top of page

Should Companies Use Off-the-Shelf Machine Learning Models?

There has been unprecedented amount of structured and unstructured data. More companies are contemplating and planning the use of machine learning (as part of artificial intelligence) to capture the value of data as a strategic asset.

Have you ever wondered about the use of off-the-shelf machine learning models such as Google Cloud, AWS? This question often arises during initial discussion of creating plans for implementing machine learning.

Machine learning allows companies to learn from data and automate decisions in real time without explicitly programmed and with minimal human intervention. This helps when companies can't have a simple (deterministic), rule-based code solution (read more here on machine learning).

Machine learning, machine learning systems, artificial intelligence
Machine Learning Systems (Source: SAS)

To save development time and effort, some providers provide off-the-shelf (machine learning) models. An off-the-shelf model is one that has been implemented by someone else (plug and play solution). Alternatively, there are also bespoke machine learning models. These models are uniquely built for your company's needs.

Deciding between off-the-shelf and bespoke machine learning models largely depends on your finances, state of data and the outcomes you expect to achieve.


Comparison Between the Models

A comparison between off-the-shelf and bespoke machine learning model is as follows:

As a summary:

  • Off-the-shelf machine learning models has lower accuracy and lower cost. The amount of data required to achieve modest accuracy can be surprisingly small.

  • Bespoke machine learning models provides high accuracy but at high cost. It is usually used where accuracy is paramount such as autonomous vehicle, healthcare.

Off-the-shelf Machine Learning Model

An off-the-shelf model is good as a fast solution and is usually used for common tasks such as face recognition or speech recognition. It is a tool for domain experts with limited data science or machine learning background.

It is sometimes used in the initial phase to guide the development of a bespoke machine learning model. It helps fill the talent gap and greatly improve accessibility to machine learning.

Machine learning, machine learning model, off the shelf machine learning, off the shelf
Azure ML is constructed by connecting datasets and analysis modules on an interactive canvas, and then deploying it (Source: Outsystems)

There are however limitations for off-the-shelf models. Since the model is developed on an external set of data, the output has lower accuracy/ precision when used on the company's own data.

Companies may also face the risk on relying on the black box without understanding the logic behind its decision-making that could risk resulting in reputational damage when things go wrong.

Companies should be aware that they may sometimes fall into the trap of getting things done quickly and at a lower upfront cost and then spend years fixing the solutions that didn't end up bringing the desired results. Usually off-the-shelf tools pay off for the short term.

Bespoke Machine Learning Model

Bespoke machine learning models are custom machine learning models. It crafts a customized solution tailored for the unique needs of each specific application. It is more accurate but takes a longer time and is costlier. It requires a large amount of data to train the models.

Ensemble Machine Learning Model

A technique that combines several base models in order to produce one optimal predictive model is sometimes used. This is called an ensemble model. It can increase accuracy. The decision tree model is the most popular and relevant ensemble machine learning model in data science today.

It should be noted that there are some areas where a 99% accuracy is not required. In these cases, companies are happy with a 70% accuracy where an average human won’t surpass that accuracy ever, and in the meantime, it gets to automate the process.


Companies now need to unlock value from big data to stay ahead of the competition. They need to decide between costs and benefits of off-the-shelf and bespoke machine learning models or an ensemble model. Creating the right balance requires in-depth thought on the required outcomes and the impact of the business.

What type/ types of machine learning models do you use? Share by leaving us a comment. If you require more information or expert advice to develop machine learning models, contact us. We want to be an extension of our clients. Subscribe to our newsletter for regular feeds.

Did you find this blog post helpful? Share the post! Have feedback or other ideas? We'd love to hear from you.



Emerj, What is machine learning?,, published 21 November 2019

Raconteur, Are off-the-shelf AI tools a good idea?,, published 12 May 2019

226 views0 comments


bottom of page