Paid acquisition across Meta, Google, TikTok, Apple Search Ads and programmatic. I build agents with Claude Code and MCP that watch the spend, catch what dashboards miss and tell you what to do next.
Three roles, three kinds of growth problems. The full story is on the experience page.
Ran app and web acquisition across Google, Apple Search Ads, Meta and programmatic, directing several agencies and shifting spend toward the channels whose users stuck around.
Ran lead generation across six platforms, bidding to what each case type was worth, with AI agents flagging the campaigns that needed attention.
Oversee Disney+ subscriber acquisition spend across 7 international markets, with a pacing monitor I built in Claude Code catching issues early.
Every morning a paid media manager checks dozens of campaigns for the same handful of problems. This agent does it in seconds and ranks what to fix by the money at risk. Pick a scenario and run it.
Estimate. Token counts are typical for this agent; API list prices as of Nov 2026 (source). The Meta and LinkedIn APIs and MCP itself don't charge per call. A Claude subscription covers the same runs for a flat monthly fee.
A subscription app only wins on paid if a user's lifetime revenue beats what it cost to get them. Move the levers and watch LTV, payback and your maximum CPI change. Example inputs, not any company's real data.
Model: paid users = installs × trial rate × trial-to-paid. Each month a share churns. Revenue is net of the app store fee, over a 24-month horizon.
7,000 sample installs, five channels, five ways to hand out credit. Pick a model and see how each channel's installs and CPA change. Then switch on the geo-lift test to see which channels actually drove installs.
Five ads for Stridewell, a fictional fitness app at $12.99/month with a 7-day free trial. Same offer, same audience, $6K each on Meta over two weeks. Only the hook changes. Rank them by a different metric and watch the winner change. Sample results.
Every channel gets more expensive as you scale it. Split a monthly budget across four channels and watch total installs and blended CPI respond. Then let the optimizer find the mix that buys the most installs. Sample response curves.
| Channel | Spend | Installs | Avg CPI | Next-install CPI |
|---|
A 2-minute walkthrough of the full version: it runs the check, writes the morning note for the head of growth and drafts the budget changes. Coming soon.