Persado Model Evaluation
Service · Audit & assessment · Marketing by Persado
Evaluate Persado's predictive intelligence on your own A/B tests.
Persado Model Evaluation is a professional assessment service that allows companies to validate Persado's predictive accuracy using their own historical A/B test data. It helps organizations quantify the impact of AI-driven content ranking before campaigns are sent.
- Launched
- 2026-09-01
Category: Predictive Analytics
- Duration
- About two business days for scoring
- Delivery
- Remote (Excel workbook exchange)
- Industries served
- Financial services, Retail, Travel & hospitality, Telecom
- Minimum campaigns
- 40
- Maximum campaigns
- 300
- Maximum messages
- 1,200 messages
- Validates predictive accuracy using historical A/B test data
- Reports Top-1 accuracy and Spearman rank correlation metrics
- Strict blind protocol where outcomes stay with the customer
- Typical turnaround time of two business days for scoring
- Supports evaluations of 40 to 300 campaigns and 1,200 messages
Overview
Persado Model Evaluation is a professional assessment service for businesses to validate the predictive accuracy of Persado's Motivation AI models using their own historical A/B test data. The service follows a strict blind protocol where the customer provides past campaign variants with results removed, and Persado's models rank them. Customers then compare the model's predicted winners against observed outcomes to calculate Top-1 accuracy, Spearman rank correlation, and run-to-run stability. This evaluation helps organizations in industries like retail and financial services quantify the potential impact of moving decision-making from after-the-send to before-the-send. The process is delivered via a structured Excel workbook and typically provides results within two business days. Persado ensures data privacy by never requesting recipient data or campaign outcomes, which remain entirely with the customer.
Evaluation Process
- Predictive Intelligence Audit: The service provides a structured evaluation of Persado's predictive intelligence using the customer's own past A/B tests.
- Blind Scoring Protocol: The model scores every variant in a provided workbook without seeing the original outcomes.
- Metric Calculation: The evaluation reports three key metrics: Top-1 accuracy, Spearman rank correlation, and run-to-run stability.
- Statistical Validation: The service includes a two-proportion test on each campaign's top two messages to set aside close calls.
Delivery Specifications
- Campaign Volume Limits: The evaluation typically covers 40 to 300 campaigns and up to 1,200 messages.
- Turnaround Time: Scored workbooks are typically returned to the customer within two business days.
- Data Privacy Guardrails: The process ensures that campaign outcomes stay with the customer and are never shared with Persado.
- Scope Constraints: The evaluation is restricted to one channel and one industry per workbook.
What's included
- 40 to 300 campaigns, up to 1,200 messages, one channel and one industry
Deliverables
- Scored Excel workbook with automated metrics calculation and a worked example
How it works
- Fill in workbook with messages, Send workbook to Persado, Persado returns scored workbook, Enter outcomes to compute metrics
Use cases
- Marketing teams validate Persado's predictive accuracy by comparing model rankings against their own historical email subject line A/B tests.
- Platform providers assess how integrating Persado's ranking model could move content decisions from after-the-send to the point of creation.
- Data analysts use the Spearman rank correlation metric to measure the agreement between predicted and observed orders of all campaign variants.
- Organizations verify model stability by reviewing five side-by-side scoring runs for every message submitted in the evaluation workbook.
Specifications
- Duration
- About two business days for scoring
- Delivery
- Remote (Excel workbook exchange)
- Industries served
- Financial services, Retail, Travel & hospitality, Telecom
- Minimum campaigns
- 40
- Maximum campaigns
- 300
- Maximum messages
- 1,200 messages
- Turnaround time
- 2 business days
- File format
- Excel workbook (approx. 600 KB)
- Data retention
- Deleted after 30 days
- Metrics reported
- Top-1 accuracy, Spearman rank correlation, run-to-run stability
- Scoring frequency
- 5 runs per message
How it works
- Fill in the workbook: The customer adds one row per variant and campaign name to the "Enter messages" sheet of the provided Excel workbook, leaving out all results.
- Send it back: The customer replies to their Persado contact with the workbook attached, keeping the results sheet empty.
- Persado returns scored workbook: Persado returns the workbook with every message scored five times within its own campaign, typically within two business days.
- Enter outcomes and calculate metrics: The customer adds unique clicks and messages delivered to the "Score vs Response" sheet to automatically compute CTR and accuracy metrics.
FAQ
What are the data volume limits for the evaluation?
The evaluation requires a minimum of 40 campaigns and can accommodate up to 300 campaigns, with a total limit of 1,200 messages.
How long does the evaluation take?
Persado typically returns the scored workbook within approximately two business days after receiving the customer's message data.
How is the accuracy of the evaluation ensured?
The protocol is strictly blind; customers withhold campaign outcomes until after Persado delivers the scores, ensuring the model ranks variants without seeing results.
Is my campaign data used for training?
Persado receives only campaign names and message text; it never asks for recipient data or campaign results, and submitted content is deleted after 30 days.
Added by AntSpace from public sources (persado.com).