Last week, one of my clients decided to be “smart.”
He connected a large language model to the Google Ads API. Gave it access. Started asking performance questions.
At first? Interesting insights. Some solid pattern spotting.
Then… chaos.
Without him realizing it, the LLM didn’t just analyze. It “optimized.”
Yes. It AI-magically made changes inside the account. Bidding targets adjusted. Broad match keywords added… Stuff no human approved.
And the reasoning behind it? Simply dead wrong!
Luckily, my Change History Alert script fired.
The next day, I got an email in my inbox:
“Unrecognized user made changes in account X.”
That “user”? The LLM, on behalf of my client.
I reverted the nonsense in minutes before the crazy changes could do harm.
Here’s the thing: AI is a brilliant intern. It is NOT a senior media buyer. Just like an LLM (sometimes) hallucinates explanations, it can hallucinate optimizations too. You have to monitor its work like you monitor your interns.
And here’s the actionable takeaway: install my change history alert script that flags ANY edits from outside your approved user list.
Even if it’s “just” an API integrated LLM.
Go ahead, install it. It only takes 5 minutes.
Your future self will thank you.
LINK: https://nilsrooijmans.com/google-ads-script-change-history-alerts/
Happy scripting!
– Nils