How Intertec Automated Offer Processing with AI-Powered Metadata Extraction
Value Delivered
+90%
More accurate and reliable metadata extraction led to more consistent analytics and reporting
+70%
Reduction of manual review, due to new process minimizing time spent on manual data sorting and review
+40%
Reduction in operational costs, due to automation of the extraction and classification process
INDUSTRY
E-commerce
PROJECT DURATION
10 years
LOCATION
Worldwide
CLUTCH REVIEW
5
Client Bio
An e-commerce platform needed to serve customers around the globe with accurate, localized content. But each department had its own system for translating product descriptions, support articles, and marketing materials. This patchwork approach led to inconsistencies and made it hard to maintain a single, unified brand message.
Situation
The client operates in a highly competitive e-commerce landscape, partnering with multiple brands and affiliates to deliver promotions and discounts to customers. Their platform receives numerous offers in different text formats, some are well-structured, while others lack critical details or come in multiple languages. As a result, determining which offers merited attention was a time-consuming task.
Because the volume of incoming offers was so high, the client’s internal teams often struggled to keep up. They spent a lot of time manually reviewing each offer to ensure it met specific standards. Inconsistent formats and missing details led to slow turnaround times, sometimes causing promising opportunities to be delayed or overlooked.
The client’s primary goal was to speed up and improve the accuracy of this process. They wanted an automated system that could parse incoming text, identify key pieces of information, and rank offers based on quality and relevance. By doing so, they hoped to reduce manual work, minimize errors, and free up resources for more strategic tasks.
Solution
Solution
Intertec partnered with the client to develop a service for metadata extraction of voucher content, a robust solution for analyzing incoming offers. Our approach involved several steps to ensure the solution fit seamlessly into the client’s workflow:
- Metadata Extraction Service: Developed a robust solution for parsing and analyzing incoming voucher offers, seamlessly integrating with the client's workflow.
- Targeted Assessment: Evaluated the data ingestion process and offer types to identify key metadata fields and address language diversity using Amazon Translate.
- Custom Name Entity Recognition (NER) Implementation: Leveraged Amazon Comprehend with custom-trained NER models to accurately extract domain-specific data.
- Seamless Deployment & Integration: Utilized AWS services (Athena, S3, VPC, EC2, ECR, ECS) with continuous integration and real-time monitoring to ensure reliable performance.
Impact
- High Volume of Low-Quality Offers
The majority of incoming offers were either irrelevant or incomplete. Sorting through these low-quality offers took valuable time away from more promising deals. - Inconsistent Formats and Languages
Offers arrived in various structures and often in multiple languages, making it difficult to apply a one-size-fits-all approach to data extraction. This inconsistency led to additional manual intervention. - Need for Accurate Metadata Extraction
The client had to capture specific fields such as discount amount, discount type, category, and target audience. Without a reliable system for extracting these details, offers could be incorrectly categorized, which affected the ranking and overall customer experience.
