Case Study: A media buying system for a leading US-based advertising agency
Explore how the SumatoSoft team developed a custom web-based system, achieved 50x faster performance, intuitive workflows, and advanced analytics, simplifying users’ daily operations.
Project details:
About the Client:
The Client is a leading US-based audio advertising agency specializing in radio and podcast advertising. Their legacy system, developed over 20 years ago using FileMaker, was outdated, low-performing, and difficult to use. It managed millions of records but faced issues like slow response times, hours-long file uploads, and frequent system failures.
Location: USA
Industry: Advertisement
Type: MIS (Management Information System)
Team: 13 members, including Project Manager, Business Analyst, QA, Frontend Dev, Backend Dev, DevOps, UX/UI Designer
Project duration: 2 years
Business сhallenge
The Client approached us with the goal of building a new high-performance and user-friendly solution capable of addressing the required data volume. High performance was a core requirement for this project.
Additional requirements:
- a new system should cover additional user scenarios that were previously unsupported.
- switching to cloud infrastructure to reduce efforts and costs associated with maintaining onsite equipment.
- the previous system relied heavily on workarounds to complete tasks. The new platform required properly implemented features.
Our solution
The SumatoSoft team developed a web-based record-keeping platform that addresses over one hundred user scenarios, ensures responses to employee requests within seconds, and meets modern standards of user responsiveness. The solution is organized into several logical modules, each handling distinct aspects of the Client’s business flow and displaying just enough information on the first screen to enable employees to make swift decisions.
The system supports three user roles: the Client’s employees, Admins, and media outlets, including podcasters and radio stations.
System overview
Advertising customers ask for ad placements on radio stations or podcasts. The employees create orders in the system and prepare detailed requests for media outlets, specifying the number of plays and media materials before sending requests to media outlets. Media outlets can accept, decline, or suggest changes to the requests. Employees can then monitor the progress of advertising campaigns, track payments, and provide additional analytics to advertising customers on campaign performance.
A comprehensive analytics algorithm was also developed, enabling the system to evaluate ad performance even for the most extensive campaigns, such as those spanning multiple states and diverse media outlets, without relying on explicit identifiers like promo codes. The system supports importing thousands of response records from Excel without system disruption.
Early User Testing
Since the Client’s employees are the core users of the system, the development team closely collaborated with them to ensure the platform met their needs. We conducted regular end-user interviews, tested prototypes, and presented multiple design options to gather valuable feedback and refine the solution.
High-Performance Requirement
We conducted rigorous performance and load testing using tools such as Locust, Jmeter. After completing the core development process, the team spent some time optimizing performance metrics and stabilizing the system. Database queries were fine-tuned, asynchronous processing was implemented, and response times for actions were reduced to seconds. These efforts resulted in a performance improvement of approximately 50 times compared to the legacy system.
Business value
Before:
- Slow response times and frequent system failures regularly disrupted the team’s workflows
- File uploads took hours, blocking employees from other tasks until processing completed
- Even basic tasks required workarounds because the legacy system lacked properly implemented features
- Measuring audio ad performance was unreliable – responses could not be attributed to specific ads without explicit identifiers
After:
- System response is now 50x faster, and the failure cycles that disrupted daily operations are eliminated
- File processing that previously took hours now completes in under a minute – a >90% reduction in wait time – while employees continue other tasks in parallel
- All operational scenarios now have purpose-built features; 100+ workflows run without any workarounds
- A custom attribution algorithm evaluates campaign performance across multi-state, multi-outlet campaigns without requiring promo codes or other explicit identifiers
Have an app idea?
Let’s start with a free quote!
See our other case studies
IoT application with sensors for industrial fridge monitoring
Graphical user interface for robot operation
IoT apps for controlling an air conditioning system
