Best practices
Search campaigns using Target ROAS will have the option to enable Smart Bidding Exploration in campaign settings or portfolio bid strategy settings. Keep the following in mind:
- Make sure you’re currently leveraging Target ROAS bidding with broad targeting (this could be non-brand broad match keywords, DSA or AI Max) and want to scale performance further by exploring new pockets of performance opportunity.
- You should be willing to accept a decrease in ROAS performance during the exploration period. Note: You will be able to indicate your tolerance for lowered ROAS performance using a custom slider. A note in the UI will show you what your new “effective” ROAS target is with Smart Bidding Exploration enabled.
- Smart Bidding Exploration will not be effective if campaigns are limited by budget. If a campaign is limited by budget when enabling Smart Bidding Exploration, an alert will appear.
- Enable Smart Bidding Exploration with your existing ROAS targets. It’s not suggested to increase ROAS target to offset Exploration, because this will limit the ability to drive increased traffic diversity. Run a campaign experiment to measure the incremental impact of Bidding Exploration.
- Other campaign changes: As with Smart Bidding more generally, making noteworthy changes (For example, changes to measurement setup, targeting, creatives) may lead to performance fluctuations. If you make these kinds of changes while using Smart Bidding Exploration, be sure to give the system time to adjust (approximately 2-3 conversion cycles) before assessing performance again.
- Avoid using CPC bid limits. For exploration, bid limits are not preferred as they restrict the ability to go after different search terms. Give the system time to recalibrate post removal of bid limits (over 2 conversion cycles or 2 weeks) & then opt into Smart Bidding Exploration.
- To test Smart Bidding Exploration, use a campaign experiment to isolate results from other factors like seasonality.
- Avoid testing Smart Bidding Exploration in brand new campaigns, instead test in a well-established campaign for optimal performance. If you do test in a new campaign, wait for the initial ramp-up period, which could be 2-3 conversion cycles or 7 days, before evaluating performance.
Experiment Best Practices
Running an experiment is recommended in order to isolate the test from other factors like changes in seasonal demand.
- Turn on Experiment sync. Keep all variables between the trial and control arm constant. Enabling Smart Bidding Exploration within campaign or portfolio bid strategy settings should be the only variable that is different between the base and trial arms.
- Don’t attempt to test other variables concurrently.
- Prior to enabling Smart Bidding Exploration in the trial arm, wait 2-3 conversion cycles or 7 days (whichever comes first) to ensure that base and trial arms have comparable spend and performance. Exclude that time period from performance evaluation.
- Next, enable Exploration in the trial phase and test for 6+ weeks, noting that it might take 1-2 additional weeks for Exploration to fully ramp.
- Don’t make major adjustments after you have opted in. Don’t adjust the Target ROAS.
- Enable Bidding Exploration in the trial arm.
- Note: To compare Traffic Diversity metrics, compare the respective Bid Strategy reports for the control and trial arm.
- It isn’t recommended to test Smart Bidding Exploration in a new campaign. For optimal performance, test in a well-established campaign
