¶ Best Practices to Set up Optimize
Best Practices to Set up Optimize
This document covers the best ways to configure AI Groups to leverage Pixis’s Optimize feature.
You can configure Optimize settings to generate and execute Bid, Budget, and Ad Creative Rotation (Meta only) recommendations for your campaigns/ad sets in the AI Group. It generates recommendations by identifying the budget constraints and target goals that you have set to achieve Cost, Result, and Revenue metrics.
Let’s understand the best practices to follow on each step of configuration.
-
Objectives
-
Optimization Goals
-
Spend Goals
-
Action Levers
¶ Step 1: Objectives
In objectives, you have to pick your source of truth and enter the key metric for which you want to optimise the AI Group. You can choose the metrics from your existing campaign data by selecting your source of truth, which can be either a third-party platform or an Ad platform.
The following are the two key metrics you give:
-
Result Metric: Here, you can pick a single KPI that you want to achieve. This can be Clicks, Conversions, Purchases, etc.
-
Your Result Metric Name should indicate your optimization KPI, for example, leads, purchases, clicks, etc.
-
-
Revenue Metric (Optional): If your results metric is Website Purchases, you can pick a revenue metric that shows the revenue you received from those purchases.
-
Add a revenue metric that clearly shows the revenue generated from the result metric. This metric will be related to sales.
-
¶ Step 2: Optimization Goals
Here, you can set the target metric value to which the AI group should be optimized.
These are the two important metrics:
¶ Target Cost
When setting up target cost, you have the option to input your target Cost per Result (CPR) or Return on Advertising Spend (ROAS). For example, if the current cost per result for a purchase is $155, you can specify that you want to aim for a lower cost, such as $140 per result. This helps the system optimize your campaigns to reach the desired cost or return on ad spend.
-
We look for the daily Target results, while Target CPR is always the average value of the ideal CPR that you would want to achieve in the daily time frame.
¶ Target Scale
Define your total target results to be achieved per day daily scaling targets, i.e., get 92 daily website purchases.
-
Always revisit your targets regularly to ensure they are adjusted according to the performance updates in your AI group. The ideal target should be 15-20% higher (in case of scaling up) than the actual result count.
-
When setting your targets, always analyze historical data to understand daily targets.
¶ Optimization Preference
AI needs to understand your priorities in terms of whether it needs to focus more on optimizing for scale or optimizing for cost i.e., achieving CPR of 140$ and getting daily 92 website purchases.
There are currently two available approaches:
-
Autonomous
-
Custom
¶ Autonomous
If you're uncertain about the most effective strategy for your specific use case, entrust it to Pixis AI. Pixis AI meticulously analyses a multitude of variables, including historical performance, target objectives, and the potential impact of recommended actions, to determine the optimal course for your AI Group. Given that each AI group may necessitate a unique approach for scaling, the AI-driven method stands out as the best practice for achieving optimal results.
Depending on the brand strategy and account health, you should choose the method that fulfills their requirements in the best possible manner.
¶ Manually
You can decide how much weight you want to give to the two available approaches using the slider. Depending on what your objective is, try to give more weight to the one that aligns better. Available approaches:
-
Cost First: This will prioritise achieving your desired CPR over scaling performance and make recommendations accordingly.
-
Scale First: This will shift the focus to achieving daily KPIs to improve campaign performance, rather than solely optimising cost.
-
Balanced: You can also try out a balanced approach just in case it’s too early for you to decide what matters more, cost or scale.
¶ Step 3: Spend Goals
Here, you can create a budget plan for daily or periodic pacing. This allows the system to utilise your budget across AI Group, either daily or over a specific period. You can also manually distribute your budget across different channels and set constraints on minimum and maximum bid and budget spending.
¶ Period
In Period, you can choose to pick either a daily or a custom date range.
-
When your AI Group has new campaigns, it's best to choose daily pacing since there's no historical data to rely on. However, as the campaign gets more data over 30 to 60 days, you can set a custom date range to optimize budgets based on the longer term.
Under Budget plan, you have two options:
-
Set a fixed budget
-
Set a budget range
-
Let Pixis use my campaign’s budget as set on ads network
If you select “Let Pixis use my campaign’s budget as set on ads network”
Pixis will incorporate the budget setting from your ad platforms and will use a deviation percentage parameter.
¶ Set a fixed budget
You can add a spend budget for daily and periodic pacing. It ensures that the budget you set is allocated across your AI Group. Your goals might be under/overachieved.
Action Constraints
When you select “set a fixed budget”, you will get the option to Add Constraints. Here, you can set bid & budget, Minimum and maximum values for both bid & budget that you would want each channel/platform to adhere to.
-
You can prioritise the channel for which you want to spend the maximum or minimum value.
For example, if you're spending 100 dollars a week across 5 channels, you should distribute the budget accordingly, ensuring that the total minimum budget for all channels does not exceed 100 dollars when inputting the budget values for each channel.
¶ Step 4: Action Levers
Using Action levers, you can choose to generate bid, budget, and creative rotation recommendations automatically to your ad platforms.
¶ Bid & Budget
You can configure the min & max % change day-on-day and frequency of recommendations. You can follow these best practices while configuring:
-
Min & Max % change: We recommend you look for the suggested frequencies over your ad platforms and understand how frequently you would want to play with your bid/budget values. Always start experimenting with the platform-suggested frequency and then try seeing the impact of other frequencies as well to know which one works best for you.
-
Always start with small values like 1-2% minimum and 4-5% maximum, and once recommendations seem to be optimising the performance consistently, then go with an aggressive mindset if required.
-
-
Frequency: Set when to apply the changes.
-
We recommend you look for the suggested frequencies over your ad platforms and understand how frequently you would want to play with your bid/budget values. Always start experimenting with the platform-suggested frequency and then try seeing the impact of other frequencies as well to know which one works best for you.
-
¶ Creative Rotation
This lever applies AI recommendations for ad creatives. It scales up well performing ad creatives and pauses ad creatives that are performing poorly.
This action is only available for Meta.
You can configure the following:
-
Lookback Period: The time period for which you would like to analyse the performance data of the ad creatives.
-
Pixis AI thoroughly evaluates the campaign performance of the selected period, taking into account various factors such as seasonality and metric-level performance. This analysis guides decision-making regarding the scheduling of creatives, determining the most effective timing for optimal campaign performance throughout the day or week.
-
-
Max test Ads: The total number of test ads (new ads) that you require to be active at any given point. New Ads are the ads that have less than six days of data in the last 30 days.
-
Pixis AI will gradually move towards the maximum number of ads you want to be active, which ultimately results in better outcomes than the instant scaling approach.
-
-
Max Optimized Ads: The total number of mature ads (ads with enough data, i.e., major ads) that you want to be active at any given point.
-
Ideally, you should consider the number of existing ads present in your AI Group and take a fraction of it as per the risk strategy for performance optimization.
-
-
Frequency: Select how many times in a week you want to apply the changes.
-
Always start experimenting with the platform-suggested frequency and then try seeing the impact of other frequencies as well to know which one works best for you.
-