; a detail in the description is not necessarily saved as a requirement. Networks explains network coverage and access.
Company filters#
The company search editor includes:
- Identity and classification: company name, company type, industry, startup sector, tags, and list membership.
- Headquarters: country.
- Size: employees and known employees by function.
- Stage and funding: revenue bands, total capital raised, last funding amount, round type and year, investors, and founding year.
- Other characteristics: customer segments, Software as a Service (SaaS) status, and domain rank.
- Existing audience relationships: employers of people in a selected list.
- Text and relevance: phrases in profile text and description similarity.
Data coverage varies by company. Revenue categories and employee counts describe different things. Funding amounts are not revenue, and a company without a known funding round is not necessarily unfunded.
Known employees by function counts people in the searchable data rather than every employee. Multiple selected functions can match separately; their counts are not added into one combined department total. SaaS and customer-segment classifications may be stored AI assessments.
What the filters mean#
| Control | Effect |
|---|---|
| Include values | Usually alternatives within that field, such as either of two selected industries. |
| Separate required filters | Combine to narrow the audience, such as an industry requirement and a headcount range. |
| Exclude values | Remove matching records from that search. Exclusions still apply when the positive part of the criterion is optional. |
| Minimum and maximum | Define numeric ranges, such as 50 to 500 employees. |
| Optional criteria | Add positive ranking preferences rather than requiring a match. |
| Similarity | Ranks relevance to the description or role. It does not require an exact match or determine eligibility on its own. |
Available controls vary by field. SaaS uses a yes/no choice; customer segments use included categories. The editor does not support every combination of AND and OR conditions. “At least 150 engineers OR at least 500 employees” is not the same as requiring both numbers. Different audience groups can express different combinations.
Saved results can also have an unknown filter assessment when the relevant information has not been established. An unchecked or unknown field does not confirm a match.
AI traits and web research#
Research traits help assess characteristics that ordinary company fields do not fully describe. For example: “Does this company publicly describe an integration partner program?”
- AI scoring assesses available company context against the trait.
- Web research investigates public sources and can return supporting evidence and citations.
AI-scored and web-research traits distinguish a supported match, unclear evidence, and a supported non-match. A required trait can reject a result with a negative assessment. Unclear or missing evidence is not the same as a negative assessment, including when the trait is required. A result can remain in the search without every research question being answered with a confirmed yes.
Optional traits affect ranking. Pending or failed research is different from evidence that the company or person does not fit. Existing searches may also contain older LinkedIn-insights traits with their own numeric thresholds.
Public research can establish only what its sources support. An article about a technology is not proof that the company uses it internally. Private budgets and buying plans may be unknown. Research explains sources, research methods, and freshness.
Results, counts, and saved records#
Results can include company details, filter assessments, research evidence, and relevance feedback. Saved, hidden, and rejected results represent different decisions. A quality assessment applies to the search run it evaluated, not automatically to later edits.
The matching pool counts companies matching the source and search conditions before later research and decisions about saving them. A lower-bound indicator means there may be more matches. Saved records count entries currently held in the connected company list.
Results arrive in batches, and a preview is a sample. An unfinished search or incomplete source can explain missing results. The company count is also separate from the number of people later found at those companies.
How search changes connect to outreach#
A saved search definition, a completed search run, and a list are separate. Applying new criteria can start work that changes the connected list as results are reassessed. Entries saved by that search can be removed if they no longer fit. Independently imported or manually added entries are not automatically reassessed against the search in the same way.
When a saved search is created within a project, the app adds that project's Do Not Contact exclusion. Search-specific exclusions still have a different scope from project-wide outreach rules. CRM screening checks records from a connected customer relationship management system. Those checks, campaign settings, the current audience, and other outreach checks continue to matter after a search result appears.
Searching or saving a record does not itself start a conversation. Campaign behavior belongs in Campaigns and personalization.
Example: companies with a partner program#
A fictional audience could use:
- Company headquarters: United States.
- Employees: 50 to 500.
- Appropriate software-industry selections.
- Optional web-research trait: evidence of a public integration partner program.
The company criteria identify the organizations. A linked people search can then look for current partnerships leaders at them. The company headquarters condition does not require those employees to live in the United States.
Working through a connected assistant#
The connected assistant supports natural-language company search and explicit-criteria previews. Natural-language search can create a saved search and return a page of results. A preview returns a count and sample without creating that saved search. More results can be returned through the supported pagination.
The natural-language search supports relevance ranking and a stricter filter mode. Additional web or LinkedIn discovery is a separate option, rather than the default for every search.
These search capabilities are different from the current create-targeting-group action, which creates a People search. They do not imply that the assistant can reproduce every company-first group setup or network selector available in the app. Using an assistant explains access and changing tool availability.
Common questions#
Can I start with a list of companies we already know?#
Yes. A company list can supply the starting businesses. A separate people search can find relevant roles within those companies, or the list can remain a research audience without outreach.
Can I find companies based on more than industry and size?#
Yes. Company criteria also cover funding, location, classifications, text, and other supported fields. AI traits can assess a more specific question, such as whether a public partner program exists, using available evidence.
Can I find companies that employ a certain number of engineers?#
Known employees by function can help. It counts people in Distill’s searchable data, not a complete workforce census. Selecting several functions checks them separately rather than adding their counts together.
Does a required research question mean every company has a confirmed yes?#
No. A supported negative assessment can reject a company, while unclear evidence can remain unresolved. The research result and its supporting evidence explain what is known.
Why is the company list smaller than the search count?#
The matching pool, assessed results, and saved company list measure different stages. Exclusions, unfinished research, duplicates, source coverage, and decisions about saving records can change the final list.
If I find the right companies, will Distill contact someone at each one?#
Company search identifies organizations. People search identifies contacts, and a campaign determines who can receive outreach. Saving a company list alone does not send messages.