DIAGNOSE · Meta only
Find what to check first in your Meta ads.
Select the symptoms you see. Get likely causes, a prioritised action list and checks to make before you change your campaigns.
Built by Jonas Sluijs, former Meta & Snap growth lead.
What this is not. This is a rules-based checklist, not an audit of your account. It cannot confirm a cause or measure incrementality. Your selections determine the recommendations. How the diagnostic model works.
How this diagnostic works
You tell it how long the campaign has been running and which symptoms you are seeing. It maps those symptoms onto the causes that commonly produce them, orders the causes by how often they explain that combination, and gives you a check for each one. Optional economics let it put your own numbers into the arithmetic.
The ordering is a set of rules written from experience, not a model fitted to your account. Two accounts with identical symptoms get identical output. Treat the result as a list of things worth checking in a sensible order, and confirm the cause in Ads Manager before you change anything.
Common questions
Does this connect to my Meta ad account?
No. There is no login, no OAuth and no ad-account connection. Nothing is read from your account and nothing is written to it. Everything you see comes from the symptoms and optional numbers you enter on this page.
What if platform ROAS looks good but I am losing money?
Reported ROAS counts conversions Meta believes it caused, inside an attribution window you chose. It can include purchases that would have happened anyway. Compare it against your own blended numbers: total revenue over total spend, and new-customer CAC. The MER and nCAC calculator works those out, the break-even calculator tells you the most you can afford to pay, and the measurement guide covers why the two numbers disagree.
Where can I read the detailed checks?
Each cause links through to a full guide. You can also browse every Meta troubleshooting guide directly, or start with the common ones: a sudden CPA increase, learning limited, low CTR and attribution windows.