A B2B paid media manager spent 12 years asking vendors to build this. Eventually he stopped asking.

Why this exists

After 12 years of running B2B Google Ads at scale, every software demo ended the same way.

Me: "We have a multi-brand, multi-MCC structure with custom offline conversion tracking and complex seasonal pacing. How does this handle that?"

Vendor: "Um. We have templates for that. But honestly, you're kind of an outlier."

This happened with Optmyzr. With Adroll. With Marin. With the smaller PPC management tools too. All of them had the same problem: built for the average use case, not for the edge case that's actually where the money is in B2B.

So I stopped waiting for them to build what I needed.

Who built it

Alex Langton. Senior B2B paid media manager. Currently runs ~$650K/month in industrial and manufacturing Google Ads accounts. 12+ years of category-specific experience — laser markers, fluid power, industrial controls, MedTech. The kind of accounts where one closed deal pays for the year of CPL nobody else wants to bid on.

The user of this suite is, deliberately, someone like me. A practitioner running real money in B2B accounts who needs tools that respect the complexity of the work. Not a marketer running a D2C funnel. Not an in-house team running brand. Not an agency running 50 accounts identically.

The three rules

Every tool in the suite follows these three rules. They are not negotiable.

1. One thing, well.

No platforms. No "all-in-one" tools. Each extension does one specific job for B2B paid media managers and exits. If a tool starts trying to do two things, it splits into two extensions. The total surface area of the suite is large; the surface area of any single tool is tiny.

2. Minimum viable permissions.

Most extensions only need storage. None of them ask to read every website you visit. The few that need page access (Scout, PageTag, Loupe) only inject content scripts on specific domains, on user click. Permissions in Chrome are a trust budget; I refuse to spend more than necessary.

3. Honest output.

Numbers come with units. Unrealistic projections get flagged as unrealistic. AI confidence scores are surfaced, not hidden. When data is partial, the tool says so. The single fastest way to lose a senior practitioner's trust is to ship a number with no context, and the second-fastest is to ship a confident number that's wrong.

How it's built

Vanilla ES6+ JavaScript. No npm, no build step, no React. Manifest V3, classic <script> tags, brand tokens as CSS custom properties. Comments matter — every exported function gets a block comment explaining what and why. The whole codebase reads like documentation because the user (me, mostly) is a vibe coder with deep domain expertise but limited engineering background.

The full conventions document lives at langton-master-handoff.md in the repo if you want the canonical reference.

How to reach me

Best place to contact me: file an issue on GitHub. Feature requests, tool ideas, bugs, "your math is wrong" — all welcome. The build queue is public; the roadmap is what's actually going to get built next.

If you're a B2B paid media practitioner and want to compare notes on running industrial accounts: also welcome.