Define an evidence standard for virtual-assistant lead enrichment

Lead enrichment often starts with a short account list and ends with dozens of fields whose origins nobody remembers. A virtual assistant can research companies and maintain source links. The assistant should not invent buying intent, infer sensitive personal traits, or present an uncertain match as a verified prospect. An evidence standard tells salespeople which fields help a real decision, which sources are acceptable, how quickly facts go stale, and what to do when sources disagree. It values an honest unknown over a polished but unsupported profile.

Choose fields that affect a sales decision

Begin with the decision the seller needs to make. If the question is whether an account fits a service territory and industry, the record may need legal or trading name, official domain, operating location, industry description, and the source date. It may not need a guessed employee's personal phone number. For each field, name its purpose, allowed source types, freshness window, and confidence states. Remove fields that nobody uses. Collecting extra data increases review work and privacy exposure without improving the sales conversation. Separate account facts from contact facts. A company may fit the target profile while the named person has left. Give both records their own identifiers and timestamps so an outdated contact does not invalidate the entire account.

Rank sources by the claim they support

An official company page can support its current product description but may use broad marketing language. A government registry may support a legal name or filing status but not current headcount. A job posting can show that a role was advertised, not that the company hired someone or approved a budget. Write source rules at the field level. Preserve the exact URL, publisher, page title, capture date, and relevant excerpt or observation. If access requires a login, record the approved system without copying restricted data into another tool. Third-party databases can be leads for research. They should not silently become the source for every field. When two reliable sources conflict, show both and route the choice instead of selecting the value that best fits the campaign.

Timestamp changeable facts

Roles, locations, products, and websites change. A field without a capture date looks current forever. Store when the source was checked and when the page itself says it was updated, if that information is available. These dates mean different things. Set review windows based on use. A campaign launching next week needs recent contact and account checks. A historical market analysis may preserve older states intentionally. The sales owner decides the acceptable age; the assistant applies it consistently. When a source disappears, do not keep the value marked verified. Preserve the prior citation, change the state to needs review, and search approved alternatives. A cached snippet alone may not contain enough context for a new claim.

Avoid inferred intent and sensitive traits

A visit to a public page, a new job posting, or a software mention does not prove that an account wants to buy. Record the observed event and let the seller decide whether it justifies outreach. Do not convert "posted an operations role" into "needs an overseas virtual assistant." Do not infer health, ethnicity, religion, political views, family status, or other sensitive information from names, photos, associations, or content. Exclude personal data that has no approved business purpose. Follow the company's privacy rules and the terms of every research source. The assistant should also avoid synthetic personalization. A sentence that pretends to understand a person's priorities can be worse than a plain introduction. Keep outreach claims tied to reviewed evidence.

Work a representative account example

Suppose a seller supplies "Northstar Labs" and asks for headquarters, industry, likely operations contact, and evidence of current growth. Search results show several companies with similar names. One official domain lists a Denver address and laboratory software. A registry result uses a longer legal name. A professional profile names an operations director but appears unchanged for two years. The assistant first resolves the account using domain, location, and legal-name evidence. The record links both authoritative sources and notes the trading-name relationship. The old professional profile is a contact candidate, not a verified current employee. The assistant marks the role unconfirmed and does not manufacture an email address. A careers page contains two open engineering jobs. The assistant records the postings and capture date. The record does not say the company is rapidly growing or has purchasing intent. The salesperson receives a usable account profile with one open contact question.

Sample records before import

Run a small batch that includes an exact company match, ambiguous name, subsidiary, stale contact, conflicting location, and account with little public information. A sales owner reviews every field and source before data enters the CRM. Track fields requested, fields supported, unresolved matches, stale sources, owner corrections, rejected personal data, and records withheld from import. Report the denominator. Do not reward completion rate in a way that encourages guesses. After import, retain the source and capture date beside the field where the system permits it. Define who reviews aging records and how corrections flow back to the research instructions. Expand volume only after the sample shows that uncertainty survives the handoff. Audit the imported sample again after a short interval. Ask a reviewer to open each cited page, confirm that the account still resolves to the intended company, and inspect whether the CRM preserved the source and uncertainty state. A broken link does not automatically make the old observation false, but it does end the claim that the field is currently verified. Record the new state and route any outreach already prepared from stale information. This second look tests maintenance, not just initial research speed. The FTC privacy and security guidance for businesses (https://www.ftc.gov/business-guidance/privacy-security) provides general background for responsible data handling. Apply relevant privacy law, source terms, contracts, and company policy with qualified advisers. To scope a research lane, review virtual assistant services (/services) or request a role plan (/contact). Bring a redacted target-account sample, the CRM field list, and the source policy.