Understanding AI Optimization Adjustments: A Detailed Explanation of AI Actions

Latest updated: 2024/7/4

 

In the article "AI summary", we provided an overview of AI's core optimization strategies and outcome metrics. This article will provide a detailed introduction into each AI optimization action.

 

Budget Dayparting

AI Optimization Logic: For ad campaigns with sufficient data and out of budget serving status, based on AMS hourly data, ad spend is controlled within time slots with higher conversion rates through budget dayparting (setting budgets by time periods).

Measurement Metrics: Whether the budget during high conversion periods is sufficient.

Value for Sellers: Assists in understanding whether AI's budget is sufficient during high conversion periods, reducing loss of order growth due to insufficient budget during peak traffic hours.

 

1. Data Filtering

• View time-based budget performance related to specified stores, ASINs, ad groups, and portfolios.

• Customize the data period.

 

2. Ad Campaign Search

Search & view a specific ad campaign.

 

3. Data Summary Card

• High Conversion Rate During High Conversion Window:

• Calculated by ad campaign, High Conversion Intervals are hours within a 24-hour day where ad campaign conversion rate is higher than the all-day average conversion rate.

• This metric measures the percentage of time in which high conversion periods did not encounter an over-budget situation.

• Campaigns with dayparting: Number of ad campaigns that have implemented time-based budget strategies.

• High Conversion Window: Time periods when an ad campaign's conversion rate is higher than the all-day average conversion rate.

 

4. Time-Based Performance Chart

The horizontal axis of the chart represents 24 hours of a day. The solid line represents the ad performance during the selected time period, and the dotted line represents the ad performance for the previous 14 days. High conversion periods are highlighted in yellow, and the percentage within the yellow box shows the proportion of budget coverage during that period.

Ideally, the higher the budget coverage during high conversion periods, the better. If there is a gradual decline, you can adjust your budget settings based on the specific performance of your ad campaigns.

Since time-based budgeting involves AMS hourly data, the earliest data in this chart can be traced back to April 8, 2024.

 

5. Ad Campaign List

This table lists the AI target, current data performance, and the target gap for each ad campaign:

• If there is a large gap between actual performance and the target, we recommend modifying the target first.

• If the actual performance is close to or exceeds the target, and there is insufficient budget coverage during high conversion periods, you can increase the budget of the ad campaign manually or set a higher budget within the Creation tab.

 

Bid Dayparting

AI Optimization Logic: Using AMS hourly data-led targeting analysis, bids adjust up or down at different times based on performance.

Value for Sellers: Analyze AI's optimization actions which use time-based bidding for different keywords, calibrate with operational experience, and improve impressions and number of orders.

 

1. Data Filtering

• View time-based bidding performance of specific stores, ASINs, ad groups, and portfolios.

• Customize the data period.

 

2. Keyword Search

Search for specific keywords to view the time-based bidding strategy of that keyword.

 

3. Data Summary Card

• CVR During High Conversion Window:

• Calculated by target, High Conversion Intervals are the time spans within a 24-hour day where the target conversion rate is higher than the all-day average conversion rate.

• This metric calculates the average conversion rate of each target in the high conversion interval.

• Campaigns with dayparting: Number of ad campaigns that have implemented time-based bidding strategies.

• High Conversion Window: Time spans when the target's conversion rate is higher than the all-day average conversion rate.

 

4. Time-Based Performance Chart

The horizontal axis of the chart represents a 24 hour day. The yellow line segments mark the hourly high conversion periods. The dotted green line is the baseline bidding coefficient for each target, which defaults to 1 (100%). The solid green line is the average bid coefficient for targets that have implemented time-based bidding strategies in each hour. If it is higher than the dotted green line, it means AI has increased the bid; if it is lower than the dotted green line, it means AI has decreased the bid. The solid blue line is the average ad performance of all targets, which can be switched between multiple viewing metrics.

Ideally, the higher the conversion rate during a period, the more likely AI will increase the bid, and the solid green line will be above the dotted green line; the lower the conversion rate during a period, the more likely AI will lower the bid, and the solid green line will be below the dotted green line.

Since there are many targets that perform time-based bidding, the bid coefficient in the chart is the result after averaging, thus, the changes in the bid coefficient are not immediately obvious. You can narrow the data range or search for specific targets to alter the view.

 

5. List of Targets Performing Bid Dayparting

 

Keyword Harvesting

AI Optimization Logic: If AI identifies high-potential keywords or ASINs in the search term report that have not yet been added to the campaign, it will consider adding them if budget allows.

Measurement Metrics: Sales brought by harvesting.

Value for Sellers: Helps users avoid the blind spots of manual operations, identify high-potential keywords that have not yet been used and adding them to existing ad campaigns when the ad campaign budget is sufficient. This opens up new, effective targets and improves both the precision and efficiency of ad delivery.

 

1. Data Filtering

- View keyword/ASIN harvesting related to specific stores, ASINs, ad groups, and portfolios.

- Customize the data period for AI to perform harvesting actions.

 

2. Data Summary Card

- Number of Harvested Keywords: The number of keywords harvested by AI within the selected time range. When checked, the page will only display the harvested keywords and their performance data.

- Number of Harvested ASINs: The number of ASINs harvested by AI within the selected time range. When checked, the page will only display the harvested ASINs and their performance data.

- Guarded Revenue: The sales generated by the above harvested keywords and ASINs within the selected time range, calculated separately for each harvested keyword/ASIN and summed up.

- Average Survival Tenure: The average number of days the above harvested keywords and ASINs were active within the selected time range.

 

3. Data Chart

The horizontal axis can switch between weeks and months to distinguish the keywords/ASINs harvested by AI in different timeframes. The vertical axis represents the performance data of these keywords/ASINs within the timeframe selected in the upper right corner.

Click on a specific week or month, and the keywords/ASINs harvested in that week or month will be displayed in the list below.

 

4. List of Harvested Keywords/ASINs

Click the trend button next to the keyword/ASIN to view daily performance data and the specific harvesting date based on when the target started generating impressions.

 

Keyword Negation

AI Optimization Logic: Based on historical search term performance, combined with TOPSIS, root word analysis, predictive models, and analysis of historical manual operations, AI calculates the contribution and cost of search terms, negating low-efficiency search terms and reducing unnecessary ad spend.

Measurement Metrics: Saved spend after negation

Value for Sellers: Users can avoid low-efficiency and invalid traffic for subsequent ad delivery and spend the budget on more efficient targets.

 

1. Data Filtering

- View keyword/ASIN negation related to specific stores, ASINs, ad groups, and portfolios.

- Customize the time range for AI to perform negation actions.

 The time range filter on this page targets the time when AI performs negation actions, not the time of the ad performance data displayed in the charts and tables.

 

2. Data Summary Card

- Total Savings: Spend generated by the keywords/ASINs negated by Xnurta AI in the 30 days prior to the AI's negation action, calculated separately for each negated keyword/ASIN and summed up.

- Number of Negated Keywords: Search terms with poor performance are precisely negated. When checked, the page will only display the precisely negated keywords and their performance data.

- Number of Negated Root Words: Search terms with poor performance have common root words and are phrase negated. When checked, the page will only display the phrase negated keywords and their performance data.

- Number of Negated ASINs: ASINs with poor performance are negated. When checked, the page will only display the negated ASINs and their performance data.

- Average Tenure: The number of days the negated keywords/ASINs were continuously delivered from the date they were added to the date they were negated by AI. Calculated for each keyword/ASIN and averaged.

 

3. Data Chart

The horizontal axis can switch between weeks and months to distinguish the keywords/ASINs negated by AI in different timeframes within the time period selected in the upper right corner. The vertical axis represents the spend and ad performance of these keywords/ASINs in the 7/30/60 days prior to the date they were negated by AI.

Click on a specific week or month to select the keywords/ASINs negated in that week or month, and they will be displayed in the list below.

 

4. List of Negated Keywords/ASINs

The list shows the performance data of each negated keyword/ASIN in the past 7/30/60 days (the specific time range depends on your selection in the data chart).

Click the trend button next to the keyword/ASIN to view daily performance data and the specific negation date based on when impressions stopped.

The past 7/30/60 days data in the table is a data snapshot recorded when AI negated that keyword/ASIN and is no longer updated; however, the data after clicking the trend button will continue to be updated.

 Due to the impact of Amazon's attribution period, there will be some changes in ad performance data, so there will be small differences between the data in the table and the trend chart.

 

Rapid Testing

AI Optimization Logic: For newly created targets, use a "small step, fast run" approach to adjust target bids, utilizing real-time AMS data to obtain clicks at minimal cost.

Value for Sellers: Users can use the "Quick Testing" action, combined with their own operational experience, to continuously optimize delivery strategies and improve overall ad delivery.

 

1. Data Filtering

- View quick test performance related to specific stores, ASINs, ad groups, and portfolios.

- Customize data period.

 

2. Keyword Search

Search for specific keywords to view the quick test situation for a single keyword.

 

3. Data Summary Card

- Average Time to First Click: The average time it takes for targets to obtain their first click through quick testing.

- Number of Rapid Tested Targets: The number of targets that underwent rapid testing within the selected date range. Newly added targets will all execute the rapid test strategy.

- Average First Click CPC: The average CPC of targets that obtained their first click through rapid testing.

- Incremental Impressions Gain: The number of ad impressions generated by newly added targets.

 

4. Rapid Test Completion Result Chart

The horizontal axis is the time period starting from when a new target is added, and the vertical axis is the number of targets that obtained their first click within that time range and the average CPC.

 

5. List of Rapid Testing Targets