Skip to content

First Google Ads Campaign Setup, Vibe-Coded Ad Management, and Connecting Claude as Your AI Media Buyer

Members Only

Most marketers running paid traffic on one platform feel a real wall when they cross over to a new one. Facebook buyers freeze in front of the Google Ads interface. Google buyers get lost in YouTube’s targeting menus. And in 2026, both of them are watching solo operators vibe code their own ad management platforms in four weeks while their five-person engineering teams are still scared to touch the codebase. This week’s mastermind covered exactly that crossover, with a live first-campaign walkthrough, a multi-network ad SaaS demo, and a clear playbook for plugging Claude into your ad accounts as your AI media buyer.

What We Covered

  • First Google Ads Campaign Setup – Live coaching for a Facebook buyer launching real estate search ads
  • Reusable Negative Keyword Lists – One list that protects every future campaign you ever launch
  • YouTube Ads in 2026 – Why placement scraping is dead and what actually works now
  • Native Network Reality Check – Real numbers on Bigo, MediaGo, Taboola, and Outbrain
  • Click Fraud at Scale – 50 percent bot traffic on some networks and how to detect it
  • Vibe-Coded Ad Management Platform – A multi-network SaaS built solo with Claude in four weeks
  • Claude as Your Media Buyer – Connecting Facebook and Google Ads APIs through MCP

The First Google Ads Campaign Trap

The biggest mistake new Google advertisers make is using their old Facebook playbook. On Facebook you can slap a landing page together and the algorithm carries the campaign. On Google, the platform punishes thin setups and rewards advertisers who use every asset extension and configuration available.

  • Bid strategy and budget audit first – Modify the columns view to expose bid strategy and bid strategy type before you do anything else
  • Drop daily budget to a low single-digit number – You will see plenty of garbage keywords on day one and you do not want to pay for them at scale
  • The “ad isn’t showing” error is normal – Cache propagation across Google’s ad serving network takes thirty minutes to two hours after enabling a campaign
  • Asset extensions matter on Google – Calls, sitelinks, callouts, structured snippets, and lead forms together fatten your ad and lift CTR even when nobody clicks the extensions themselves[1]Use as many asset types as possiblesupport.google.com
Pro Tip: Build one default negative keyword list with generic junk like “free”, “admin”, and “login”, save it as a reusable list, and attach it to every new campaign. You only have to do this once.[2]Create and apply negative keyword lists to campaignssupport.google.com

YouTube Ads: Demand Gen Replaced Placement Scraping

YouTube ads used to be a placement game. You scraped every video in your niche, dropped your ads on top of those placements, and reaped traffic from people already watching content related to your offer. That workflow is gone. Google now funnels everyone into Demand Gen campaigns, which behave like Facebook ads where you hand the algorithm your creative and trust it to find your audience.[3]Video Action campaigns are being upgraded to Demand Gensupport.google.com

  • In-market audiences barely work alone – Targeting people in-market for pest control still served against Minesweeper streams and cat videos
  • Demographic and retargeting layers carry the weight – Age, gender, household income, and warm audiences are the only signals that consistently produce
  • Broad-market offers still scale – Medicare, ACA, and gift card promos can spend hundreds of thousands monthly on YouTube because the audience is wide enough for the algorithm to find
  • Tight verticals need creative work – Auto insurance and pest control are harder because the buyer’s mindset has to align with the moment of the ad

The Reality of Native Networks in 2026

Outside of Google and Facebook, the native ad ecosystem is a mixed bag. The cheap inventory is real, but so is the bot traffic, and most networks need significantly larger test budgets than the major platforms.

  • Bigo Ads ships in-app traffic at extremely low CPM and CPC; about half is junk but the converting half can profit at four-figure daily spend
  • MediaGo placements on MSN and weather.com burned through real money before reaching the cost-per-lead target
  • Taboola and Outbrain need five-figure test budgets and only really work with broad-market offers because their interest targeting is weak
  • Microsoft Ads ships massive bot traffic despite claiming aggressive filtering; custom click tracking exposes it quickly
  • Google’s YouTube and search remain the cleanest filtered traffic on the open internet
Warning: Even networks that should be filtering aggressively are 30 to 50 percent bot traffic. If you are not running your own click fraud detection, half your spend on those networks is funding scraping companies.[6]2025 Imperva Bad Bot Report (bots = 51% of web traffic)imperva.com

Join the Vibe Coding & Marketing Mastermind

Weekly strategies, updates, and vibe coding resources.

Vibe Coding an Entire Ad Management Platform Solo

Maybe the most striking part of the session was watching one operator demo a multi-network ad management SaaS he built solo with Claude over four weeks. The platform connects to Facebook, Google, Bigo, MediaGo, Newsbreak, Taboola, and Outbrain through their APIs, automates negative keyword detection with DeepSeek and web search, generates ad copy and creative, optimizes bids by state and hour-of-day performance, and runs an AI chat interface on top of the whole thing.

  • Replaces fragmented tooling – Creative library replaces Google Drive, AI image and video generation replaces stock providers, competitor analysis replaces ad spy tools
  • Bring your own API keys for AI – Krea AI, OpenAI, DeepSeek, and others plug in via personal keys instead of platform-level subscriptions
  • Slack-driven workflow over IDE – Each Slack channel becomes a project, each conversation becomes a session, and token usage drops compared to one long Cloud Desktop chat
  • What used to take a ten-person team – The kind of platform that previously required a funded engineering org now ships solo over a month of focused work
Key Insight: Established engineering teams sitting on legacy code are often slower than solo vibe coders because they are scared to let AI touch the codebase. The compounding speed advantage shifts to whoever moves first.

Refactoring an Entire Codebase to TypeScript Over a Coffee Break

The rest of the session covered a real-world refactor story that demonstrates the speed AI brings to maintenance work. A SaaS originally built on a year-and-a-half-old framework was migrated to the current major version in roughly thirty minutes. While that ran, the same project was then refactored from JavaScript to TypeScript in another thirty minutes back-to-back. Yes, both refactors introduced bugs, but every bug was fixable through one more conversational round trip.

  • TypeScript reads more like English – AI works faster and cleaner in it because the language itself is closer to natural language
  • Refactor cost is approaching zero – The kind of upgrade that used to be a two-month sprint becomes an afternoon
  • Bugs are conversational, not catastrophic – Each bug surfaces, gets reported back, and gets fixed in the same chat that introduced it

Connect Claude as Your AI Media Buyer

The closing call to action of the session is the practical step every advertiser should take this week. Apply for Google Ads standard API access today because approval takes one to two weeks. Get the Facebook Ads API key approved next, even if the process is convoluted. Then expose them through an MCP server and Claude becomes your media buyer.

Want the Full Walkthrough?

This post covers the highlights, but the full session includes the live first-campaign walkthrough with bid strategy and asset extension setup, the actual ad management SaaS demo with screen sharing, the rate-limit configuration for Facebook API, and detailed Q&A on YouTube Demand Gen targeting. Our weekly mastermind sessions are available to members with full video recordings, and we cover new campaigns, strategies, and optimizations every week.

Join the Vibe Coding & Marketing Mastermind

Join the Vibe Coding & Marketing Mastermind

Weekly strategies, updates, and vibe coding resources.

Leonidas
FORMULATED BY

Leonidas

I build websites, plugins, automations, and entire product lines using nothing but AI and natural language. Google Ads, Microsoft Ads, Facebook Ads, Claude, ChatGPT, Gemini. I run the Weekly Mastermind where I show exactly how to build a real business with vibe coding.

🕐 Vibe Mastermind in 00d 00h 00m 00s
Call LIVE — Join Now