Media Buying System for a Leading US-Based Advertising Agency
Challenge
The client is a US-based audio advertising agency specializing in radio and podcast advertising. Its legacy record-keeping system, built on FileMaker more than 20 years ago, held millions of records but responded slowly, failed often, and blocked employees for hours during file uploads.
Even basic tasks required workarounds. The agency needed a high-performance, easy-to-use replacement that covered previously unsupported user scenarios, moved to cloud infrastructure to cut on-site maintenance, and measured ad performance without promo codes.
The client is a US-based audio advertising agency specializing in radio and podcast advertising. Its legacy record-keeping system, built on FileMaker more than 20 years ago, held millions of records but responded slowly, failed often, and blocked employees for hours during file uploads.
Even basic tasks required workarounds. The agency needed a high-performance, easy-to-use replacement that covered previously unsupported user scenarios, moved to cloud infrastructure to cut on-site maintenance, and measured ad performance without promo codes.
Solution
SumatoSoft built a web-based record-keeping platform organized into logical modules that cover more than 100 user scenarios for three roles: agency employees, admins, and media outlets such as radio stations and podcasters.
Employees create orders and send placement requests that specify plays and media materials; outlets accept, decline, or suggest changes; the team then tracks campaign progress and payments. A custom attribution algorithm evaluates performance across multi-state, multi-outlet campaigns, and the system imports thousands of response records from Excel without disruption.
The team ran regular end-user interviews and prototype tests, then load-tested the platform with Locust and JMeter, tuned database queries, and added asynchronous processing.
SumatoSoft built a web-based record-keeping platform organized into logical modules that cover more than 100 user scenarios for three roles: agency employees, admins, and media outlets such as radio stations and podcasters.
Employees create orders and send placement requests that specify plays and media materials; outlets accept, decline, or suggest changes; the team then tracks campaign progress and payments. A custom attribution algorithm evaluates performance across multi-state, multi-outlet campaigns, and the system imports thousands of response records from Excel without disruption.
The team ran regular end-user interviews and prototype tests, then load-tested the platform with Locust and JMeter, tuned database queries, and added asynchronous processing.
Results
- System response about 50x faster than the legacy system, with the failure cycles that disrupted daily work eliminated
- File processing cut from hours to under a minute, a reduction of more than 90%, while employees keep working in parallel
- 100+ advertising workflows run on purpose-built features without workarounds
- Campaign performance measured across multi-state, multi-outlet campaigns without promo codes or other explicit identifiers
- System response about 50x faster than the legacy system, with the failure cycles that disrupted daily work eliminated
- File processing cut from hours to under a minute, a reduction of more than 90%, while employees keep working in parallel
- 100+ advertising workflows run on purpose-built features without workarounds
- Campaign performance measured across multi-state, multi-outlet campaigns without promo codes or other explicit identifiers