Consumer health
A lead pipeline that runs itself
The problem
A growing consumer health company needed a steady flow of qualified B2B leads across three very different markets: clinical partners, wellness and longevity brands, and corporate wellness buyers. The obvious move was to hire a sales development rep or hand it to an agency. Both are expensive, and neither solves the real issue, which is that “a good lead” means something different in each of those three markets. They wanted the pipeline full without adding headcount they’d have to manage.
The stack
| Tool | What it does | What it costs to run |
|---|---|---|
| Apollo | Finds companies matching each target market | Their existing subscription |
| Claude | Scores every company against the ICP, 0 to 10 | A few dollars a month in usage |
| Monday.com | Holds the scored leads as a working board | Their existing board |
| Klaviyo | Emails the qualified leads, segmented by market | Their existing platform |
| GitHub Actions | Runs the whole thing on a schedule | Free |
What was built
A pipeline that wakes up every 72 hours and does the work a junior rep would do, faster and without missing a cycle. It searches Apollo across all three target markets, then scores each company with Claude against a six-part rubric tuned to what actually makes a good fit in that market. Companies that clear the bar land in Monday.com and get pushed straight into the right Klaviyo list to start outreach. The ones in the maybe range are set aside for a human to look at before anything gets sent. Everything below the line is dropped, quietly.
Three layers of deduplication make sure the same company is never scored twice and the same person is never emailed twice, even months apart. And the whole thing is theirs to tune: the team changes how selective the pipeline is by editing one settings file, no code, no call to me.
Outcome
The company got a top-of-funnel that runs on its own, into the tools they already used, for a flat build fee and a few dollars a month to operate. Set against the alternatives, a sales development hire runs $60,000 or more a year, and an agency would have quoted a multiple of the build fee, that math is not close. The pipeline does the sorting so their team only spends time on leads worth a human’s attention.
What was harder than expected
The client’s first instinct was to fully automate it, scrape straight to email, no human in the loop. I pushed back. Their definition of a good lead was still thin, and thin criteria plus full automation is a fast way to burn your email reputation before you’ve built one. So we did the unglamorous part first: real, specific criteria for each of the three markets, then a scoring threshold that sends the uncertain ones to a person instead of the outbox. The lesson holds for every automation like this. Pointing AI at a vague target just sends bad email faster. The work that mattered wasn’t the code, it was getting sharp about who’s actually worth reaching.