How I Automated My Entire Outreach Pipeline (And You Can Too)

I'm going to walk you through something I built that fundamentally changed how I approach client acquisition. Not a theory. Not a framework from a business book. An actual system I built and ran, with real numbers, real failures, and real lessons.

The short version: I built a pipeline that automatically finds businesses that need a better web presence, generates custom demo websites for them, and sends personalized outreach showing them exactly what their new site could look like.

The result: 150+ custom demo sites generated and personalized outreach sent at a scale that would have been physically impossible to do manually.

Here's exactly how it works, what it cost, what went wrong, and what I'd do differently.

The Problem I Was Trying to Solve

I run Be Curious Labs as a one-person operation building AI automation and web solutions for small businesses. My ideal clients are local service businesses: pool companies, contractors, landscapers, home service providers. The kind of businesses that are great at what they do but often have terrible websites (or no website at all).

The traditional outreach approach looks like this:

1. Search Google for businesses in your target market
2. Visit each website manually
3. Write a personalized email explaining how you could help
4. Send it
5. Follow up a few days later
6. Repeat 500 times

At maybe 15-20 minutes per lead (research + email drafting), that's 125-167 hours of work to reach 500 businesses. That's over three weeks of full-time work just on cold outreach. For a solopreneur who also needs to do the actual client work, that's not realistic.

So I automated the entire thing.

Step 1: Finding Leads at Scale

The first challenge was identifying businesses that actually need help. Not every business is a good prospect. I needed ones with poor or missing web presences that would benefit from a modern site.

The Approach

I used Claude Code to build a scraping system targeting business directories and search results for specific service categories in specific geographic areas. Think "pool companies in Orlando" or "landscapers in Clermont."

For each business found, the system captures:

• Business name and contact info
• Current website URL (if they have one)
• Google Business profile data
• Basic info about their services

The Filtering

Raw scraping gives you a lot of noise. Not every result is a real prospect. The system filters based on:

• No website or broken website: highest priority
• Outdated website (looks like it was built in 2010): strong prospect
• Decent website but missing key features (no mobile optimization, no contact form, no reviews): moderate prospect

This step alone saved dozens of hours. Instead of manually Googling and clicking through results, I'd wake up to a filtered list of businesses that actually matched my criteria.

Step 2: Generating Custom Demo Sites

Here's where it gets interesting. Cold emails that say "I can build you a great website" are easy to ignore. But cold emails that include a link to a fully built demo of what their new website could look like? That stops people in their tracks.

How the Demo Generation Works

For each qualified lead, the system:

1. Gathers content: Pulls information from their existing web presence, Google Business profile, and directory listings
2. Generates copy: Claude writes service descriptions, about sections, and calls-to-action tailored to their specific business
3. Builds the site: A template system generates a complete, responsive website using their business name, services, location, and color scheme
4. Deploys it: The finished demo goes live at a unique URL under our demo subdomain

Each demo site isn't a generic template with their name slapped on it. It's a genuinely personalized site with their services, their location, and copy that speaks to their specific customers.

For our higher-touch prospects, we go further. An AI agent team (UI Designer + Frontend Developer running through Claude Code or Codex) creates fully custom designs with unique color palettes, typography, and layouts pulled from the business's actual branding and social media presence.

The Tech Stack

• Claude for all the copywriting. Given a business name, their services, and their location, it generates surprisingly good website copy. The key is detailed prompts that capture the right tone for each industry.
• Claude Code and Codex for building the generation scripts, processing scraped data, and creating custom site templates programmatically
• QwikCity for our custom builds. Static site generation gives us fast, SEO-friendly pages.
• S3 + CloudFront for hosting. Every demo site gets uploaded to S3 and served through CloudFront. Cost per site is effectively zero. Fractions of a penny per month.
• Wildcard subdomains: Each demo lives at something like joes-pool-service.demo.becuriouslabs.com. A wildcard DNS record and CloudFront distribution handle the routing automatically.

The total hosting cost for 150+ demo sites? Under $5/month. S3 storage and CloudFront bandwidth for static sites is absurdly cheap.

Step 3: Personalized Outreach

This is where everything comes together. A demo site is worthless if nobody sees it.

The Email System

For each lead with a generated demo, Claude crafts a personalized email. Not a mail merge with {FIRST_NAME} tokens. An actually personalized message.

Here's roughly what a real email looked like:

Hi Joe,

I came across Joe's Pool Service while researching pool companies in the Winter Garden area. I noticed your current website might not be showing off the quality of work you do.

I took the liberty of putting together a quick concept for what a modern site for your business could look like: [joes-pool-service.demo.becuriouslabs.com]

It's fully responsive, loads fast, and is designed to help customers find and contact you easily. No obligation. I just thought you might find it interesting.

If you'd like to chat about making something like this your actual site, I'm happy to jump on a quick call.

Best, Logan

Every email references their specific business, their location, and their industry. The demo link makes it tangible. They can click and immediately see what we're proposing. Not a vague promise, but a finished product.

The Sending Infrastructure

• Amazon SES for email delivery. High deliverability, $0.10 per 1,000 emails.
• Custom sending domain with proper SPF, DKIM, and DMARC records. Critical for avoiding spam filters.
• Rate limiting: sends are throttled to avoid triggering spam detection. No blasting 500 emails at once.
• Unsubscribe handling: every email includes a proper unsubscribe link. This isn't optional. It's legally required and ethically necessary.

Step 4: Tracking and Follow-Up

Sending emails is only half the battle. You need to know what happens next.

What We Track

• Email delivery status: Did it actually arrive? Any bounces?
• Demo site visits: When someone clicks the demo link, we know. This is the strongest signal of interest.
• Response tracking: Replies get flagged for immediate personal follow-up

The Follow-Up Logic

• Demo visited, no reply within 48 hours: Send a brief follow-up. "I noticed you checked out the demo. What did you think?"
• No demo visit, no reply within 5 days: One gentle follow-up, then done. No one likes being hounded.
• Reply received: Immediately flagged for personal response. This is where the human (me) takes over.

The system knows when to push and when to back off. Aggressive follow-up sequences are a great way to get marked as spam and tank your sender reputation. We keep it to two touches maximum.

The Results: What Worked

Let's talk numbers.

• 150+ demo sites generated across multiple service industries
• Personalized outreach sent at scale, what would have taken weeks of manual work happened in days
• Demo click-through rates significantly higher than typical cold email. Having a tangible demo to show makes a real difference
• Multiple conversations started from a single outreach campaign
• Total infrastructure cost: under $20/month (SES + S3 + CloudFront)

The biggest win wasn't any single metric. It was the ability to test the entire approach quickly. Instead of spending months manually reaching out to 50 businesses, I could reach hundreds and learn what worked within weeks.

The Results: What Didn't Work

I promised honesty, so here it is.

Cold Email Is Still Cold Email

Even with personalized demos, you're still reaching out to people who didn't ask to hear from you. Response rates for cold outreach are inherently low. The demo dramatically improved engagement compared to generic cold emails, but it didn't turn cold outreach into warm leads.

The businesses that responded were genuinely impressed by the demos. The ones that didn't respond... well, most people ignore most emails. That's reality.

Not Every Demo Was a Winner

AI-generated content is good, but it's not perfect. Some demos had slightly off copy, like a pool company described as offering "aquatic maintenance solutions" instead of just "pool cleaning." These needed human review.

I learned to build a review step into the pipeline. Fully automated end-to-end sounds great in theory, but a 5-minute quality check on each demo saved me from sending embarrassing outreach.

Industry Targeting Matters Enormously

Some industries responded much better than others. Home service businesses with outdated or missing websites were the sweet spot. Businesses that already had decent sites (even if they weren't great) were much harder to convert. They didn't feel the same urgency.

The lesson: automate the process, but invest your strategic thinking in targeting. A perfectly automated pipeline pointed at the wrong audience still produces zero results.

Deliverability Is a Constant Battle

Even with proper SPF, DKIM, and DMARC, email deliverability requires ongoing attention. Warming up a new sending domain takes time. Sending too fast triggers rate limits. Different email providers have different spam thresholds.

I spent more time on deliverability than I expected. If I were starting over, I'd invest in domain warming earlier and send even slower at the beginning.

What I'd Do Differently

If I rebuilt this pipeline from scratch today:

1. Start with fewer, higher-quality leads. Instead of casting a wide net, I'd spend more time on targeting criteria upfront. Ten perfect-fit prospects beat a hundred maybes.

2. Add a human review step earlier. Full automation is tempting, but the quality improvement from a quick human check is worth the time investment.

3. Warm the sending domain for longer. I was too eager to start sending. A properly warmed domain delivers better results for months afterward.

4. Test messaging before scaling. I should have sent 20 emails manually first, iterated on the messaging, and then automated the version that worked. Instead, I automated first and iterated in production.

5. Build the follow-up system from day one. The initial launch focused on first-touch emails. The follow-up logic came later. Since most conversions happen on the second or third touch, I left opportunities on the table early on.

The Bigger Picture

Here's what I want you to take away from this, beyond the specific technical details.

This pipeline isn't magic. It's a combination of existing tools (web scraping, AI content generation, static site hosting, email delivery) connected together with automation. Each individual piece is accessible and affordable. The value is in the system design.

The traditional approach to outreach is labor-intensive, inconsistent, and doesn't scale. The automated approach trades upfront engineering effort for ongoing, scalable execution. Once the pipeline is built, the marginal cost of reaching one more prospect is nearly zero.

And that changes the math on everything. You can test more markets, try more messaging angles, reach more prospects, and learn faster. All without working more hours.

You Don't Need to Build All This Yourself

I built this because I'm an engineer and I genuinely enjoy this stuff. Constructing the pipeline was half the fun.

But you don't need to love infrastructure to benefit from this approach.

If you're a service business doing manual outreach and you're ready to scale, this kind of system can transform your client acquisition. The technology is proven. The costs are minimal. The ROI is measurable.

At Be Curious Labs, this is exactly the kind of system we build for clients. Not a generic SaaS tool. A custom pipeline tailored to your specific business, your target market, and your outreach style.

We handle the scraping, the AI content, the demo generation, the email infrastructure, and the tracking. You handle the conversations with interested prospects.

Want to see what an automated outreach pipeline could look like for your business? Check out our work or reach out to get started.

The leads are out there. The question is whether you'll find them one by one, or build a system that finds them for you.