Lead Generator

Most "lead generator" tools hand you a spreadsheet of names and let you figure out who's actually worth calling. This template does the opposite: it asks for your product and ICP in plain language, runs one cost-capped Apify search, scores every result against real evidence (not assumptions), and only writes to Google Sheets once you've approved the shortlist. This template acts as a starting point: open any agent and use Use AI to help draft or update its instructions. Adding new agents or connecting new tools is done manually in the canvas.

What it does: Turns a plain-language ICP into a capped, evidence-scored B2B lead list, delivered to Google Sheets only after you approve it. Best for: Agencies, B2B founders, freelance consultants, and SDRs who need a real outbound list, not a raw contact dump. Apps used: Apify, Google Sheets. Setup time: About 10 minutes.

Last verified from workflow config August 2026. Includes common failure modes and fixes.

Problem this solves

Most DIY B2B lead generation setups fail in the same three places:

• Before: Uncapped scraper runs quietly rack up charges because nobody set a spend ceiling. After: every run has a hard maxTotalChargeUsd cap and a per-conversation lock so a rerun can't double-charge.
• Before: A list of 500 contacts with no reason to call any of them first. After: each lead carries a scored fit and why-now trigger with a source URL and date, not a guess.
• Before: The same person gets contacted twice because nobody checked the sheet first. After: the agent reads existing rows and a contacted ledger before writing anything.
• Before: A generic ICP returns a pile of consumer or local-business "leads" that were never reachable in the first place. After: the agent says up front when your audience doesn't fit a B2B database, before it spends anything.

How do you build a B2B lead generation agent yourself?

Strip away the branding and every B2B lead generation tool does the same five jobs: capture an ICP, source candidates from a database or scraper, dedupe and score them, enrich the top few with real evidence, and hand off the approved list to a CRM or sheet. Whether that's Apollo, Clay, or a template like this one, only who owns each step changes.

Building it yourself typically means putting together:

• A scraping layer: an Apify actor, a Playwright/Scrapy script, or a paid database API, plus proxy and rate-limit handling if you're scraping directly.
• A scoring layer: code (often a Claude Code or Codex-generated script) that normalizes raw rows, removes duplicates, and ranks candidates against your own fit criteria.
• An enrichment layer: bounded website and LinkedIn lookups for your top candidates only, because enriching everything burns budget fast.
• A handoff layer: writing the approved batch into a CRM or sheet without creating duplicates on repeat runs.

The real cost of the DIY path shows up after launch, not before it: teams building this from scratch report scrapers that "fetched thousands of items even though strict limits were configured," with platforms that kept charging for runs nobody meant to start. That's the exact failure mode this template's spend cap and per-conversation run lock exist to close.

If you'd rather not maintain that pipeline yourself, this template pre-wires the same five layers into one agent, with Apify handling sourcing and enrichment and Google Sheets as the handoff. The setup guide below covers the roughly 10-minute version. If your leads are local businesses rather than B2B contacts at companies, the Web Lead Extractor template covers that route instead.

Setup guide

Just as you click on 'Use this template', you are taken to the Agent Space, where tools and skills are already connected. Give the skils a couple of seconds to auto-install and then all that's left is to connect your account to the tools(Google Sheets).

Setup steps for the template and the starter prompts to get the template running are attached within the agents space for your ease.

the Agent Space Looks like this:

Step 1: Add the template and connect Apify and Google Sheets

Open the template and go to the Apps tab on the two tool nodes: one for Apify, one for Google Sheets. Connect your own account to each and complete any permission prompts Apify shows for enabling its allowlisted lead-sourcing actor.

Both tools ship disconnected by default, so nothing will run until you authorize them. If you don't wish to connect your account to Apify, Apify can use SketricGen credits to run.

When you click on Connect Google Sheet a window opens through which you can connect your Google Account using OAuth.

Step 2: Give the agent your product context

Send any first message. The agent

• checks your project files for a product website or description
• crawls the official site once if that context is thin
• drafts an editable Business & Audience Profile: primary customer, economic buyer, pains, use cases
• a proposed ICP (titles, seniority, industries, company size, geography, exclusions).

If your project holds more than one client or brand folder, it asks which one this list is for before doing anything else.

Step 3: Confirm the ICP, lead cap, and enrichment depth

Review the drafted profile and edit anything that's wrong. On first use, the agent batches two open decisions into a single question: your per-run lead cap (it shows a recommended default) and enrichment depth (how many top candidates get deep research, default five). Nothing paid starts until you confirm the profile or naturally ask for a count, like "give me 100 leads," which also sets the cap.

Step 4: Run a capped search and review the scored shortlist

Once confirmed, the agent runs one cost-capped Apify search, checkpoints the raw results immediately, then deeply researches only the top candidates. You'll see up to five fully enriched rows in chat (fit score, contactability, tier, why-fit, why-now, readiness) plus a compact list of the rest, with counts for how many were found, enriched, and still provisional.

The reuslting Google sheet includes all they key details for your leads and you can then start on actual acquistion through cold outreach pipelines or through direct connect.

Customization knobs: per-run lead cap, enrichment depth (top 3 or top 5, website-only or website plus LinkedIn), ICP fields (titles, seniority, industries, size, geography, exclusions), and destination spreadsheet. Open the agent and use Use AI to help refine its instructions if you want to change tone or add extra qualifying questions. Adding new agents or tools is still a manual canvas step.

Step 5: Approve and deploy

Test it by asking "who have we already contacted, skip them" on a repeat run to confirm suppression is working before you rely on it daily. Deploy it simply in the agent space by pushing publish now, or connect it to other apps simply using OAuth or username + password connect (depending on each app). You can also set-up schedule runs so that the template runs once a week or so on.

Common issues and fixes

• Apify or Google Sheets shows as disconnected mid-run. Reconnect the account from the Apps tab; the agent won't start a paid search on a disconnected tool.
• The ICP is too broad and returns weak matches. The agent will say so and offer the nearest reachable audience before spending, rather than return junk. Narrow titles, industries, or geography and try again.
• A second search request in the same conversation gets skipped. This is the one-paid-run-per-conversation lock working as intended; it reuses the first run's results unless you explicitly ask for a new search.
• Consumer, creator, or local-storefront ICPs come back thin. The connected source is a database of people at companies, not a local-business directory. The agent flags this rather than fake a fit.
• You don't want to connect your own Apify account. SketricGen credits can run the actor instead; connect your own account later if you want runs billed to your own Apify usage.

Who it's for

• Lead generation and marketing agencies running repeated, capped prospecting across multiple clients. The profile is stored per client, so nothing bleeds between accounts.
• B2B and SaaS founders building their first outbound pipeline who want a cost ceiling before they hand it to an LLM.
• Freelancers and independent consultants scoping a client's ICP without paying for a full database seat.
• SDRs who need a fast, scored shortlist instead of a raw export to work through manually.

Not a fit: consumers, creators, students, and local storefronts. The connected source is a B2B people-at-companies database, and the agent will tell you plainly rather than return low-quality matches.

What the agent does

This template can:

• Discover your product from project files first, crawling your site only once if context is thin.
• Draft an editable Business & Audience Profile covering buyer, pains, use cases, and a proposed ICP.
• Route your audience to the right B2B source type, or say plainly when your audience isn't a fit.
• Run one cost-capped Apify search with a hard spend ceiling and a per-conversation lock against double-charging.
• Checkpoint raw results immediately after the paid run, before any enrichment starts.
• Deduplicate and score every candidate by persona and seniority before spending anything further.
• Enrich only the top few with bounded website and LinkedIn research, never the full batch.
• Separate fit from contactability so a clean email never masks a weak match.
• Ask for explicit approval before writing anything, showing counts found, enriched, and provisional.
• Suppress prior contacts automatically on every future run.

How it works

1. You send a message describing your product, or just say "give me 100 leads" if the profile is already confirmed.
2. The agent builds or confirms your Business & Audience Profile, routing your audience to a reachable B2B source.
3. It confirms the lead cap and enrichment depth the first time, then remembers your answer.
4. It runs one capped Apify search, polls it, and reads only that run's own results.
5. It normalizes, dedupes, and scores every candidate, saving a checkpoint before enrichment starts.
6. It deeply researches only the top candidates, scoring fit and contactability as two separate numbers.
7. It shows you the shortlist and waits for approval before writing anything to Google Sheets.

Requirements

You'll need an Apify account with the template's allowlisted lead-sourcing actor enabled, and a Google Sheets account to receive the approved list. You'll also need to give the agent your product context, either through project files or a live website URL, so it can draft an accurate ICP.

No CRM or outreach-sending app is wired in by default. The Sheet is the handoff point, not an inbox. If you want to route approved leads further, SketricGen's library of 2,000+ connectable apps covers CRMs and outreach tools as an optional add-on, and a template like Structured Output is a useful reference if you want to adapt the evidence/scoring JSON contract into your own downstream pipeline.

Apps used

Use cases

Agency running parallel client searches Confirm a separate profile per client, then run capped searches for each without mixing data between accounts. Best for: agencies managing multiple B2B clients at once.

SaaS founder building a first outbound list Describe your product once, confirm the drafted ICP, and get a scored shortlist instead of a raw contact export. Best for: early-stage B2B/SaaS teams without a dedicated SDR yet.

Freelance consultant scoping a client's ICP Use the drafted Business & Audience Profile as a deliverable on its own, then run the search once the client signs off. Best for: independent consultants pitching or kicking off outbound work.

SDR getting a fast, scored shortlist Skip the raw database export and start from a shortlist that already separates fit from contactability. Best for: sales reps who need a working list this week, not a training project.

Once a shortlisted lead turns into a scheduled call, pairing this with a template like Pre-Meeting Prep briefs you on the contact beforehand.

Example prompts and outputs

Prompt: "Give me 100 leads." Output: A shortlist showing up to five fully enriched rows in chat (Name, Title, Company, Size, Fit, Contactable, Tier, Why Fit, Why Now, Evidence, Readiness), plus the rest listed compactly, with a note on how many were found, enriched, and still provisional. Ends with an approval choice: approve all, approve specific rows, or adjust the ICP.

Prompt: "Who have we already contacted? Skip them." Output: The agent checks the contacted ledger and destination sheet first, then excludes anyone already present from the next run's results with no manual list-cleaning required.

Why you need this template

• Spend safety by default: a hard cost ceiling and a one-paid-run-per-conversation lock mean an interrupted turn can't double-charge you.
• Evidence over guesswork: every why-fit and why-now claim is checked against a name-swap test and a real source URL, not assumed.
• Nothing writes without your approval: the Sheet only gets touched after you sign off on the shortlist.
• No repeat outreach: a contacted ledger and a read-before-write check stop the same person from being contacted twice.