The Limits of Historical Bidding
Smart Bidding works by learning from historical conversion data. It sees that certain audiences, at certain times, on certain devices, tend to convert. It bids accordingly.
The limitation: historical data reflects past conditions. When conditions change, the algorithm is slow to adapt. It needs new conversions to learn new patterns.
By the time Smart Bidding sees that demand has shifted, the shift has already happened. You've either missed the opportunity or overspent into declining demand.
Smart Bidding typically takes 2-4 weeks to fully adjust to significant demand changes. For seasonal products or trend-driven categories, that's too slow.
What Is Demand Sensing?
Demand sensing uses leading indicators to predict demand before it materialises in conversion data. Instead of waiting for sales to tell you what happened, you use signals that predict what will happen.
External Signals
Weather forecasts, social trends, economic indicators, competitor actions. These external factors influence demand before conversion data reflects it.
Internal Signals
Search query trends, site search volume, email engagement, wishlist activity. Your own data shows intent before purchase.
Operational Signals
Inventory positions, supplier lead times, cash flow requirements. Business context that should influence advertising decisions.
Predictive Signals
Different categories benefit from different predictive signals:
Signal by Category
Implementation Approaches
Demand sensing can be implemented at different levels of sophistication:
Level 1: Manual Adjustment
Monitor external signals manually. Adjust budgets and targets based on anticipated demand. Simple but effective for predictable seasonal patterns.
Level 2: Rules-Based Automation
Create automated rules triggered by external data. "If temperature forecast exceeds 25°C, increase garden category budget by 30%."
Level 3: Machine Learning
Build predictive models that incorporate multiple signals. The model learns which signals predict demand for which categories and adjusts automatically.
Start simple. Level 1 implementation often captures 70% of the value. Complex ML approaches need significant data and engineering investment.
Hybrid Strategy
Most accounts benefit from combining historical bidding with demand sensing:
Evergreen Products
Stable demand patterns. Historical bidding works well. Let Smart Bidding optimise based on conversion data.
Seasonal/Trend Products
Demand shifts faster than learning. Apply demand sensing signals via budget adjustments and target modifications.
New Products
No historical data exists. Demand sensing based on category trends and competitive signals is the only option.
The future is predictive. Brands that build demand sensing capabilities now will outperform those relying solely on historical optimisation.
Preparing for Predictive
Steps to build toward demand sensing capability:
1. Identify Your Leading Indicators
What predicts demand in your category? Weather, events, social trends, economic factors? List the signals that should matter.
2. Build Data Pipelines
Get predictive data into your decision-making flow. Weather APIs, trend monitoring tools, internal search data. Accessible data enables action.
3. Test Correlations
Validate that your signals actually predict demand. Retrospective analysis: did weather correlate with garden sales? Build evidence.
4. Start Simple
Manual budget adjustments based on obvious signals. Prove the concept before investing in automation.
Frequently Asked Questions
What is demand sensing in Google Ads?
Demand sensing uses real-time and predictive signals to anticipate demand before it appears in historical data. Weather forecasts, trending topics, inventory positions, and economic indicators can predict demand shifts faster than waiting for conversion data.
Can Smart Bidding do demand sensing automatically?
Smart Bidding uses historical conversion data and some real-time signals like device and location. But it doesn't incorporate external demand signals like weather, events, or inventory positions. Demand sensing requires feeding additional signals into your strategy.
When should I use historical bidding versus demand sensing?
Use historical bidding for stable, evergreen products with consistent demand patterns. Use demand sensing for seasonal products, weather-sensitive categories, and trend-driven inventory. Most accounts benefit from a hybrid approach.
Ready for Predictive Advertising?
We help brands identify predictive signals and build demand sensing into their Google Ads strategy.
Book a Strategy DiscussionGet our insights in your inbox
Plain-English thinking about Google Ads. No spam, unsubscribe anytime.