Overview
Nextant partnered with a global software marketplace to transform thousands of unstructured product listings into trusted, campaign-ready collections. We developed a cloud-based AI merchandising application that evaluates category and industry fit, ranks offers, and explains every recommendation. The solution replaced a slow, subjective curation process with a scalable workflow that helps marketing teams activate campaigns faster and with greater confidence.
Challenge
As the marketplace catalog expanded, campaign curation became increasingly difficult to manage with consistency, speed, and confidence. Thousands of listings needed to be reviewed and grouped into meaningful collections by category and industry, but the existing process depended heavily on manual interpretation and keyword matching.
This created several business challenges:
- Manual tagging could not scale: Reviewing and classifying thousands of listings by hand was slow, time-consuming, and difficult to standardize across reviewers.
- Curation decisions were subjective: Different reviewers could interpret the same listing differently, creating inconsistency across campaigns and reducing confidence in the final collections.
- Keyword search produced unreliable results: Marketplace listings often include broad marketing language, which can make keyword matches appear relevant even when the underlying offer does not truly fit the intended category or industry.
- Campaign teams lacked explainability: Without clear rationale behind why a listing was included, excluded, or flagged for review, teams had limited confidence in the quality of curated collections.
- Growth increased operational pressure: As the number of listings continued to expand, the manual process became harder to maintain and less effective as a long-term operating model.
For the marketplace team, this was not just a catalog management issue. It affected campaign speed, quality, discoverability, and the ability to confidently promote the right solutions to the right audiences. The organization needed a smarter, more scalable approach that could combine automation with transparency and human oversight.
Solution
Nextant developed a live, cloud-based AI merchandising application that evaluates how well marketplace listings align with specific categories and industries. Rather than relying on keyword matching alone, the solution analyzes the substance of each offer, identifies its likely use cases, and applies a consistent scoring framework to generate transparent recommendations.
The application combines AI engineering, business rules, and human oversight into a repeatable workflow. Each listing is classified as Relevant, Needs Review, or Reject, allowing teams to automate high-volume screening while retaining control over ambiguous or judgment-based decisions.
How the Solution Works
The application evaluates each marketplace listing through a structured, multi-layered scoring process:
- AI-powered extraction: Large language models analyze each listing to identify its core capabilities, use cases, target audiences, and business relevance.
- Semantic analysis: The solution evaluates meaning and context rather than relying solely on keyword overlap, helping identify listings that genuinely align with the selected category or industry.
- Hybrid scoring: LLM-based evaluation, semantic similarity, and configurable business rules are combined to produce balanced and consistent recommendations.
- Explainable classification: Each result includes a clear rationale explaining why the listing was classified as Relevant, Needs Review, or Reject.
- Human-in-the-loop review: Listings with uncertain or borderline scores are surfaced for review, allowing subject-matter experts to apply judgment where it adds the most value.
This approach helps teams apply the same evaluation logic across thousands of listings while preserving transparency, governance, and human oversight.
How Users Interact with the Application
The application translates AI scoring into a practical merchandising workflow that campaign and marketing teams can use directly:
- Ranked listing results: Users can review listings based on their category or industry fit, prioritizing the strongest matches first.
- Transparent recommendations: Each classification includes supporting reasoning, making it easier to validate results and resolve borderline cases.
- Chat-based exploration: An embedded AI assistant allows users to ask questions about listings, categories, recommendations, and scoring outcomes.
- Second-opinion validation: Users can request an additional AI assessment for offers that require deeper review or further confirmation.
- Marketing copy generation: The application can help transform listing information into campaign-ready descriptions and messaging.
- Reusable campaign collections: Approved listings can be organized into curated collections that teams can reuse and adapt across future campaigns, industries, and audiences.
By combining automated scoring with intuitive review and campaign activation capabilities, the application helps merchandising teams move from unstructured marketplace data to trusted, campaign-ready collections more efficiently.
Results and Impact
By automating category and industry scoring end to end, the solution helped the marketplace team move from a slow, subjective manual process to a scalable, explainable AI-powered operating model.
The impact included:
- Faster campaign curation: What previously required days of manual review can now be supported through an on-demand scoring workflow.
- Greater consistency across listings: AI-assisted scoring reduces variability and helps teams apply the same evaluation logic across thousands of offers.
- Improved trust in results: Clear explanations make it easier for users to understand and validate recommendations.
- Better use of expert time: Teams can focus attention on strategic decisions and borderline cases instead of repetitive review.
- Scalable campaign activation: Curated collections can be reused and adapted across categories, industries, and future campaigns.
- Stronger customer discovery: Better-organized listings make it easier to surface relevant solutions to the right audiences.
- A foundation for future AI merchandising: The solution creates a repeatable model that can evolve as the marketplace, categories, and business priorities change.
The marketplace team now has a scalable AI merchandising engine that transforms unstructured listings into curated, campaign-ready collections. Instead of relying on manual tagging and keyword search, the team can use an explainable scoring system that supports faster decisions, higher-quality campaigns, and greater confidence in marketplace discovery.
Why Nextant
This case study reflects how Nextant helps organizations move from AI experimentation to practical, measurable business impact. Many companies know AI can improve efficiency, but struggle to identify the right use cases, design responsible workflows, and implement solutions that teams will actually use. Nextant bridges that gap by combining business strategy, AI technical expertise, Microsoft ecosystem knowledge, and execution discipline.
For clients, this means Nextant can help:
- Turn complex business problems into practical AI solutions: Nextant starts with the business challenge, not the technology, ensuring AI is tied to measurable outcomes.
- Design solutions users can trust: Explainability, transparency, and human oversight are built into the experience so teams can adopt AI with confidence.
- Move from idea to implementation: Nextant brings the consulting, technical, and delivery expertise required to build working solutions, not just roadmaps or prototypes.
- Scale with the Microsoft ecosystem: With experience across Azure AI, Power Platform, Dataverse, Microsoft 365, and automation technologies, Nextant helps organizations maximize the value of their existing Microsoft investments.
- Balance speed, quality, and cost efficiency: Nextant’s integrated U.S. and nearshore delivery model provides flexibility, scalability, and strong execution capacity.
- Create solutions that drive adoption: Beyond building technology, Nextant focuses on workflows, governance, enablement, and change management so AI becomes part of how work gets done.
Organizations looking to apply AI to merchandising, sales, marketing, operations, knowledge management, or business process transformation need more than a tool. They need a partner that understands how to connect business priorities, user needs, data, technology, and adoption into one practical solution.
Nextant brings that combination of strategic thinking and hands-on delivery. We help clients identify where AI can create real value, build solutions that work in the real world, and scale those solutions into measurable business impact.
Technology Used
- Semantic Search & RAG
- Azure OpenAI (GPT-5, Mistral)
- Azure Serverless (Functions)
- Dataverse / Power Platform
- Full-Stack Web App
