Scribd
Upload
2 views3 pages

Amazon Ads Python Take Home Assignment

The document outlines a take-home assignment for building a Python service that processes Amazon Sponsored Products keyword performance data. It includes tasks for data handling and aggregation, bid optimization logic, deduplication, and architecture questions. Submission requires a GitHub link, a README, modular code, and unit tests.

Uploaded by

mmed pod
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
2 views3 pages

Amazon Ads Python Take Home Assignment

The document outlines a take-home assignment for building a Python service that processes Amazon Sponsored Products keyword performance data. It includes tasks for data handling and aggregation, bid optimization logic, deduplication, and architecture questions. Submission requires a GitHub link, a README, modular code, and unit tests.

Uploaded by

mmed pod
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Amazon Ads Data & Optimization - Take Home

Assignment

Objective

Build a small Python service that simulates pulling Amazon Sponsored Products keyword
performance data, performs data transformations, and applies bid optimization logic.

Part 1: Data Handling & Aggregation

• Load provided JSON performance data.

• Normalize into a Pandas DataFrame.

• Aggregate by campaignId, adGroupId, and keywordId.

• Compute CTR, CPC, CVR, ACoS, and ROAS.

• Handle edge cases (zero clicks, zero sales, missing values).

Sample Input Structure

[
{
"campaignId": "123",
"adGroupId": "111",
"keywordId": "9001",
"keywordText": "vitamin c serum",
"matchType": "EXACT",
"impressions": 10000,
"clicks": 120,
"cost": 240.0,
"sales": 800.0,
"orders": 20,
"date": "2026-01-10"
}
]

Part 2: Bid Optimization Logic

• Target ACoS = 25%.

• If sales = 0 and clicks ≥ 20 → decrease bid by 20%.

• If ACoS > 1.5 × target → decrease bid by 15%.

• If ACoS < 0.7 × target → increase bid by 10%.

• Otherwise → no change.
• Enforce min bid = 0.2 and max bid = 5.0.

Expected Output Format

[
{
"keywordId": "9001",
"oldBid": 1.2,
"newBid": 1.02,
"reason": "High ACoS"
}
]

Part 3: Deduplication & Data Freshness

• Simulate duplicate rows for the same (keywordId, date).

• Keep only the latest record per (keywordId, date).

• Explain how to enforce uniqueness in PostgreSQL.

• Explain how to prevent duplicate inserts at the application layer.

Part 4: Architecture Questions (Written)

• How would you handle Amazon Ads API rate limits?

• How would you design retries and backoff?

• How would you store and refresh OAuth tokens securely?

• How would you structure this inside a Django project?

• How would you implement a rolling 30-day window (excluding last 2 days)?

Submission Requirements

• GitHub repository link.

• Clear README with setup instructions.

• Well-structured and modular code.

• Include at least 3 unit tests.

Evaluation Criteria
• Code quality and structure.

• Correct metric calculations.

• Edge case handling.

• Optimization logic clarity.

• Architectural thinking.

You might also like