Show HN: I built my first MCP to manage Google Ads (adchestra.com)

2 points by MichalKrk ↗ HN
My co-founder and I tried running Google Ads for our Shopify store, but we could not get a positive ROAS (we spent 150USD on a single conversion).

As we also could not afford an agency, we turned to AI (IMO Codex >>> Claude Code).

We found that Google’s Ads MCP could only read data, but it could not make changes. So we built our own hosted MCP with read and write access.

It can inspect performance, find wasted spend, create or update campaigns, and work with Google Analytics and Tag Manager.

15 comments

[ 0.28 ms ] story [ 13.4 ms ] thread
I will definitely try this, how much experience with ads do i need to have to use it?
Interesting, can I make it do scheduled tasks, for example sending xampaign performance report every monday?
Does it automatically create campaigns + organize them (proper naming conventions). I am lazy when it comes to organizing my ads.
Yes, it can create and organize campaigns, but the naming conventions depend on the AI agent you use (from my experience most LLM models do a pretty good job already).

We haven’t added our own marketing best practices to the MCP yet. We might likely add them later based on user feedback.

This is really cool, I do such a terrible job managing my ad campaigns... too time consuming.
Haha exactly, would love to hear you thought if you get to try it out ;)
The write access is the interesting part here. How are you handling approval boundaries for campaign changes—does the agent propose changes first, or can it execute within predefined limits?
Great question! The agent asks you with every call to our MCP whether you would like to "Deny"/"Allow once"/"Allow always", that way you can keep read calls on auto-approve and review the write calls one by one to stay on the safe side :)
Interesting. Is it a commercial, hosted wrapper around the open-source AdLoop MCP server project?
Congrats on the launch. I'm working on something nearby. With AI everyone ships a site fast and the content is ready-made, so in the end they all look alike, and that shows up in signups. Rawuser reads where a visitor came from and why, and shows a page written for that, in your own design, untouched. If it's interesting, I'd like to show you how it looks on adchestra.com. Agshin