Before you begin
- Write an account hypothesis that explains why a company characteristic or event is relevant to your offer.
- Separate required conditions from preferences and exclusions.
- Choose a region, company range, customer type, and business context in which the characteristic matters.
- Define the person or persona you would look for at a matching company.
- Decide what happens after a result appears: monitor, review, draft, add to a tracker, or hand off.
Choose between company search and monitoring
| Use case | Use it for | Do not use it as |
|---|
| AI Company Search | Finding companies that match characteristics that are hard to express with static filters alone. | A guarantee that every result is a qualified account. |
| News or signal monitoring | Watching defined public events such as expansion, partnerships, hiring, go-to-market changes, or an expressed operational need. | Proof of buying intent or a replacement for company and persona scoring. |
Write a verifiable company search prompt
Describe observable characteristics instead of a general ideal customer profile. A useful prompt states what the company offers, who it serves, where it operates, its approximate size if relevant, and what must be excluded.
- Offer or business model: what the company provides
- Customer type: who buys or uses the offer
- Region: headquarters or operating market, if relevant
- Size: an appropriate range for company size or maturity
- Evidence: a characteristic that can be checked using available data
- Exclusions: industries, business models, regions, or named patterns that do not fit
Example structure: “Find companies in [region] that offer [offer] for [customer type], show evidence of [verifiable characteristic], and exclude [clear exclusions].” Treat examples as guidance, not hidden requirements.
Name the event you want to detect and explain why it matters. Suitable event categories may include expansion, partnerships, relevant hiring, new go-to-market initiatives, changes related to funding, or a public statement about a problem your offer addresses, where such sources are available.
Set a sensible time window and separate evidence from interpretation. For example, a job posting may support the statement “The company is hiring for this role.” It does not automatically prove “The company has budget for our product.”
Steps
- Write the account hypothesis. Describe the company pattern or event and why it may make an account worth reviewing.
- Choose AI Company Search or monitoring. Use search for descriptive company discovery and monitoring for new events over time.
- Describe required characteristics. Add offer, customer type, region, size, evidence, and any other conditions that must be true.
- Add explicit exclusions. Name regions, industries, company types, or patterns that make a result unsuitable.
- Define event and time window for monitoring. Use concrete events and a period that fits your sales cycle. Avoid vague requests for “interesting news.”
- Choose the relevant persona. Specify which role should be reviewed if a company or signal matches.
- Set a small initial limit. Generate or monitor a sample that one person can review before expanding scope.
- Review and revise. Compare results with known positive and negative examples and change only one part of the prompt at a time.
- Assign the next action. Route each accepted result or priority to monitoring, manual review, a draft, a tracker, or an approved CRM handoff.
Review results before using them
- The result matches the required company characteristics, not just a general keyword.
- The available evidence supports the factual claim.
- The company and intended persona fit the target segments.
- Exclusions, duplicates, status in other lists, and blacklists have been checked.
- The next action matches the signal priority and does not overstate buying intent.
- The first sample is good enough before increasing the limit or monitoring scope.
A signal is evidence of an event, not proof of intent. Always validate fit, context, and source quality before using the result in outreach, scoring, or CRM notes.
Troubleshooting
| Problem | Solution |
|---|
| Results are too broad | Add region, customer type, company size, business model, explicit exclusions, and two representative positive examples. Remove vague terms like “innovative” unless you define how they can be verified. |
| Results are too narrow | Remove unnecessary conditions, separate must-haves from preferences, and test only one change at a time. |
| A monitored signal sounds relevant but has no clear action | Define the action before activation: monitor, manually review, prepare a draft, add to a tracker, or hand off to CRM depending on priority. |
| The prompt returns companies outside the target market | Explicitly name excluded regions, industries, customer types, and company patterns, then review a new sample. |
| A result contains a claim that cannot be verified | Open the available source context. Do not use the claim in outreach or scoring until a trustworthy source supports it. |
Frequently asked questions
What makes a good prompt for AI Company Search?
A good prompt describes observable company characteristics, a target context, explicit exclusions, and verifiable evidence. It avoids vague adjectives without a clear definition.
Should I include examples in the prompt?
Examples can clarify the intended pattern. Also describe the shared characteristics so the search does not depend on names alone. Use approved, non-sensitive examples.
Does a news signal prove buying intent?
No. A signal is evidence of an event, not of buying intent. Check company fit, persona fit, context, and the appropriate next action.
How often should I revise the prompt?
Review it after a small sample and whenever the result pattern changes. Change only one condition at a time so you can see what improves or reduces quality.
Should I use monitoring or a tracker?
Use the workflow shown in your workspace for the outcome you need. Monitoring detects defined events; a tracker may support ongoing account or profile workflows where available.
What should I do with an unverifiable claim?
Do not use it as fact in outreach, scoring, or CRM notes. Keep the result for manual review or remove it until a trustworthy source confirms it.
Related articles
Reference: https://digital-now.atlassian.net/wiki/spaces/SS/pages/248479766
Keep reading
Choose the right list source
Choose your list source by starting with what you already have and what you want the list to do next. If you begin with the source name alone, it is easy to pick something too broad, too narrow, or not available in your workspace.
Use LinkedIn list sources safely
LinkedIn activity can help you build timely lists from profile visits, followers, events, posts, engagement, or prepared LinkedIn URLs. Use the source that matches the specific signal, then qualify the people before you treat them as outreach candidates.
Upload people, companies, and job titles via XLSX
Use an XLSX upload when you already have a defined set of people, companies, or job titles. A clean file and a careful preview help Signalist match the right records and reduce avoidable duplicates, credit use, and manual correction.
