# Training-Serving Skew

> Live data and training data are made in two ways. So a good model does badly.

**Field:** Artificial Intelligence  
**Category:** MLOps & Deployment  
**Also known as:** Train-Serve Skew, Training/Serving Skew, Online-Offline Skew

## Think of it like

It is like practising football on grass. Then the real game is played on ice.

## Definition

Training-serving skew is when the data a model sees in real use differs from the data it was trained on, because of how the system was built. Often two pieces of code compute the same input in different ways. For example, training may use prices in dollars while the live app sends cents. The model still answers, but the answers are worse. It is a bug in the system, not a change in the world.

## Example

Emma's model was trained on ages in years. The live app sends ages in months, so every customer looks 12 times older and the scores go wrong.

## Why it matters

Nothing crashes, so it can hide for months. A model that tested well quietly gives poor answers.

## Common mistake

Calling it data drift. Drift is the world changing. Skew is your own two pipelines not matching.

## Three principles

- **Same input, two ways.** Training and live code do not match.
- **Silent harm.** No error, just worse answers.
- **Share the code.** One feature step for training and serving.

## Related terms

- [Data Drift](https://learn.benibix.com/term/ai/data-drift)
- [Feature Store](https://learn.benibix.com/term/ai/feature-store)
- [Model Monitoring](https://learn.benibix.com/term/ai/model-monitoring)
- [Feature Engineering](https://learn.benibix.com/term/ai/feature-engineering)
- [Real-Time Inference](https://learn.benibix.com/term/ai/real-time-inference)

## Told apart from

- [Training-Serving Skew vs Data Drift](https://learn.benibix.com/compare/training-serving-skew-vs-data-drift): People think training-serving skew is just data drift, and it is not. Drift is the world changing after the model was trained, so the live data slowly moves away. Skew is your own system building the same input two ways, one for training and one for live use. Drift needs new data and a retrain. Skew needs a bug fix.

![Four panels. A model trained in PyTorch is saved as one ONNX file and runs on a phone, a browser and a server. A big slow model becomes a small fast one. Live requests go to the old model and a copy goes to a new model whose answers are only logged. Training data and live data, side by side, with one input that does not match.](https://learn.benibix.com/diagrams/ai7-serving-path.svg)

---

Source: https://learn.benibix.com/term/ai/training-serving-skew
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-17
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
