Intentwise's Bidding Algorithm
Intentwise's bidding algorithm is designed to help you optimize your campaigns by reducing the gap between your target and observed ACOS (advertising cost of sales). Let's explore how it works and the strategies it employs.
Understanding Your Target ACOS
Target ACOS is the goal you set for your campaigns to match your actual ACOS. Intentwise's bidding algorithm considers factors like past keyword performance, keyword volume, and other data points to generate bid recommendations, helping you improve your campaign performance.
Learn how to set Target ACOS and apply bid recommendations.
How It Works
Machine Learning Models: Each account is assigned one of twenty machine learning models based on attributes like cost per click and revenue per click.
Dynamic Look-Back Window: Keywords/targets are assigned a dynamic look-back window, considering up to four weeks of historical data.
Predicted ROAS: Predicted return on ad spend (ROAS) is determined for various bid values around the current bid.
Bid Change Caps: Bid change caps are set based on the bid optimization profile (Conservative, Neutral, Aggressive) configured for each campaign. Learn more.
Strategies
No Click, No Impression Strategy: Targets with zero clicks or impressions are regularly bid up to collect performance data.
Target Re-cycle: Targets that previously met goals but have seen a drop in spend and bid value in the last 90 days are given another opportunity
Operational Overrides
Manual Bid Updates: If you manually update bids, the algorithm won't recommend changes for the next three days.
Campaign Level Performance: The algorithm prioritizes campaign-level performance over keyword/target level performance.
Bid Downs Before Sales Events: Bid decreases are minimized three days before a major sales event.
Best Practices
Algorithm vs. Rules: Avoid using algorithmic bidding and automated bid change rules on the same campaigns, as they operate independently.
Bid Strategy: Avoid using “bid up and down” as a bid strategy in conjunction with algorithmic bidding.
