AdAdvisor’s recommendation engine is the core of the platform. When you hit “Generate Recommendations,” a multi-agent AI pipeline analyzes your campaign data across multiple time windows, compares it against your business metrics, and produces specific, prioritized actions you can take to improve performance.
What triggers recommendation generation?
Recommendations are generated on-demand. You click the Generate Recommendations button on either the dashboard or the Ads Manager page. The AI runs its analysis and delivers results within a couple of minutes.
You can generate recommendations once every 24 hours. The button shows when your next generation will be available.
What data feeds the AI?
The recommendation engine looks at a lot of data before making suggestions:
- Daily metrics across multiple time windows: 3 days, 5 days, 7 days, 15 days, and 30 days
- Statistical analysis for each window: averages, medians, standard deviations, trends, volatility
- Metrics include: ROAS, CPA, CPC, CPM, AOV, spend, revenue, impressions, frequency, CTR, and conversions
Your business context
Historical recommendations
The AI checks what it recommended before, how you responded, and what happened after. If you accepted a recommendation, the AI looks at how your performance changed as a result. That feedback loop helps it make better suggestions over time. If you rejected something, it factors that in too. And it won’t suggest the same thing again while it’s still waiting for results from a previous recommendation.
Types of recommendations
Here’s an example of what a recommendation card looks like in AdAdvisor:
The AI can suggest many different actions, grouped into categories:
Budget actions
| Action | What it means |
|---|
| Scale Up | Increase daily budget for a high-performing entity |
| Scale Down | Decrease budget for an underperformer |
| Pause | Temporarily pause an entity |
| Shutdown | Pause a campaign permanently |
Creative and audience actions
| Action | What it means |
|---|
| Refresh Creatives | Replace or update ad creatives showing fatigue |
| Fix Audience Fatigue | Adjust targeting to combat saturation |
| Fix Funnel | Optimize the conversion funnel |
| Fix AOV | Address declining average order value |
Structural actions
| Action | What it means |
|---|
| Duplicate Campaign | Clone a top performer into a new campaign |
| Duplicate Ad Set | Clone a top-performing ad set |
| Create Campaign | Build an entirely new campaign |
| Create Ad Set | Add a new ad set to an existing campaign |
| Create Creatives | Generate new creative assets |
| Clean Up | Remove underperforming ad sets and ads |
Monitoring
| Action | What it means |
|---|
| Maintain | Keep doing what you’re doing. The AI is monitoring but no changes needed. |
Priority levels
Every recommendation is assigned a priority:
| Priority | Meaning |
|---|
| Critical | Act now. Performance is suffering or money is being wasted. |
| High | Act soon. There’s a clear opportunity or a growing problem. |
| Medium | Worth doing when you have time. An optimization that will improve results. |
| Low | Nice to have. Minor improvement or something to keep an eye on. |
What’s next?
