Skip to content

Quick start

A mechanical fifteen-minute walkthrough that takes you from sign-in to scored accounts ready for review. This page gets you operational; for the why-behind-each-step framing and the rhythm you’ll settle into after the first run, read How LeadHunter works — before, after, or in parallel.

You’ll get the most out of the next fifteen minutes if you have these ready:

  • A LeadHunter account at leadhunter.humans2agents.com. Sign up there first if you haven’t.
  • A product you’re selling, with a public website you can paste in and a one-or-two-paragraph description in your head.
  • 2–3 example accounts (their websites) of customers or prospects you’d love to win. These calibrate scoring more than anything else.
  • A starter list of accounts — could be a small CSV from your CRM, a list of bookmarks, or just “I want to find every bike shop in Berlin.”

If you only have the first two, you can still get to scored accounts in this session — you’ll discover the rest via Google Maps in step 4.

A Company groups everything: products, accounts, campaigns, custom fields, saved filters, team members. Most users run one. If you’re an agency, run one per client.

  1. Sign in. You’ll land on the dashboard.
  2. Open the Company switcher in the top bar.
  3. Choose New company, give it a name (e.g. “Acme Outbound”), save.

Everything you do from now on is scoped to whichever company is active in the switcher — accounts, campaigns, statistics, custom fields, the lot.

A Product is what you’re selling. Its description, website, and example URLs feed the AI’s scoring model.

  1. Go to Products → New product.
  2. Fill in:
    • Name — e.g. “FleetMate”.
    • Website URL — your product’s public landing page.
    • Description — one or two paragraphs. The clearer the better; the AI uses this to draft the ICP and to score accounts. “AI-powered fleet management” won’t fly. “FleetMate is a maintenance scheduling tool for small commercial fleets (5–50 vehicles); we focus on independent garages and last-mile delivery operations in the EU.” is closer.
    • Communication language — default language the AI drafts outreach in. Campaigns can override.
  3. Save.

The Ideal Customer Profile describes who you want to sell to: sector, geography, firmographics, buying signals, anti-patterns.

  1. Open your new Product.
  2. Click Generate ICP. LeadHunter visits your website + reads your description and drafts a structured ICP — usually in 10–20 seconds.
  3. The draft opens in Pending approval. Review every field. Edit anything that’s off.
  4. Click Approve. The Product is now ready for scoring.

The single biggest lever on scoring quality after the ICP: add 2–3 example good URLs (websites of accounts you’d love to win) and 2–3 example bad URLs (accounts that look like fits but aren’t) to the Product. The model anchors heavily on these. Skip them and scores read fuzzy.

Three paths — pick whatever fits your starting data:

  • Paste a URL or domain under Accounts → Lookup. Works for a Google Maps URL, a website, or a bare domain like acme.com. Pick Quick mode for fast extraction; Deep mode (slower, more tokens) also extracts decision-maker contacts from team / about pages.
  • Upload a CSV or XLSX under Accounts → Import. The wizard previews the columns, the AI suggests a mapping (org_name → name”), you confirm, and LeadHunter runs deduplication against your existing database. Dry-run mode is recommended for anything over ~50 rows.
  • Search on Google Maps with a query like “bike shops in Berlin”. Each result becomes an Account, or merges into an existing one via the dedupe stack.

For this first run, aim small — 5–20 accounts is enough to see the loop working. The point of the first session is to validate the scoring + outreach flow on something you can actually skim. Once that loop feels right, scale up: a sharp scoring pass on a few hundred well-targeted accounts beats a noisy pass on thousands.

If any of these came in through Adwords forms, Instagram DMs, referrals, or other inbound channels, set the acquisition channel on each — they’ll auto-start at contacted instead of prospect. See Track inbound leads.

  1. Campaigns → New campaign, pick the Product you just made. Name the audience, not the product (“Berlin bike shops Q2”, not “Spring campaign”).
  2. Add some accounts — pick a saved filter for bulk-add (the reach estimator tells you how many will land before you commit), or add them one by one from the campaign’s accounts panel.
  3. Wait for scoring. LeadHunter scores each account against the Product’s ICP and writes typed reasons. With 5–20 accounts on a first run it’s a couple of minutes; a few hundred takes 5–15. You can keep working while it runs. Each account ends with:
    • An ai_score — 0–10.
    • A score_labelexcellent (≥8), moderate (5–7), or mismatch (<5).
    • 2–4 reasons, each tagged positive (concrete fit), negative (concrete gap), or neutral (context).
  4. Sort by score descending. Open the top accounts. Read the reasons — a 9/10 for the wrong reasons is still wrong.
  5. Approve the strong ones, reject the misses, set an override score where the AI was clearly off.

Every approve and reject you make is a calibration signal. Once a Product crosses ~10 reviewed accounts, those verdicts start feeding back into the scoring prompt — every subsequent batch gets sharper. The first campaign is when this matters most — be thorough.

Open an approved campaign-account, click the Conversation tab.

  1. Pick a contact (or add one).
  2. Mode AI draft, channel email (or whatever surface you’ll actually send from — channel matters because the dashboard funnel attributes responses against it later).
  3. The AI writes a fresh first-touch message in the account’s language. Edit until it reads right.
  4. Send through your own email client / LinkedIn / IG / WhatsApp.
  5. Click Log sent in LeadHunter — the message is recorded, outreach_status flips to sent, the account’s lifecycle status advances prospect → contacted, and the dashboard funnel counts the account as initiated.

When a reply comes back, paste it via Inbound paste. The funnel counts the account as responded, cohort-attributed back to the original initiation date.

A few common first-run hiccups:

  • The Generate ICP button finishes but the ICP reads vague. Most often the Product description is too thin. Two paragraphs of specific positioning — what you sell, who it’s for, who it isn’t for — beats one line every time. Regenerate after editing.
  • The first batch of scores all cluster around 5–6. Same root cause as above, plus missing example URLs. Add 2–3 good and 2–3 bad URLs on the Product and re-run scoring.
  • Accounts I added by hand show as duplicates. That’s the dedupe stack working — if the same Google Place ID, phone, or website already exists, the new data is merged into the existing row. Look in Accounts → Duplicates if you want to see what merged where.
  • My CSV import skipped a column. The wizard’s column-mapping step is where you tell LeadHunter what each column means. Re-run with dry_run=true first to see what would land before committing.

You’ve now got an end-to-end loop running. From here:

  • Log spend as it accrues — Adwords invoices, agency fees, internal labor hours. LeadHunter computes cost-of-acquisition per outreached / responded / closed customer. See Track campaign costs and CAC.
  • Read the funnel — the dashboard home shows initiated / responded / closed across today / 7d / 30d. Track week-on-week, not the latest day (cohort attribution makes the right edge low by design).
  • Settle into the rhythm — daily outreach + reply pasting; weekly fresh-score reviews; monthly duplicate cleanup + ICP refinement. See How LeadHunter works → The recurring rhythm.