The ad spend report that writes itself
a specific n8n and Claude workflow for daily spend reporting across paid channels, and why it exists to buy back hours, not to replace the decisions.
managing $300k a month in ad spend across channels means someone is checking numbers every single day, and for most of a growth operation’s history that someone is a person clicking through three different ad platform dashboards before they’ve had coffee. this is the workflow that replaced that specific hour, not the strategy behind it.
What the workflow actually does
once a day, it pulls yesterday’s spend and results from each platform, checks the numbers against the pass or fail thresholds set before the test started, and flags anything that crossed a threshold in either direction. a Claude call reads the pulled numbers and drafts a plain english summary, three or four sentences, the same length as what I’d type by hand if I were doing it manually at 7am. that summary lands wherever I actually read updates, not buried in a dashboard that only gets opened when something already feels wrong.
Why a summary and not a dashboard
a dashboard answers questions you already know to ask. a short written summary tells you the thing you didn’t know to look for, because it’s built to say what changed, not to display everything and let you go find it. most growth operations already have enough dashboards. what they’re missing is someone reading them daily and saying what actually matters in one sentence, which is a task a workflow can do at 7am without needing to have opinions about anything except the numbers in front of it.
Where the decision still happens
the workflow flags. it does not decide. crossing a pass or fail threshold means a real look at the creative and the audience before anything gets killed or scaled, because a threshold catches the symptom, not the reason. the 150+ creatives tested monthly only work as a system because every kill decision gets a human look first. automate the flag, not the call.
What this bought back
roughly an hour a day that used to go to manually checking three dashboards before anything else happened, across every active account. that hour moves to the part that actually needs a person: reading the flagged results and deciding what to do about them. it’s a smaller change than it sounds, and it’s most of what “AI-run growth ops” actually means in practice, once you strip the marketing language off it.
the broader pattern, including where I stop automating entirely, is in the n8n + Claude growth automations playbook.