How would you approach the cost stack or workflow for scaling automated outreach with scraping and GPT‑4 parsing?
GPT‑4 token costs are low—about $0.40 per million input tokens, so even a blended rate of $0.50 per million is cheap. The bigger cost is scraping. Firecrawl costs $19/month for 3,000 pages or $99/month for 100,000 pages, making the latter far more cost‑efficient. At roughly $1.70 per thousand pages you can scrape large volumes affordably. Combine a cheap scraper (e.g., Firecrawl’s $1‑$2 per thousand pages) with GPT‑4 for parsing, and keep an eye on token usage. Overall the workflow should stay well under $30 for 2,000 qualified emails, likely closer to $3 when optimized.
I recommend using Apify to scrape leads, which provides Apollo scrape databases for about $150 to $2 per thousand leads. These lists have a high bounce rate, so run them through a verification tool like NeverBounce, Million Verifier, or Mail.so. After verification, you'll have around 1,000 emails costing $2 to $3. For sending, my go-to stack is Zap Mail at about $3 per mailbox per month. To send 1,000 emails a month, you need roughly two mailboxes, costing $5 to $10 monthly. Add a domain at $10 per year (~$1/month). So far, you're at about $10. Then you need a sending plan: Zap Mail's $37 or $39 per month plan brings the total to roughly $50 per month. If you want to scale beyond 1,000 emails a month, remember that cheap cold email only makes sense if you're sending sufficient volume; otherwise, reconsider the effort.
Use a done-for-you mailbox provider; I recommend Zapmail (with affiliate code Nyx for 10% off). Get 9-12 mailboxes at about $3.50 each per month; with the discount that's roughly $33 total. Buy domains—one domain per three mailboxes—so about 3-4 domains at ~$10 each, adding ~$35. Set up Instantly: the $37 plan works if you're bottlenecked, bringing the total to about $105, leaving $15 for Make.com (or n8n) for personalization. You can also scrape leads for free using Apollo or Apify, then personalize them for a few dollars per thousand. Overall you can start with around $70.
He outlines a low‑cost cold‑email stack for a landing‑page agency. First, scrape leads on Appify (or similar) – a typical scraper costs about $150 for ~3,000 leads, though many are old or low quality, so expect roughly half to be usable. Next, run those leads through a verification service such as Million Verifier to weed out bad addresses; verifying ~10,000 emails costs around $20. Then set up mailboxes with a service like Airmail at $4 per mailbox; for ten mailboxes that’s $40/month, plus a growth/warm‑up tool (~$40/month). You’ll also need to warm up domains for about 21 days. Overall, you can expect to spend roughly $100 per month to run the outreach stack. The process is: acquire leads, verify them, configure mailboxes, send cold emails daily, and monitor deliverability.
Great question. I was in your exact shoes not long ago. While running this channel I still had a few retainers paying $3,000‑$3,500 a month, but the work was highly custom, technical, and required endless revisions – a massive pain. I kept taking on too much, not the right kind of work. You’re essentially running unscoped, general‑purpose retainers without clear deliverables, so clients pay well but you can’t scale. My recommendation is to standardize your retainer packages. Treat your cold‑outreach agency as a product: pick one to two lead sources, automate or semi‑automate lead scraping, and create three tiered packages (e.g., 5k, 10k, 20k leads per month). Bundle the number of mailboxes/outreach volume into each package, price them with modest discounts for larger tiers (e.g., 25% off moving from 5k to 10k, another 25% from 10k to 20k). This gives clients a clear value proposition and lets you charge proportionally without a linear price jump. Standardize the validation/verification process with SOPs: pull an Apollo search URL, scrape leads into a Google Sheet, run them through a verification service, and keep everything repeatable. By limiting lead sources, automating scraping, and offering defined packages, you eliminate endless custom work and can scale efficiently.