How to Describe Your ICP for an AI Lead Finder: Examples and Best Practices

January 08, 2026

The prompt is everything. When using an AI lead finder, the quality of your results depends entirely on how well you describe what you’re looking for. Get it right, and you’ll receive a list of precisely targeted prospects in seconds. Get it wrong, and you’ll waste time sifting through irrelevant results or missing ideal customers entirely.

AI lead finders work fundamentally differently from traditional database filters. Instead of selecting from predetermined categories, you describe your ideal customer profile in natural language. The AI interprets your description and finds companies that match. This flexibility is powerful, but it requires a different skill set than checkbox-based filtering.

The anatomy of an effective ICP prompt

Effective prompts share common characteristics. They’re specific enough to exclude irrelevant matches but not so restrictive that they eliminate good prospects. They focus on observable characteristics that the AI can actually evaluate. They describe what you want, not what you don’t want.

A strong ICP prompt typically includes several elements:

Company type or business model. What kind of business are you targeting? SaaS companies, agencies, manufacturers, retailers? This foundational element shapes the entire search.

Specialization or focus area. Beyond general business type, what specific area should they work in? E-commerce, fintech, healthcare technology, B2B services? The more precise your specialization description, the better your results.

Size indicators. Employee count remains a useful proxy for company stage and capacity. Ranges work better than exact numbers: “10-50 employees” rather than “exactly 25 employees.”

Geographic focus. Where are these companies located or where do they operate? “US-based” is different from “serving the US market.”

Additional qualifiers. Any other characteristics that matter to your targeting. Recent funding, specific technologies, target customer types, or growth indicators.

15 prompt examples across industries

Learning by example accelerates prompt writing skill. Here are effective prompts for various targeting scenarios:

Technology and SaaS

  •  “B2B SaaS companies in the US building HR or recruiting software with 20-100 employees”
  •  “Developer tools startups in Europe that have raised Series A funding in the last 18 months”
  •  “Cloud infrastructure companies targeting enterprise customers with engineering teams of 50+ people”

Agencies and Services

  •  “Digital marketing agencies in the UK specializing in e-commerce clients with fewer than 30 employees”
  •  “Web development agencies that build custom Shopify stores for fashion and lifestyle brands”
  •  “SEO agencies in North America focused on B2B SaaS clients”

E-commerce and Retail

  •  “Direct-to-consumer brands in the beauty and skincare space with active social media presence”
  •  “E-commerce companies selling sustainable or eco-friendly products in Europe”
  •  “Subscription box companies in the US with between 10 and 50 employees”

Healthcare and Life Sciences

  •  “Biotech startups working on gene therapy or CRISPR applications in the US”
  •  “Digital health companies building patient engagement or telemedicine platforms”
  •  “Medical device companies focused on diagnostic equipment with FDA clearance”

Financial Services

  •  “Fintech startups in Europe building payment processing or money transfer solutions”
  •  “Wealth management firms targeting high-net-worth individuals with AUM over $100M”
  •  “Insurtech companies using AI for claims processing or underwriting”

Each prompt is specific enough to return targeted results while remaining broad enough to capture the full range of relevant companies. Notice how they combine multiple characteristics without becoming overly restrictive.

With an AI lead finder using natural language like Findymail, you can enter prompts exactly like these and receive qualified leads matching your description.

What to include and what to avoid

Effective prompts focus on externally observable characteristics. The AI analyzes publicly available information: websites, company descriptions, job postings, press releases. It cannot access internal data.

Include:

  •  Industry or sector
  •  Business model (B2B, B2C, marketplace, etc.)
  •  Company size (employee count ranges)
  •  Geographic location or market focus
  •  Technology focus or specialization
  •  Target customer type
  •  Business stage (startup, growth, enterprise)

Avoid:

  •  Revenue figures (not publicly available for most companies)
  •  Specific tools or software they use internally
  •  Hiring plans or growth rates (unless evident from public job postings)
  •  Profitability metrics
  •  Internal organizational structure details

For example, “companies using Salesforce as their CRM” is a problematic prompt because CRM choice is internal information. Instead, try “B2B companies with sales teams of 10+ people” which describes the same target indirectly through observable characteristics.

Refining your results

Your first prompt rarely produces perfect results. Effective prompt writing is iterative.

Start broad, then narrow. Begin with a prompt covering your general target market. Review the results to see what’s included that shouldn’t be. Add qualifiers to exclude those unwanted matches.

Learn from unexpected results. If irrelevant companies appear, understand why the AI included them. Their presence reveals how the AI interpreted your prompt. Adjust wording to clarify your intent.

Use multiple searches. Rather than crafting one perfect prompt, try several variations. “E-commerce agencies” and “Shopify development firms” and “online retail consultancies” might each surface different but relevant companies.

Test edge cases. Try prompts at the boundaries of your ICP. This helps you understand where the AI draws lines and whether those boundaries match your actual requirements.

Intellimatch from Findymail uses semantic search to understand prompts beyond literal keyword matching. This means your descriptions are interpreted intelligently, matching companies based on meaning rather than exact terminology.

Common mistakes and how to fix them

Several patterns consistently produce poor results.

Too vague. “Tech companies” is essentially useless. There are millions of tech companies. Add specificity: “Tech companies building what specifically, of what size, where?”

Too specific. “SaaS companies in Austin, TX founded in 2019 by Stanford graduates building AI-powered CRM with exactly 37 employees” will return zero or nearly zero results. Relax constraints that aren’t essential to your targeting.

Negative framing. “Companies that are NOT in healthcare” doesn’t help the AI understand what you do want. Focus on positive characteristics.

Internal data requests. “Companies with revenue over $5M” or “businesses that use Slack” ask for information the AI cannot access. Describe the same target using external indicators.

Conflicting criteria. “Enterprise software companies with 10 employees” contains inherent tension. Enterprise focus typically requires more scale. Ensure your criteria work together coherently.

Advanced techniques

Once you’ve mastered basics, several advanced approaches can improve results:

Layered searches. Run your ICP through multiple complementary prompts. “B2B SaaS in fintech” plus “payment technology startups” plus “financial software companies” together capture more of your market than any single prompt.

Temporal qualifiers. When relevant, add time elements. “Startups that have raised funding in the last 12 months” or “companies that have recently expanded to European markets” add precision.

Comparative descriptions. “Companies similar to [well-known company]” can be surprisingly effective. The AI can identify characteristics of known companies and find similar ones.

Exclusion by inclusion. Instead of “not enterprise,” describe the size range you want: “with 10-100 employees.” This achieves exclusion while giving the AI positive direction.

Putting it into practice

The best way to improve prompt writing is practice. Start with your current ICP and translate it into a natural language description. Test it. Review results. Refine.

Most users find their prompt writing improves dramatically within their first five to ten searches. The patterns become intuitive quickly. What feels unfamiliar at first becomes natural.

The payoff is substantial. Well-crafted prompts return highly targeted prospect lists in seconds. Poor prompts waste time with irrelevant results or miss ideal customers entirely. The skill is worth developing.

You can try these prompts with a free trial to see how natural language targeting performs on your specific ICP. Experiment with different descriptions and compare results.

More must-read stories from Enterprise League:

Related Articles