Overview

Spend Optimization in Prescient AI empowers marketers to make data-driven budget allocation decisions, ensuring every dollar contributes to maximizing Return on Ad Spend (ROAS) or minimizing Customer Acquisition Cost (CAC). By leveraging scenario modeling and forecasting, users can evaluate potential budget adjustments, compare projected outcomes, and refine their strategy before making any changes.

Key Capabilities:

Scenario Planning & Forecasting: Simulate different budget allocations and predict their impact.
Campaign Selection & Budget Control: Choose which campaigns to optimize and define spend constraints.
Optimization Recommendations: AI-driven insights on how to allocate budgets for maximum impact.
Performance Visibility: Compare historical trends with projected outcomes.
Confidence Scoring: Assess the reliability of recommendations based on data density and historical performance..

Optimization Management

The Optimization page provides an overview of all created optimization scenarios, categorized by status for easy tracking:

Scenarios: Active optimizations that are saved and available for review.
Drafts: Every optimization is auto-saved as a draft to ensure scenario tests are not lost.
Archived: Optimizations that are no longer in use but remain accessible for reference.

Public vs. Private Optimizations

Optimizations can be set as Public (visible to the entire organization) or Private (visible only to the creator). By default, all optimizations are Public, but this can be adjusted in the optimization settings.

Creating an Optimization Scenario

To create a new optimization scenario:

Navigate to the Optimization page.
Click the + Create New Optimization
Configure your scenario settings, including target metric, campaigns, and budget range.

Scenario Setup

Scenario Target: Select the metric you want to optimize:
- Shopify: Revenue/ROAS or New Customers/CAC
- Amazon: Revenue/ROAS or New Customers/CAC

📘
Info
Amazon optimization is only available if you have an Amazon storefront connected.

Scenario Name & Description: Add a name and description to track the optimization's purpose.
Scenario Basis: Optimizations use a Fixed Budget approach, ensuring allocations fit within specified constraints.
Forecasting Timeframe: Choose 7, 14, or 28 days for projected performance.

Campaign Selection

Users can select which campaigns to optimize from a list of all modeled campaigns, organized by channel.

Campaigns can be selected at the channel level or individually.
Use the search bar to quickly find specific campaigns.
Apply Saved Views to filter campaigns based on predefined views from the Performance Page.
Lock specific campaigns at their current budget using the lock icon, ensuring they remain fixed during optimization.

Budget Selection

The final step is defining the budget for the optimization scenario.

By default, the previous period’s budget is used as a baseline (e.g., if optimizing for 14 days, the system pre-fills the last 14 days’ spend).
Adjust budgets manually or scale dynamically using the slider (minimum: $0, maximum: 200% of the baseline).
The manual input box allows custom spend allocation.

Optimization Outcomes & Projections

Once an optimization scenario is created, the Outcome screen provides a breakdown of expected performance.

Scenario Details Panel

Displays key scenario details, including name, description, and visibility settings.
Adjust the estimated budget and re-run the optimization with updated allocations.
Update the visibility of the optimization by setting it as public or private.

Performance Projections

The Projections section provides a comparison of three key spend scenarios:

No Change: Forecasted outcomes if spending patterns remain the same.
Previous Period: Outcomes if spend is identical to the prior period.
Optimal: Projected performance using the recommended optimized spend.

Metrics shown include:

Total Spend
Total Revenue or New Customers
Total ROAS or CAC
Expected Performance Impact (highlighted in green for positive impact, red for negative).

Confidence Score

Each optimization is assigned a Confidence Score, assessing the reliability of recommendations based on:

Data volume: The historical data available for the selected campaigns.
Variance: The range of potential outcomes (upper vs. lower bounds).
Spend consistency: The density of historical spend patterns at various levels.

Confidence Level	Score Range
Low Confidence	0 - 50%
Medium Confidence	51 - 69%
Medium-High Confidence	70 - 79%
High Confidence	80 - 100%

🚧
Warning
Confidence Scores reflect data reliability, not the probability of success. A 10% confidence score means Prescient has limited data to model predictions accurately, not necessarily that the forecast is unlikely.

Metric Rate

The Metric Rate option at the top of the Outcome Screen allows users to control how values are displayed for the optimization:

Total: Displays cumulative spend, revenue, and customer acquisition figures for the full selected timeframe.
Daily: Shows average daily values for each metric, providing a more granular view of performance trends.

Optimization Recommendations

Prescient provides spend recommendations at three levels:

Channel-Level: Suggested spend adjustments for entire marketing channels.
Tactic-Level: Insights based on user-defined tactic labels.
Campaign-Level: Specific spend recommendations for individual campaigns, categorized based on whether an increase or decrease in spend is recommended.

Each recommendation includes:

Projected Performance Impact: Expected changes in revenue/ROAS or new customers/CAC.
Saturation Curves: Visual representation of diminishing returns at different spend levels.

Accepting, Declining, & Implementation Date

Users can accept or decline optimization recommendations within the Campaigns View of the Outcome Screen.

Accepting a recommendation:

Users must specify an implementation date when accepting a recommendation.
Implementation Window: Accepted changes must be implemented within 7 days of the scenario date.
Accepted campaigns will appear in the Tracking tab for monitoring.

Declining a recommendation:

Campaigns do not need to be accepted or declined.
If a recommendation is neither accepted nor declined, it will simply not be tracked.

Customizing Recommendations

The optimizer is designed to create a mathematically optimal budget allocation based on Prescient’s attribution model and saturation curves. However, because it doesn’t account for campaign-specific context—such as its role within your marketing funnel, strategic goals, or upcoming creative adjustments—it may occasionally suggest optimizations that don’t align with your broader strategy.

To address this, users can modify recommendations after they have been generated using the Customize Recommendations button. This opens a window displaying:

All campaigns included in the optimization
The allocated spend after optimization, including the percentage of the total budget assigned to each campaign
A lock icon, allowing users to freeze spend levels for specific campaigns before re-optimizing

To manually adjust a budget, enter a specific dollar amount or budget percentage in the text box. The lock icon will engage automatically when a value is modified. Clicking on the Allocated $ box reveals:

Spend over the last 28 days
Forecasted "no change" spend
Optimal spend amount based on Prescient’s recommendation

Once all adjustments are made, scroll down and click "Save Changes." A confirmation window will appear, displaying the revised budget with two options:

Reallocate the difference across campaigns based on optimization logic.
Keep the manually entered budget as the final allocation.

Once a selection is made, the optimization will automatically rerun with your changes.

Columns Management & Exporting

Columns: Customize which columns are displayed in the Recommendations section using the Columns button at the top. This allows you to focus on the most relevant metrics for your optimization analysis.
Export: Optimization recommendations can be exported for external analysis, ingestion, or further customization using the Export Outcome Info button.

Tracking Optimization Performance

The Tracking tab allows users to monitor actual performance against forecasted outcomes.

Tracking Availability

Tracking is only available for campaigns with accepted recommendations. The button on the Optimization page will display "See Tracking" or "Tracking Complete" depending on if the scenario is still in progress.

Tracking Metrics & Data

When you click a tracking button on the Optimization page, you'll be taken to a new tab on the optimization outcome screen - Tracking. This tab displays:

Whether tracking is In Progress or Complete.
The start and end date of tracking.
An updated Projection table that shows actual "In Flight" totals since the optimization tracking started, showing:
- Actual MMM Revenue or MMM New Customers
- Actual Spend vs. Optimal Spend
- Performance deviations over the elapsed days

Creating an Optimization from Forecasting

Quick, single campaign optimizations can be created from the performance page via the Forecast tab within campaign drawers.

To initiate tracking, click Track within the forecast tab.
To activate tracking:
- Navigate to the Optimization Page
- Accept the recommendation
- Set an implementation date

Best Practices

Use the best practices below to maximize Prescient’s Optimizer and balance its recommendations with your real-world marketing constraints.

Step 1: Initial Setup for Holistic Budget Allocation

Define Goals & Metrics: Clarify your primary objective—ROAS or CAC, for example—and confirm whether you’re optimizing for Shopify vs. Amazon.
If You Have Multiple Objectives or Stores: In cases where you’re running optimizations for two eCommerce stores (e.g., Shopify + Amazon) or two objectives (e.g., CAC + ROAS), run a separate optimization for each goal. Compare recommendations; if there’s no overlap, prioritize your primary objective.
Understand Your Budget: Prescient projects current budgets by looking at spend up to the second-to-last day of the observed period for 7-, 14-, or 28-day scenarios. If your actual planned budget differs significantly (±25% or more), adjust the Estimated Budget slider. Note that large deviations may trigger an error message.

Step 2: Choosing Which Optimizations to Take

Look for High-Impact Changes: Identify where you get the largest potential performance lift for the least additional spend.
Confidence Score Is a Guide—Not a Verdict: Low confidence often reflects limited historical data or new campaigns. If data is sparse, consider gradual budget increments to test performance.
Don’t Feel Obligated to Accept All Recommendations: The Optimizer aims to find a mathematically efficient budget mix. You can accept or reject individual suggestions based on strategic priorities, audience constraints, or upcoming creative changes.
Use the Saturation Plot & Forecaster to Spot Diminishing Returns: If Prescient suggests a significant budget increase on a campaign that already shows diminishing returns, evaluate the trade-off carefully.

Step 3: Customize Recommendations As Needed

Leverage the ‘Customize Recommendations’ Feature: If certain campaigns can’t scale due to audience limits or if you have a campaign in test, lock them at their current spend. Then re-run the optimization to see updated allocations.
Combine with Forecasting Insights: If you need to drill deeper into one particular campaign (especially for quick checks), the individual Forecasting tab can offer a detailed view of potential spend curves and performance.

Step 4: Implement Changes & Track Performance

Allow Time for Stabilization: Especially if budgets changed by ±20% in the past 2–3 days, wait about 7 days before running a new optimization. This ensures the data better reflects the new spend levels.
Follow 1–2 Conversion Cycles Before Re-evaluating: For upper-funnel and video/CTV campaigns, consider waiting 3–4 cycles to see the full impact.
Use Prescient’s Tracking: Accept and specify an implementation date to track recommended changes within Prescient. Consistency in spend level is key for meaningful results.

When to Use Forecasting vs. Optimization

Optimizer: Ideal for holistic budget allocation and finding the most efficient mix across multiple campaigns or channels.
Forecaster: Best for exploring scale potential within a single campaign or testing the upper threshold of spend.
Keep Contextual Factors in Mind: Prescient’s tools don’t account for audience or impression-share limitations—use your own judgment on feasibility.

Handling Contradictory Data

Check the Attribution Drawer: If Prescient’s ROAS or revenue numbers differ from your platform or MTA reporting, it may be due to Amazon revenue or Halo effects that aren’t captured in other tools.
Contact Your CS Rep: If you see unexpected inefficiencies in your Optimizer scenario (e.g., ROAS dropping despite recommended changes) or encounter any error messages, reach out for support.

Revisiting & Frequency of Optimization

Optimize Every 2–4 Weeks: Or after any major budget shift. This timeframe typically allows you to gather enough post-implementation data to gauge performance changes.
Pair Budget Cycles & Major Campaign Milestones: Re-run scenarios ahead of new budget cycles, significant promotions, or strategic shifts to keep your spend allocations aligned with current goals.