Disco Like

Disco-Like

Lookalike company discovery. Give it 3-10 seed domains you know are a good fit; it returns hundreds of similar companies by domain, industry, and business characteristics. Useful for expanding from a small known-good list to a much bigger TAM without manual research.

When to use

• You have 3-10 customer domains you love, want "more like these"
• You want to expand a small client list into a full TAM
• You have an ICP description but don't want to manually build Prospeo filters
• Competitive / adjacent-market expansion

When NOT to use

• You need PEOPLE, not companies (use Prospeo or Blitz after this)
• Your ICP is extremely narrow or nascent (<5 seed examples exist)
• Budget is tight — DiscoLike charges per call + per record; see cost section

Two search modes

Mode A — Seed domains (most common)

npx tsx scripts/discover.ts --domains "clay.com,apollo.io,outreach.io" --country US --limit 500 --out lookalikes.csv

DiscoLike finds companies with similar characteristics (industry mix, employee count range, business type, tech stack) to your seeds.

Mode B — Natural-language ICP

npx tsx scripts/discover.ts --text "B2B SaaS companies selling outbound sales software to RevOps teams" --country US --out lookalikes.csv

Uses DiscoLike's text matching. Less precise than seeds, but useful when you don't have named comparables.

Hybrid mode

npx tsx scripts/discover.ts --domains "clay.com" --text "outbound automation" --country US --out lookalikes.csv

Combines both — starts from seeds, expands via text semantics.

Negation (exclude existing customers / competitors)

npx tsx scripts/discover.ts \
  --domains "clay.com,apollo.io" \
  --negation-domains "yourcompany.com,yourbigcustomer.com" \
  --country US \
  --out lookalikes.csv

Always include your own domain + existing customers + known-unfit competitors. Saves enrichment cost downstream.

Inputs

• DISCOLIKE_API_KEY (env) — from DiscoLike dashboard
• Either --domains or --text (at least one required)
• Optional: --negation-domains, --country, --limit, --max-companies

Outputs

CSV with columns: domain, company_name, industry, headcount_range, headcount, location_country, location_state, location_city, linkedin_url, description, source

All rows have source=discolike so you can mix with other list-builder outputs without collisions.

Cost

• $0.10 per API call + $2.00 per 1,000 records returned
• Default page size: 100 per call
• A 500-company discovery = ~5 calls + 500 records ≈ $1.50
• A 10,000-company discovery ≈ $10 + $20 = $30

Compare to Prospeo, which charges per export. DiscoLike is typically cheaper per company-discovered but more expensive per enriched contact (DiscoLike gives companies, not people).

Required step: Qualify with /icp-prompt-builder

This is a required step. Do not skip it.

Before pulling 5,000 companies, run DiscoLike on a small sample (50-100), then invoke /icp-prompt-builder:

1. Evaluate which of the 50 are actually good ICP fits
2. Refine your ICP description / negation list based on what DiscoLike returned
3. Only then scale to 5,000+

Why required: DiscoLike lookalike results are only as good as your seed domains. If 80% of the first 50 are wrong, you need to change seeds, not pay to pull more. At $0.10/call + $2/1K records, a wrong-seeded 10K pull costs $20-$30 in DiscoLike fees AND cascades into wasted email-finder fees downstream. Qualifying the first 50 catches bad seeds before they become expensive.

Recommended flow

1. /icp-onboarding → nail down seed companies (your best 5 customers)
2. /disco-like --domains="seed1,seed2,..." --limit=100 --out=sample.csv → sample run
3. /icp-prompt-builder → score the sample, tune ICP prompt
4. If sample quality is high, scale: /disco-like ... --limit=5000 --out=full.csv
5. /blitz-list-builder --domains-file=full.csv → find decision-makers at each
6. /email-waterfall → fill in emails
7. Upload to Smartlead

API details (reference)

Base URL: https://api.discolike.com/v1

Auth: x-discolike-key header

Endpoints:

| Method | Path | Purpose | |---|---|---| | GET | /count?domains=X&text=Y | Total matching companies (before paying to pull) | | GET | /discover?domains=X&text=Y&country=Z&limit=100&offset=0 | Paginated lookalike results | | GET | /bizdata?domain=X | Detailed data for a single domain |

Data returned per company:

• domain, name, description
• industry_groups (weighted dict — script takes top industry)
• employees (range string like "51-200")
• address (country, state, city)
• social_urls (script extracts LinkedIn company URL)

Rate limit: Conservative — script throttles at 5 concurrent, 10 req/sec. No 429s observed on normal runs.

Common gotchas

• Seed domains must be clean bare domains. clay.com works, https://clay.com/ doesn't.
• Text mode is fuzzier than you think. "Outbound sales" returns SaaS, agencies, consultancies — broad. Tighten with seeds.
• No people data. DiscoLike is company-level. Always chain with Blitz or Prospeo for contacts.
• Non-US coverage varies. US has deepest data. EU/APAC coverage is thinner; count may be misleading.
• Check the count FIRST. Before paying for 10,000 records, run /count to confirm the universe actually has 10,000. Many narrow ICPs top out at 500-2000.

Scripts

• scripts/discover.ts — main search + CSV output
• scripts/count.ts — pre-check universe size before paying
• scripts/bizdata.ts — single-domain lookup

What to do next

Run /icp-prompt-builder on your 50-company sample (required step above). Then either:

• /blitz-list-builder to find owner contacts at each filtered domain, OR
• /list-quality-scorecard directly if this is companies-only and you'll enrich another way

Or wait: if the 50-sample ICP fit was poor (<40% matches), don't scale. Change your seed domains and re-run with better inputs.

Related skills

• /icp-onboarding — defines the seed domains you'll use
• /icp-prompt-builder — quality-check the first 50 results before scaling
• /blitz-list-builder — chain to find contacts at each discovered company
• /email-waterfall — fill missing emails after Blitz
• /cold-email-starter-kit → 06-list-building-prospeo.md for broader list-building patterns

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: growthenginenowoslawski
• Source: growthenginenowoslawski/coldoutboundskills
• License: MIT

Install and usage instructions live in the source repository linked above.