They Requested It. I Built It. Nobody Ever Used It.

2026-05-27 · Source: Towards Data Science · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Intermediate, medium

Summary

Data science models frequently fail to achieve adoption despite being requested and developed, a common issue for data professionals. This problem stems from three core challenges: models often act as "black boxes" lacking explainability, development cycles extend too long, and solutions are not easily integrated into existing workflows. In healthcare, for instance, clinicians prioritize trusted processes over complex, unexplainable predictive models, even if highly accurate. Prolonged development can lead stakeholders to find alternative solutions, underscoring the need for rapid iteration and frequent communication. Furthermore, models must seamlessly fit into users' daily operations, such as integrating predictions directly into electronic health record software like Epic, rather than introducing workflow friction.

Key takeaway

For Data Scientists and ML Engineers aiming for model adoption, prioritize user trust and workflow integration over raw predictive accuracy. You should develop concise model briefs explaining complex features in business terms and push for rapid v1 deployments. Continuously communicate progress and ensure predictions seamlessly fit into existing tools, like Epic in healthcare, to avoid abandonment and maximize impact.

Key insights

Data models fail adoption due to explainability, slow delivery, and poor integration into user workflows.

Principles

• Explainability often outweighs raw accuracy for user trust.
• "Don't let the perfect get in the way of the good."
• Seamless integration is critical for model adoption.

Method

Create an "easily digestible model brief" (slides) defining population, target, features, and performance, using business-centric language. Communicate progress frequently.

In practice

• Develop a model brief defining terms in business context.
• Prioritize rapid iteration (v1) over perfect initial solutions.
• Integrate predictions directly into existing user systems (e.g., Epic).

Topics

• Model Adoption
• Explainable AI
• Data Science Workflow
• Healthcare AI
• User Experience
• Iterative Development

Best for: AI Product Manager, Product Manager, Data Scientist, Machine Learning Engineer, Director of AI/ML

Related on AIssential

• Explaining AI Without Code: A User Study on Explainable AI — Integrating XAI into no-code ML platforms improves usability and trust, but must balance novice accessibility with expert diagnostic depth.
• April 2026 PDC State of Data Modeling Survey Results Are In! — Effective data modeling hinges on training, clear requirements, dedicated ownership, and sufficient time, not primarily on new tools.
• Atheni secures £350K to help teams use AI more effectively — Effective AI adoption stems from embedding role-specific guidance directly into daily workflows, fostering "master craftspeople."
• H2O MLOps Enterprise Model Registry & Hugging Face | Part 8 Integration | Part 8 — A robust model repository is essential for managing and governing a large, diverse portfolio of AI models in production.
• From Radiology to Drug Discovery, Survey Reveals AI Is Delivering Clear Return on Investment in Healthcare — Healthcare and life sciences are rapidly moving from AI experimentation to execution, driven by strong ROI and open-source adoption.
• Enterprises can now train custom AI models from production workflows — no ML team required — Enterprise AI application interactions generate valuable training data that most organizations currently fail to capture.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Towards Data Science.