DL-020 · Case study
Amazon data platform for an advertising agency
An Amazon Ads agency wanted its own data layer instead of rented reporting tools. We built a platform that loads seller, vendor, Ads and DSP data into BigQuery.
- Client
- Name withheld
- Location
- Germany
- Industry
- Amazon Ads agency
- Capabilities
- 03Build02Stabilize
01
Challenge
- Client reporting depended on third-party tools the agency did not control.
- Each new client account meant manual setup across seller, vendor and Ads APIs.
- Exports to the warehouse were too slow for daily agency reporting.
02
Environment
- Many agency clients across Seller Central, Vendor Central and Amazon Ads.
- Sponsored Products, Brands, Display and DSP reporting on several marketplaces.
- Reports consumed in BigQuery and Looker Studio dashboards.
03
Intervention
- Admin portal for self-service client onboarding with paired SP-API and Ads authorisation.
- Raw report layer plus modelled tables, exported to the agency's own BigQuery.
- Rebuilt the BigQuery exporter and fixed the warehouse cost drivers.
- Per-seller concurrency so one account's backlog cannot stall the rest.
04
Outcome
- The agency owns its Amazon data pipeline instead of renting one.
- Export throughput rose tenfold, to tens of millions of rows a day.
- Daily warehouse spend fell by about 90%.
- 01SELLER CENTRALSP-API→DATA PLATFORMLambda / SQS
- 02VENDOR CENTRALSP-API→DATA PLATFORMLambda / SQS
- 03ADS + DSPAds API→DATA PLATFORMLambda / SQS
- 04ADMIN PORTALClient onboarding- -DATA PLATFORMLambda / SQSclient auth
- 05DATA PLATFORMLambda / SQS→BIGQUERYRaw + modelled10x export
- 06BIGQUERYRaw + modelled→LOOKER STUDIODashboards
Results, as published.
- 10x
- Export throughput to BigQuery
- ~90%
- Lower daily warehouse cost
Stack and services.
Tell us what has to work.
A senior engineer reads every brief. Describe the system, what it runs on today and what happens when it fails.
Discuss a project→