Overview

Amazon has broadened its Marketing Stream service to include data from Demand‑Side Platform (DSP) campaigns, delivering performance metrics on an hourly basis. The change, announced in early 2024, lets sellers and agencies that run both DSP and Sponsored Ads see a unified, near‑real‑time view of their ad spend. This faster insight enables quicker budget shifts, more precise dayparting, and automated alerts that can protect margins during high‑stakes events.

Key Points

• Hourly DSP metrics — Impressions, clicks, spend, CTR, and viewability from DSP campaigns now arrive every hour, matching the cadence already available for Sponsored Products, Brands, and Display.
• Push‑based delivery — Data are sent as structured JSON to an Amazon SQS queue in the advertiser’s AWS account, eliminating the need to poll APIs for daily aggregates.
• Instant change alerts — Real‑time notifications fire when a campaign’s budget, status, or targeting parameters are altered, allowing teams to react to both intentional edits and accidental misconfigurations.
• Single data pipeline — DSP and Sponsored Ads information flow through the same SQS queue, simplifying dashboard construction and cross‑channel optimization logic.
• Automation‑ready — The hourly stream can trigger Lambda functions, budget‑pacing rules, or pause actions without human intervention, reducing manual monitoring workloads.

How Marketing Stream Works

1. Data extraction — Amazon’s ad servers compile performance numbers for each DSP and Sponsored Ads campaign every hour. For example, a DSP campaign that generated 12,000 impressions and $450 spend between 2 PM and 3 PM will be packaged into a JSON payload.
2. Push to SQS — The JSON payload is automatically pushed to an Amazon Simple Queue Service (SQS) queue that the advertiser has created in their AWS account. No API calls are required; the queue receives the data the moment it is ready.
3. Processing layer — A Lambda function (or similar compute service) reads the message from the queue, transforms the fields into a normalized schema, and writes the result to a data lake such as Amazon S3 or a data warehouse like Redshift. For instance, the function may convert the raw “clicks” field into a numeric column and calculate a derived “cost per click” metric.