📗 This is the user manual for the Suite for ChatGPT Ads, the tool in the Ninja Scripts panel. Here you can search it, listen to it, highlight it and save it to favourites. It is updated with every release of the tool.
A lookout that checks the account after every read and opens — or closes — alerts about what's worth changing. Each recommendation says what's happening, with which figures and what to do, and most can be applied with a button.
First things first: the target cost per conversion
It's a field on this screen and it's the single thing that unlocks most of the tool. Without it:
- there are no bidding recommendations (nobody can say whether a bid is high or low without knowing what «right» is),
- the budget isn't filtered by profitability,
- and the 🩺 Audit cannot judge money: its score stays capped.
It's your business number: what you're willing to pay for a conversion. Saving it recalculates the recommendations straight away.
The rules, and what each one requires
None of them opens without enough data. The minimums aren't decorative caution: with less, the recommendation would be a coin toss.
| Recommendation | When it appears |
|---|---|
| Switch to maximise conversions | ≥ 30 conversions in 30 days and ≥ 12 in 14, in clicks or fixed-bid groups of campaigns that measure conversions |
| Go back to clicks or check your measurement | Maximising conversions with fewer than 10 in 30 days and ≥ 100 in spend |
| Move the fixed bid | ≥ 5 conversions in 14 days. Down if cost exceeds your target by 20 %; up if it's 20 % below with ≥ 50 clicks |
| The budget is capping you | At the cap on ≥ 5 of the last 7 days, with ≥ 3 conversions and meeting your target |
| The budget is more than enough | Average spend below 50 % of the daily cap and no delivery issues |
| Hours that don't pay | Hours with ≥ 3 % of spend and zero conversions that together add up to ≥ 20 %, with ≥ 10 conversions in 14 days |
| Expensive segment | A platform, device or country with ≥ 20 % of spend and no conversions (or ≥ 15 % at 2.5× the account's cost) |
| Lab stalled | An active ad group with ≥ 2,000 impressions in 30 days and no live or recent test |
| Low relevance | An active ad group with ≥ 500 impressions and an index below 45 |
| Quality event | Lead Rating has already sent ≥ 30 qualified leads in 30 days and there's a campaign measuring conversions |
How it decides on clean data
- Anomalous days don't count: a day that spent less than 20 % of the median or more than three times it (a half pause, an odd spike) is dropped before the maths.
- Complete days only: today is out.
- Medians are sorted. Sounds obvious; it was a real bug in another Ninja product and was fixed in both places.
Dismissing and restoring
A recommendation you don't care about is dismissed and leaves the list. It isn't lost: it can be restored. And if the situation resolves itself, the recommendation closes itself — there's no list of stale alerts to tick off.
What it doesn't do
There is no automatic mode. No recommendation applies itself, not overnight, not «when it's obvious». You press the button.