Smart Assistant Function Description

I、Function overview

Based on the data of the current segment waterfall configuration, Smart Assistant automatically determines whether the corresponding ad source needs to be turned on/off, and whether it is necessary to increase bidding or conventional ad source.

Example:

1. Add or Remove Ad Sources

2. Dynamic Waterfall Requests

After Smart Assistant is enabled, different waterfall configurations will be requested based on device-level information. Smart Assistant supports dynamically adjusting the waterfall configuration for each request based on the value of each device.

This can reduce invalid ad requests and improve ad impression efficiency.

 

Ⅱ、Feature Usage

1. Feature Whitelisting

Please contact your AM to have the feature whitelisted before you can access the corresponding feature entry.

If the Smart Assistant entry appears in the left navigation bar, or if the Smart Assistant entry and related pages are visible on the Mediation page, the feature has been successfully whitelisted.

 

2. Recommended Placements for  Smart Assistant

(1) Entry

After the feature is whitelisted, you can view the notification for eligible placements on the Mediation page.

The recommendation will only be displayed after the placement reaches a certain level of impressions. If the number of impressions is too low, accurate optimization cannot be performed. If you do not see the recommended entry but need to enable Smart Assistant, please follow the process described in 3. Manage Smart Assistant Full-Scale Switches.

 

(2) Enable Smart Assistant

When the Mediation page displays the notification: “The current placement is recommended to use Smart Assistant for managed optimization to increase monetization revenue.” Click Enable Smart   to open the enablement dialog. After clicking it, you can choose to enable an A/B test or enable it for all traffic.

After clicking it, you can choose to enable an A/B test or enable it for all traffic.

① Placement A/B Test

If you select A/B Test, a Smart Assistant experiment will be created by default. You can then select the segments to be included in the experiment. It is recommended to select all segments. If you only need to run the experiment for specific segments, select them as needed. You can also adjust the segment allocation ratio.

After confirming the configuration, click Switch On to start the experiment.

After the experiment is successfully created, you can click More Actions to view summary data, adjust the allocation ratio, or edit the A/B test.

 

② Enable for All Traffic

After selecting Fully Enable, you can directly select the segments where Smart Assistant should be enabled. Smart Assistant will then be enabled for the selected segments.

3. Manage Smart Assistant Full-Scale Switches

If your placement does not have a recommendation to use Smart Assistant and you do not need to run an A/B test, you can manage which segments under the current placement should have Smart Assistant fully enabled through the Smart Assistant entry on the Mediation page.

Smart Assistant can also be used without an A/B test. This means that Smart Assistant can be fully enabled for a single segment or for all segments under a placement.

(1) Configuration Entry

From the Mediation page, select the placement where you want to enable Smart Assistant and click the Smart Assistant configuration entry. You can also disable it if it is no longer needed.

(2) Batch Operations for Segments

After entering the configuration page, you can batch enable or disable Smart Assistant for multiple segments.

(3) Configuration Completion

After Smart Assistant is enabled, the system will use the ad source information already enabled for the current placement to automatically create third-party ad sources.

If you need to adjust the ad source configuration, modify it on the Smart Assistant Settings page, as described in 4. Configure Smart Assistant below. 

Note: If there are Networks with missing or incorrect configurations, the Smart Assistant icon will be displayed in orange. The corresponding ad sources will only be automatically created after the configuration has been confirmed to be correct.

 

4. Configure Smart Assistant

(1) Configuration Entry

After Smart Assistant is enabled, you can manage the Networks and detailed configuration information for which Smart Assistant performs optimization for the current placement.

Go to Mediation → Smart Assistant → Smart Assistant Settings to view the automatic optimization strategy for the current placement and configure the Networks for which ad sources should be automatically recommended and created.

(2) Automatic Optimization Strategy

We will optimize the corresponding segments based on your settings on the automatic optimization page.

① Default Automatic Enable/Disable of Ad Sources

② Ad Source Price Range Limits

Supports setting limits for the floor prices of waterfall ad sources generated by Smart Assistant.

③ Custom Ad Source Naming Rules

If Auto Create Ad Source is enabled, you can use custom naming rules to name the Taku ad sources and ad placements of third-party ad platforms that are created.

Some ad platforms have restrictions on the maximum length of placement names. It is not recommended to add more than three fixed combination words.

④ Reuse Ad Sources Across A/B Test Segments

If Yes is selected, when Smart Assistant determines that an ad source with price C needs to be added, and another test segment already has an ad source with price C, the existing ad source will be reused directly. If you need to strictly distinguish the third-party API data of different test segments, select No.

⑤ Adjust Advanced Features (Whitelisting Required)

If Yes is selected, Smart Assistant will manage the Setting content of all segments under the current placement where Smart Assistant is enabled and dynamically adjust the configuration based on the data.

If you want to configure Setting manually, select No.

Setting will affect your revenue, so please select this option carefully.

⑥ Data Performance Preference

Different strategy settings will be applied based on your selected performance preference.

By default, the strategy balances eCPM and impressions. It will adjust impressions and eCPM based on the current status of your placement to maximize revenue.

(3) Ad Source Recommendation Configuration

① View and Select Networks

On the Ad Source Recommendation Configuration page, you can view the Networks currently supported for configuration. You can select the Networks that should participate in ad source recommendations based on your actual monetization requirements.

② Enable Networks and Complete Configuration

Only Networks with the switch enabled will participate in ad source recommendations. After enabling a Net

work, continue to complete the basic information and ad source configuration information for that Network. Smart Assistant can recommend ad sources based on actual data only after the relevant configuration has been completed correctly.

③ Diagnose Configuration Issues

A diagnostic feature is provided on the left side of the page to quickly check the configuration status of each Network.

If a Network has missing or incorrect configuration, you can quickly locate the configuration items that need to be completed through the diagnostic results, ensuring that the ad source recommendation feature works properly.

We recommend enabling Auto Create Ad Source. You can select the Networks for which you want Smart Assistant to recommend ad sources. After selecting a Network, if the Network supports Auto Create Ad Source and you have completed the relevant ad source configuration, Smart Assistant will automatically create new third-party ad sources based on the actual data. If a Network does not currently support Auto Create Ad Source, Smart Assistant will recommend ad sources that need to be added manually. You can view them later under Manually Add Recommended.

Note: After modifying a Network configuration, click Save Config to save the changes. 

Description: Automatic creation of ad sources requires the corresponding parameters to be configured and requires management API permissions for the Network. For details, refer to the corresponding Network's help documentation.

 

(4) Configuration Completed

After completing the configuration according to the process above, the Smart Assistant logo will be displayed in green.

If the logo is displayed in orange, it indicates that some configurations have not been completed. Please complete the configuration according to the process above.

 

5. View Manually Add Recommended

(1) Entry

After Smart Assistant is enabled, if there are ad sources recommended for manual addition, you can access the waterfall configuration through Mediation → Smart Assistant → Manually Add Recommended.

 

(2) Configuration Page

6. Other Features

(1) Disable Ad Sources

When the floor price configured through automatic creation does not take effect on the third-party platform, Smart Assistant will disable the ad source based on the relevant conditions.

This prevents low-eCPM ad sources from taking traffic when the configured eCPM does not take effect.

For example, when a Baidu ad source is newly configured, if the traffic volume is insufficient or Baidu cannot achieve the configured eCPM, the corresponding Baidu ad source will be disabled.

(2) Delete Ad Sources

When newly created ad sources by Smart Assistant generate relatively few impressions and little revenue, both the Taku ad source and the third-party ad source will be deleted.

For example, IDs that have been frozen by YLH or IDs that generate no revenue will be deleted to prevent the number of IDs created by the account from reaching the upper limit.

(3) Reuse Ad Source Configuration

If the current placement already has third-party ad sources created using Auto Create Ad Source, when Smart Assistant is enabled, it will use the configuration of the automatically created ad source ranked first in the current segment by default to create subsequent ad sources.

Bidding ad sources will be prioritized, followed by standard-priced ad sources.

(4) Automatic Optimization Configuration at the Segment Level

Please contact your AM to have the feature whitelisted before you can access the corresponding feature entry. After the feature is whitelisted, you can click to enter the automatic optimization configuration page for Segment A.

After adjusting the configuration, you can choose Save as Current Segment or Save and Apply. Subsequent optimization will follow the configuration logic accordingly. 

Specific Use Cases:

① Segments are configured based on region, and different regions have different Network requirements.

② Segments are configured based on user value, and specific ad source price ranges need to be configured.

③ Segments are configured based on channel, and different channels require different ad source naming rules.

7. Notes

🌟🌟 After enabling Smart Assistant, it is recommended that you do not manually adjust the configuration of the current segment or control group, as this may affect the comparison of A/B test data.

🌟🌟 Please note: If the Smart Assistant logo is displayed in orange, make sure to edit the configuration and click Save Config. Otherwise, automatic optimization will not take effect, and no new ad sources will be recommended.

 

Ⅲ、Smart Assistant  Module

A new Smart Assistant module has been added to the left navigation bar. It contains two sections: Management and A/B Test. The relevant features are as follows:

1. Management

(1) Enable/Disable Smart Assistant

The Management page displays placements where Smart Assistant has been enabled or is recommended. It supports comparing data by date to view changes in data performance, and also supports batch enabling/disabling Smart Assistant for the corresponding placements.

Supports selecting placements or segments and directly batch disabling/enabling Smart Assistant using Batch Disable Smart Assistant or Batch Enable Smart Assistant.

When enabling Smart Assistant, you can choose between A/B testing and enabling it for all traffic.

Note:

When enabling Smart Assistant for multiple placements in bulk, only full-scale enablement is supported. A/B testing for Smart Assistant is not supported in this case. Automatic Smart Assistant A/B testing is only supported when A/B testing is enabled for a single placement.

(2) Configuration Issues for Enabled Placements

A list of placements with configuration Exception will be displayed. Clicking a placement will quickly redirect you to the configuration page to complete the required configuration. 

Based on the issue identified on the configuration page, click To Configuration to quickly locate the configuration item that needs to be completed.

(3) Manual Waterfall  Recommendation

A consolidated list of ad sources that need to be added manually is displayed.

The recommended list is updated every day.

(4) Batch Management Entry

Batch-manage the Smart Assistant status of all Apps using the Batch Config Across Apps button.

2. Smart Assistant A/B Data

After selecting A/B testing when enabling Smart Assistant and starting the experiment, the experiment will be displayed in the A/B Test module.

You can quickly view the data of all placements with ongoing Smart Assistant experiments and edit the experiments.

 

Frequently Asked Questions

Q1. How can I verify the effectiveness of Smart Assistant?

You can use the A/B Test feature supported by Taku for verification. You can also observe the changes in revenue at the placement level before and after enabling Smart Assistant.

Note: A large number of developers have previously verified ARPU increases ranging from 5% to 60%. 

Q2. Which platforms support automatic ad source creation? 

Mainland China: Pangle, YLH, Kuaishou, Baidu, and Gromore. Overseas: Mintegral, Liftoff (Vungle), Pangle, and Bigo. 

Q3. If a placement already has multiple segments configured, can Smart Assistant be used?

Yes. However, the number of impressions at the segment level is preferably more than 5,000.

If a segment only has a few hundred daily impressions, enabling Smart Assistant is not recommended because the amount of data available for algorithmic learning is too small.