Stack · Published 2026-05-06 · 3 min read
A fast ad account audit: the CPA versus spend chart
I have been following Curtis Holland for a long time. Curtis invents new ways of visualising ad account data. Highly recommended.
One of his charts plots CPA against spend by creative. The task: understand what is happening in an ad account in thirty seconds and find the problems.
How it works. The X axis is total spend per creative. The Y axis is CPA.
Each dot is a creative.
Two lines divide the chart into four quadrants
- the horizontal line is your target CPA (this is the user's LTV),
- the vertical line is the spend threshold at which a creative finishes learning.
The four quadrants. Testing or killing — high CPA, low spend. Creatives with a high CPA and little spend. Plenty of dots in this area means you have a healthy process for testing new creatives.
Cut — high CPA, high spend. Creatives that missed the KPI but still burned budget. If there are many dots here, the account has a systemic problem: the buyer has given up on the project.
Scale — low CPA, low spend. Effective creatives that were never given budget. They work, but they have not been scaled. If there are many dots here, you are underusing creatives that already work.
Winners — low CPA, high spend. Winning creatives: CPA below target, spend high. If you see a lot of dots in this square, you have a confident pipeline for producing and scaling working creatives. If there are few, you have most likely already run into weekly CPA degradation.
How to choose the thresholds
- The horizontal line (target CPA) is usually known to the team. It is the user's LTV or the target acquisition cost.
- The vertical line (spend threshold) is something people often do not think about, although it matters. At what level of spend do you decide to switch a creative off? The topic deserves its own post, which I will publish tomorrow.
Why you need this
One quick glance at the chart shows
- how much budget is draining into ineffective creatives,
- whether there are working creatives that have not been scaled,
- how efficiently the testing process is running.
This chart can be animated in Claude Code — then you see a creative's path from test to scaling or shutdown. This is where neural networks show their full worth: we get a representation of the data in a form nobody had thought of before.
In Module 10, AI in User Acquisition, of the UA Manager 2026 course, we go through how to generate this and other charts for auditing an account.
The UA Manager 2026 course: Meta and Google Ads at full strength. Starts 15 May, finishes 15 July. Registration and details at the links.
Or just write to me: @SiberianPaaanda.