Model Results

Revenue & CAC

The Prescient platform measures the relationship between revenue and customer acquisition costs (CAC) across different campaign types. Understanding how these metrics interact is crucial for interpreting your results and optimizing campaign performance.

Reading Your Results

Campaign Type Differences

When analyzing revenue and CAC metrics, you'll see distinct patterns between campaign types:

Retargeting Campaigns

Generally show higher conversion rates
Display lower CAC values
Generate more immediate revenue
Limited impact on customer base growth

Prospecting Campaigns

Typically show higher CAC values
Display lower immediate conversion rates
Generate long-term revenue effects
Critical for customer base expansion

Key Metrics Displayed

The platform shows several key relationships:

Direct revenue attribution by campaign type
CAC trends over time
Audience growth measurements
Campaign efficiency metrics
Seasonal performance patterns

Understanding Metric Interactions

Revenue Impact Time Frames

Revenue effects appear in two distinct time frames:

Immediate Impact

Direct conversions
Short-term revenue attribution
Primarily from retargeting campaigns

Long-term Impact

Audience pool growth
Future conversion potential
Primarily from prospecting campaigns

CAC Interpretation

When analyzing CAC in your results:

Higher CAC isn't necessarily negative for prospecting campaigns
Lower CAC in retargeting reflects warm audience targeting
Seasonal variations affect CAC across campaign types
Consider CAC alongside audience growth metrics

Limitations and Considerations

Important factors to consider when analyzing these metrics:

CAC calculations exclude halo effects
Revenue attribution has different windows for different campaign types
Seasonal factors can affect both metrics
Audience pool size influences retargeting efficiency

Best Practices for Analysis

When evaluating revenue and CAC relationships:

Compare metrics within campaign types rather than across types
Consider both immediate and long-term revenue effects
Account for seasonal variations in performance
Evaluate CAC in context of overall growth objectives

Common Questions

Q: Why do my prospecting campaigns show higher CAC?

A: Prospecting campaigns target cold audiences, requiring more investment to convert new customers compared to retargeting warm audiences.

Q: How should I balance high CAC prospecting with efficient retargeting?

A: Focus on total revenue impact and audience growth rather than minimizing CAC. The platform's optimization tools can help find the right balance for your specific goals.

Q: Why do revenue and CAC metrics vary seasonally?

A: Seasonal factors affect both customer behavior and advertising costs, leading to natural variations in both metrics throughout the year.

New Customers

New customer metrics in Prescient provide insight into customer acquisition across your marketing campaigns. The platform employs sophisticated de-duplication algorithms to ensure accurate customer counting across channels and campaigns.

Reading Your Results

New Customer Counting

When viewing new customer metrics in your dashboard:

Customer Definition: A new customer is someone who makes their first purchase based on their unique customer ID (if ingested directly from the platform) and first recognized order date. This is not based on a specified time period or window - it's determined by the actual first order date in the data we have available from the ingestion process.
Modeled New Customers from Paid Media: New customers are attributed to specific marketing campaigns based on their probabilistic conversion pathway based on Prescient's models. There is no 'ground truth' for new customers driven by specific marketing campaigns which is why our statistical methods continue to be valuable signal for our users.

Specific Customer Definitions by Platform

Shopify & SFCC (Salesforce Commerce Cloud):

Based on unique customer_ID
Determined by the first recognized order date
We ingest all available data allowed by the API which is typically the full store history (not limited to the modeling window that we use)

Amazon:

Based on unique buyer/marketplace email. Emails can only be collected for Amazon Seller Central and not for Amazon Vendor Central.
Determined by the first recognized order date
We ingest all available data allowed by the API (typically 18-24 months).

Google Analytics:

No new customer data is available through this integration.

Custom Integration:

When connecting to a data warehouse or other form of ingestion method for a custom source, the definition of a new customer is determined directly by the user.

Data Behavior Patterns

Several normal patterns appear in new customer data:

Daily Fluctuations: Numbers may vary in the first few days after spend
Attribution Windows: Customer counts may adjust as attribution windows close
Spend Impact Delays: Customer acquisition often lags behind spend by 1-3 days

Understanding Metric Fluctuations

Common Patterns
You may notice these typical behaviors in your new customer metrics:

Initial Reporting: Day 1 numbers are preliminary
Stabilization Period: Data typically stabilizes by day 4
Attribution Updates: Customer counts may shift as the platform processes delayed conversions

Why Metrics Change

Several factors influence new customer metric fluctuations:

Attribution windows closing
Delayed conversion effects
De-duplication processing
Spend impact lag time

Limitations and Considerations

When analyzing new customer data:

Initial day metrics are subject to change
Customer counts exclude halo effects
De-duplication may affect historical numbers
Attribution windows vary by channel

Best Practices for Analysis

When working with new customer metrics:

Allow 3-4 days for data stabilization
Consider spend timing when analyzing results
Look for consistent patterns rather than daily fluctuations
Account for attribution windows in your analysis

Common Questions

Q: Why do my customer numbers change over several days?

A: Customer numbers may fluctuate for up to 4 days due to attribution windows and delayed conversion effects from marketing spend.

Q: How does the platform handle duplicate customers?

A: The platform uses de-duplication algorithms to identify and merge duplicate customer records across channels and campaigns.

Q: When should I consider my new customer numbers final?

A: While numbers can continue to adjust slightly, day 4 numbers are typically stable and can be considered reliable for analysis.

Halo Effects

Halo effects represent the indirect revenue impact of your marketing campaigns beyond direct conversions. In your Prescient dashboard, you'll see halo effects measured as specific dollar amounts attributed to different indirect revenue streams.

Reading Your Results

Direct vs. Halo Revenue
When viewing campaign results, you'll see two distinct types of revenue:

Direct Revenue: Revenue from users who directly clicked through your ads and converted
Halo Revenue: Revenue from indirect pathways, broken down into several categories:

Organic traffic
Direct traffic
Branded search
Amazon organic sales (for omnichannel brands)

Important Considerations

When interpreting halo effects in your results:

Metric Independence: Halo effects are calculated separately from metrics like CAC and New Customer counts. These metrics currently exclude halo effects due to model complexity.
Amazon Attribution: For omnichannel brands, halo effects can show how off-Amazon marketing (e.g., Meta campaigns) influences organic Amazon sales.
Channel Interactions: Different marketing channels may show varying levels of halo effects. For example:
- Social media campaigns often show strong organic traffic halos
- Video advertising frequently generates direct traffic halos
- Brand awareness campaigns typically impact branded search volume

Limitations and Caveats

To accurately interpret your results, be aware that:

Halo effects represent probabilistic attributions based on statistical modeling
The relationship between campaigns and halo effects may vary over time
External factors (seasonality, market conditions, etc.) can influence halo effect measurements
Current model iterations do not include halo effects in new customer calculations or CAC metrics

Using Halo Effect Data

When using halo effect data for decision-making:

Consider both direct and halo revenue when evaluating campaign performance
Look for patterns in which channels generate stronger halo effects for specific revenue streams
Use halo effect insights to inform upper-funnel marketing strategies
Account for the exclusion of halo effects when analyzing customer acquisition metrics

Common Questions

Q: Why don't I see halo effects in my CAC calculations?

A: Currently, CAC calculations exclude halo effects due to complexity in reconciling these effects with customer acquisition data.

Q: How are Amazon organic sales attributed?

A: The model identifies statistical relationships between off-Amazon marketing activities and changes in organic Amazon sales patterns, attributing these effects as halos when significant correlations are found.

Q: Can halo effects change over time?

A: Yes, the strength and distribution of halo effects can vary based on numerous factors including seasonality, campaign type, and market conditions.

Attribution Percentages

The Prescient platform shows the percentage of your total revenue and new customers that can be attributed to your marketing activities. These percentages help you understand how much of your business performance is explained by your measured marketing efforts.

Reading Your Results

Attribution Percentages

You'll see key attribution metrics in your dashboard:

R-MMM Paid %: The paid media attributed revenue over total forecasted revenue. This is % of impact from paid media to your overall revenue (as opposed to seasonality, trend, and other non-explained factors).
C-MMM Paid %: The paid media attributed new customers over total forecasted new customers. This is % of impact from paid media to your overall new customers (as opposed to seasonality, trend, and other non-explained factors).

Expected Ranges

Typical attribution ranges you might see:

Revenue attribution: 70-90% of total revenue
New customer attribution: 40-60% of new customers

These ranges are normal and reflect different measurement approaches for revenue and customers.

Understanding Different Attribution Levels

Revenue vs Customer Attribution

Several factors explain why revenue and customer attribution percentages differ:

Measurement Approach

Revenue attribution uses detailed revenue data
Customer attribution uses more constrained identification methods

Business Factors

Promotional periods affect revenue and customer acquisition differently
Average order values fluctuate with marketing activities
Multi-platform sales (e.g., Amazon + direct) impact attribution differently

Attribution Caps

Important points about attribution limits:

Total attribution cannot exceed 100% of actual revenue
If you see higher percentages, please contact support
Attribution gaps are normal and expected

Limitations and Considerations

When analyzing attribution percentages:

Not all marketing activities may be measured
External factors influence both metrics
Platform limitations affect customer tracking
Some revenue sources may not be fully captured

Best Practices for Analysis

When working with attribution percentages:

Compare trends rather than absolute numbers
Consider seasonal and promotional impacts
Account for multi-platform sales effects
Review both metrics for complete performance understanding

Common Questions

Q: Why is my revenue attribution higher than customer attribution?

A: This is normal due to different measurement approaches and the complexity of customer identification across platforms.

Q: What does it mean if my MMM % of Revenue exceeds 100%?

A: This indicates a technical issue that should be reported to support, as attribution cannot exceed total revenue.

Q: Should I expect to see 100% attribution?

A: No, it's normal and expected to have some portion of revenue and customers that cannot be attributed to measured marketing activities.