A virtual assistant cannot read the unwritten part of your sales plan. If the instruction is “find decision makers at growing companies”, the researcher has to invent what growing means, choose the businesses and guess which role matters. A tidy spreadsheet can hide all three guesses.
The useful deliverable is a reviewable research file for one defined offer. This article provides an operational handoff you can adapt after choosing a segment. For the earlier decision about whether outside research is worth paying for, use the outsourced prospect research guide.
Start with the decision
Write down what you want to decide after reviewing the work. “Approve a small set of companies for manual account research” is an actionable objective. “Generate new clients” is too far downstream to serve as an acceptance criterion for this task.
Fictional worked example: a consultant called Northbank Operations offers a fixed-scope review of how small agencies hand projects from sales to delivery. Northbank wants to investigate agencies that publicly describe both a sales function and a delivery team. The invented company, offer and batch sizes below illustrate a process; they are not a customer case study or a recommended market benchmark.
The researcher is not being asked to establish that any agency has a broken handoff. A public service page might support an agency’s type of work. A team page might show an operations role. Neither proves a current problem, budget or willingness to buy. Make that boundary part of the brief before the first row is collected.
Write a brief someone can use
Give the assistant a short document that remains beside the spreadsheet. Include one example of a relevant company and one example you would exclude, with the reason for each. If you cannot explain the difference, resolve the segment first with the Service-to-Segment Decision Sheet.
For Northbank, the working brief might look like this:
| Brief item | Fictional instruction |
|---|---|
| Decision | Prepare candidates for the consultant’s manual review, before any contact. |
| Company scope | Agencies whose own website describes project delivery as well as client acquisition work. |
| Geography | English-language businesses in the market the consultant can actually serve. |
| Exclusions | Recruitment-only firms, software vendors and businesses already in the consultant’s contact ledger. |
| Role hypothesis | Someone responsible for operations or delivery; record the actual role rather than forcing a title match. |
| Evidence | A company URL and a short quotation or faithful note showing the stated inclusion condition. |
| Unknowns | Team ownership, current process, budget and buying authority stay unknown unless supported. |
| Pilot | Return five candidate companies before continuing. Five is this example’s working choice, not a quality standard. |
| Boundary | Research only. No invitations, messages, account creation or purchases are included. |
Also agree the delivery format, review time and what happens when a source cannot be opened. An assistant should be able to mark “not established” without feeling obliged to fill a cell. A believable unknown is more useful than an invented answer.
Keep access arrangements explicit. Use the sources and permissions agreed for the work. Do not put account passwords inside the brief or a shared CSV, and do not assume that a research assignment also authorises outreach.
Separate export fields from research
There are two different records in this handoff: a collected profile record and your interpretation of that record. Keep them distinguishable so another person can check the evidence.
The published Leadsforlinked header-only CSV contains these actual profile-export columns:
full_name, first_name, last_name, linkedin_url, headline, company, position, location, avatar_url, created_at, civility.
That download contains no personal rows. It shows a file structure, not complete records or a promise that every field will be populated. It also does not include a verified-email column.
For Northbank’s research, the assistant could maintain a separate review sheet with company_source, observed_condition, source_checked_on, role_hypothesis, review_status and review_note. Those are suggested columns you add yourself; they are not Leadsforlinked export fields or an automatic enrichment feature. Use the profile URL as a reference when reconciling the two files, and review duplicates before combining anything.
This separation prevents a subtle error: turning “operations” in a headline into a statement that the person owns this particular purchase. Record the observed text in one place and the proposed relevance in another. The consultant can then approve or reject the interpretation without losing the original observation.
Review the first small batch
Ask for the pilot before the assistant repeats the method. Review every candidate in that first handoff against the brief. Start with company fit, then the source supporting inclusion, then the role hypothesis, and finally the file’s usability.
Use clear dispositions rather than a vague quality score:
| Disposition | What the reviewer does next |
|---|---|
| Keep for investigation | The stated criteria are supported; buying need still needs discovery. |
| Return for clarification | A source or observation is missing and can reasonably be checked. |
| Exclude | The business conflicts with a stated exclusion; explain which one. |
| Leave unresolved | Public evidence does not settle the point; do not quietly count it as a match. |
If three of the fictional five companies are ambiguous, do not simply ask for more rows. Compare the ambiguous examples with the brief. Perhaps “agency” is too broad, or the website does not distinguish delivery from account management. Change the instruction once, note the version and have the assistant recheck the pilot before expanding.
Hand back a decision, not just a file
Close the review with a short note: which candidates remain, what changed in the brief and which questions require a conversation rather than more research. This makes the next batch repeatable and prevents rejected companies from returning under slightly different names.
If your next task is to collect a few professional profiles for this review, Leadsforlinked’s workflow is LinkedIn People search, the Chrome extension, an audience in the app and a CSV export. Inspect the actual records against your brief before expanding. Existing contacts may already be sufficient; importing another file is not a substitute for a clear research decision.
The next useful question is then specific: “Can this company explain who owns the sales-to-delivery handoff?” That is a discovery question you can ask appropriately. It is not a fact the assistant should manufacture in the spreadsheet.