Four offline checks before giving a marketing agent API access

Description and use

Four offline tools expose common marketing agent errors before live account access. Each tool uses synthetic data to validate CPA calculations, price versus package changes, CRM coverage, and local email eligibility checks.

Problem

Marketing agents can misinterpret data and make wrong decisions even when producing valid JSON. The goal is to expose such errors before connecting an agent to a live account.

Outcome

The four tools collectively demonstrate how to detect common agent misinterpretations—incorrect CPA calculation, price versus package changes, incomplete CRM reads, and local eligibility checks—without accessing live data.

Workflow

• Download and unpack each tool’s ZIP file from GitHub releases.
• Run the provided Python command for each tool with the synthetic example data.
• Observe the command’s exit code and output values.
• Compare the observed values with the expected results described in the article.
• Report any mismatches and explain unsupported conclusions the agent might make.
• Do not connect to live accounts or perform any real data operations.
• Use the results to refine agent prompts or workflows before live deployment.

Evidence

Each has synthetic input and a result you can reproduce with Python 3.9+ and the standard library.

Source

The fictional input contains five rows and four grouped placements. The result has 97 clicks, 6,000 RUB cost and 5 conversions. Total CPA is 1,200.00 RUB.

Source

In the synthetic strategy-session package, the price changes from 10,000 to 12,500 RUB: +25%. Delivery changes from four to six business days, and interview quantity changes from two to one.

Source

Deal 101 has the complete empty read. Deal 105 has incomplete coverage: its absence status cannot be established.

Source

The six invented rows produce one eligible-under-supplied-rules row and five excluded rows.

Source