A sourcing list can look useful while leaving a recruiter with the hardest work. The title looks right, but the person has moved. The company name points to a parent instead of a subsidiary. An email exists, but another recruiter already owns the conversation. Adding more enrichment to that list does not resolve those decisions.
Clay can support candidate research through company search, people search and enrichment. The practical question is where those tools fit between a hiring brief and an approved ATS record. This guide recommends a workflow that keeps evidence visible and stops uncertain rows before they become messages. It does not assume that finding a profile proves job fit or gives permission to contact someone.
Start with a hiring brief that a search can use
Write down the work the person will do, the relevant experience, the locations you can hire in and the exclusions that matter. Separate required evidence from useful context. A title can be a search clue without being a hiring requirement. Ask the hiring manager to explain acceptable adjacent backgrounds before building the first table.
Choose whether the search starts from companies or people. If the brief identifies a specific set of employers, Clay recommends Find People at These Companies. If there is no company list, its standalone people search creates a new table from criteria such as job title. Keep the reason for choosing the starting pool beside the search configuration.
Build a small search and inspect the misses
Clay recommends title keyword lists rather than relying only on broad function and seniority dropdowns. Its title match modes include Is similar to, Contains and Is exactly. Similar matching can find title variations, but it can also return irrelevant profiles. Use excluded title keywords for obvious mismatches, then inspect actual results before expanding the search.
For company searches, Clay documents AND logic across different filter types and OR logic within a filter. That distinction matters when a brief combines industry, size and other company attributes. If you need either of two different combinations, use separate searches. Do not assume that adding another filter broadens the pool.
Keep a short rejection note for false positives. Was the company wrong, the title too broad, or the experience missing? Change the responsible filter rather than tightening everything at once. This gives the recruiter a search they can explain and revise when the hiring manager changes the brief.
Use the right company identifier
Clay recommends company LinkedIn URLs rather than domains where possible for people searches against a company list. A domain lookup can resolve to the wrong entity, especially around subsidiaries or renamed businesses. Check the intended company before using it as evidence about a person's employment.
Keep the original company input, the resolved company and the candidate profile identifier in separate columns. This is a recommended design choice, not a claim that every ATS uses the same fields. It makes a mapping error visible before it spreads through the rest of the workflow.
| Stage | Keep in the review row | Stop when |
|---|---|---|
| Search | Brief version and profile identifier | The result is outside the agreed pool |
| Employment check | Expected company and current experience evidence | The company conflicts or evidence is missing |
| Enrichment | Returned value, provider and check time | Identity or contact quality is uncertain |
| ATS handoff | Existing record, owner and next action | Ownership or contact preferences conflict |
Check current employment before using it
Clay describes people search as snapshot data. It can lag behind profile changes. Its documentation recommends running Enrich Person and checking current experience against the expected company. That is a useful check for company-based sourcing, but it should not turn missing evidence into a confirmed match.
Use explicit review states such as confirmed, conflicting and unknown. A recruiter can inspect conflicting rows or keep them for a different search. Unknown should remain unknown. Do not let an AI step fill the gap with a confident sentence about a candidate's current responsibilities.
Enrich only the fields needed for the next decision
Choose the missing field before choosing a provider chain. You may need enough background to review fit before you need any contact details. Keeping those steps separate reduces unnecessary collection and makes it easier to see which lookup actually helped the recruiter.
Clay describes waterfall enrichment as a sequential chain that stops at the first accepted result or when the chain is exhausted. Set an acceptance condition for the field and test it against your own target pool. Do not assume a provider result is correct because it arrived first. Preserve the provider and check time beside the accepted value.
Reconcile the list with your ATS
Before creating records, compare candidate identifiers against the ATS and the team's suppression list. Clay supports excluding existing people through sources such as a table, URL list or CSV. Use that capability as an early filter, then recheck against the current ATS state at handoff. A search exclusion is only as fresh as its input.
Define how to handle an existing candidate, an active applicant and a person assigned to another recruiter. Keep contact preferences separate from search relevance. A strong match should not overwrite a previous request not to be contacted. If the ATS write is uncertain, stop and reconcile it before retrying a create action.
Review the handoff before adding message generation
- Confirm that the row matches the agreed hiring brief.
- Check employment evidence and unresolved identity questions.
- Locate the existing ATS record and its owner if one exists.
- Review contact preferences and recent outreach.
- Approve the proposed record update and next action.
Only add message drafting once that queue works. Let the draft use verified role evidence, not invented personal details. Keep the recruiter responsible for the claim made in the message. Test the workflow with intentional duplicates and incomplete records so the stop conditions are exercised before a larger run.
Mistakes to avoid
- Enriching the whole search before reviewing relevance.
- Treating a title match as proof of the required experience.
- Confusing snapshot search results with confirmed current employment.
- Dropping source identifiers when moving rows between tools.
- Creating a new ATS record every time a workflow is retried.
- Sending outreach because an email field is no longer empty.
How Remqa builds recruiting research workflows
Remqa builds AI agents for recruiting teams around the systems and review steps the team uses. A useful starting scope is one role, one research queue and one approved ATS handoff. Bring a sample list and the decisions recruiters make on each row. We can map the checks, exceptions and ownership before adding more automation. For the surrounding integrations, see our guide to connecting a recruiting tool stack.