How our AI Recommendations works?

Recommendations are the AI generated insights that can be used to improve your campaigns. These insights are based on user behavior data for your campaigns and ad products. 

How are Recommendations generated?

Our machine learning algorithms have been trained on millions of rows of data. The algorithms understand the behavior patterns for your account and determine appropriate thresholds for you to take actions. The ability for our algorithms to function on just your accounts data ensures that the recommendations are personalized to you and not based on some generic benchmark which may not be appropriate for you. 

Our algorithms identify outliers for both - good and bad performance to bring the recommendations for you.

What if I don’t agree with the recommendations

We operate with the principle of “Human in the Loop” for building our algorithms. That means, you have full control over what to take the action on. For each recommendation, you get an option to ignore it.

Our algorithms learn your preferences each time you ignore a recommendation building a positive feedback loop.

---

Types of Outcomes for Recommendations

Recommendations can result in 2 types of broad outcomes - 

1. Actions to maintain an existing campaign

2. Actions to expand reach of existing campaigns / ad products

   
   

1. Maintain Existing campaigns

The actions here are aimed at improving existing campaigns with respect to Sales / ACOS / Impressions. These data backed actions here include:

• Adjusting target bids

• Removing targets

• Adjusting placement strategy

• Streamlining campaigns by removing non-performing ad products

• Identifying strategies for campaigns running out of budget often

Optimization	Backend strategies
Adjust Target Bids	Keyword Targets with limited impression share Search terms not included in targeting Product targets with low impression share Product targets from Auto Campaigns Keywords not added to all campaigns for a product Keywords not working well
Removing Targets	Search terms resulting in wasted budget in manual and auto campaigns Target Keywords with extremely high ACOS Target Products with extremely high ACOS Product - keyword combination resulting in wasted budget
Placement Strategy	Placement wise breakdown of performance Boosted placement positioning not resulting in good returns
Ad product performance	Ad products in specific campaign with extremely bad performance
Budget out campaigns	Campaigns where you should increase the budget Day parting strategy to reduce ACOS Day parting strategy to ensure more coverage for good campaigns throughout the day

2. Expand Ad reach

The actions here aim to expand the reach of good campaigns and ad products by creating additional budgets for a new campaign / ad group. These include:

• Identifying new targets

• Identifying new campaign opportunities

Expansion	Backend strategies
Identifying new targets	Search terms resulting in good outcomes Product keyword combinations that can be leveraged for additional sales Product targets that can be leveraged for additional sales
New campaign opportunities	Create new campaign / ad group for good products which are not getting enough budget Product combinations that can work for you

These recommendations can be segregated by 

• Campaigns

• Actions (Categories)

• Goals