AI & Data Best Practices
A Step-by-Step Approach to Running AI and Machine Learning Projects
Following a step-by-step repeatable, documented, and iterative approach will ensure project success. These steps provide an AI project template that provides guidance on each phase in development and provides an agile, flexible approach that allows you to iterate your project as needs change or you develop experience with AI systems.

Download the Step by Step Checklist for AI Projects
How many Steps are in the Stepwise Approach to AI and Machine Learning?
The CPMAI methodology for AI project success consists of six phases that are followed on each project iteration:- CPMAI Phase I: Business Understanding – “Mapping the business problem to the AI solution.”
- CPMAI Phase II: Data Understanding – “Getting a hold of the right data to address the problem.”
- CPMAI Phase III: Data Preparation – “Getting the data ready for use in a data-centric AI Project.”
- CPMAI Phase IV: Model Development – “Producing an AI solution that addresses the business problem.”
- CPMAI Phase V: Model Evaluation – “Determining whether the AI solution meets the real-world and business needs.”
- CPMAI Phase VI: Model Operationalization – “Putting the AI solution to use in the real-world, and iterating to continue its delivery of value:”
The First Step – Phase I: Business Understanding
The first step in any AI project is gathering an understanding of the business or organizational requirements. Successful AI projects make sure they are addressing a real problem with a solution that’s best suited for an AI approach. If you don’t know which problems are best suited for AI, you should check out our Seven Patterns of AI that shows how different applications of AI are best suited to solve different problems. In this step you should focus on understanding the project objectives and requirements from a business perspective, identify which patterns of AI you’re addressing, identifying which parts of the problem should be solved in a non-AI way, key performance metrics and measures, ethical guidelines and boundaries, and a preliminary, iterative plan designed to achieve the objectives. It’s important to remember that there is a solution to every problem, but AI is not the solution to every problem.During this first step of your project you should determine if AI is the right solution to your problem. If it is, then figure out what portions of the project require / do not require AI. With this solution in mind, you can outline your criteria for project success, understand potential data-centric needs, determine what skill sets are going to be needed for successful project completion, and other critical factors for project success.Second Step – Phase II: Data Understanding
Since AI projects are at their core data projects, a firm understanding of the data environment and availability is absolutely essential to AI project success. As such, the next step in an AI project is understanding the data needed to solve the problem with AI. The most important part here is understanding what data is required to address the business problem, whether or not that data is available, and what format it is in. In the Data Understanding phase of the AI project iteration, you should look to address three key data requirements for AI projects: the availability and sources of data to meet business needs, the quality of that data and need for enhancement or augmentation, and the environments in which data is needed for training and real-world inference. Determining what data is necessary to achieve your objectives laid out in step one, determine the quantity and quality of our data, what (if any) external data will be needed, as well as addressing ongoing data gathering and preparation.Third Step – Phase III: Data Preparation
The statement “Garbage in is garbage out” is a fundamental truth of AI systems. It’s astounding to realize how many AI projects fail because the data is not of sufficient quality to meet the business needs. By focusing on high quality data from the previously identified sources that meet business needs, you will have greatly increased your odds of AI project success. The third step in an AI project is Data Preparation. Once you have figured out what data you have, now you need to make sure it’s usable for your project. Included in this step are things like data cleansing, data aggregation, data augmentation, data labeling, data normalization, data transformation and any other activities for data of structured, unstructured, and semi-structured nature. This step addresses three key data preparation requirements for AI projects: wrangling data from the sources and transforming it to its required state, data cleansing to eliminate critical data flaws, and data augmentation and enhancement including data labeling to add necessary meaning and context to the data so that AI systems can properly learn from the data. In this step you should be addressing how your data needs be transformed to meet your project requirements, means by which data quality can continuously be monitored and evaluated, if and how you will be using and modifying third-party data, data labeling requirements, as well as creating necessary data engineering pipelines.Fourth Step – Phase IV: Model Development
With the right data prepared in the right ways to solve the right problems, we’re ready to build our first AI iterations that can solve our business problems. By the time we are ready to build our very first model you’ve already determined the business needs, the data requirements, and gotten the data in the right format and quality. If you haven’t, then you need to revisit these steps before getting here. In the fourth phase of an AI Project iteration, Model Development, you’re focused on selecting the right approaches and algorithms for the model. In this step it’s important to determine appropriate algorithm selection, settings, and hyperparameters. It’s also important to determine if you will use third-party models and/or extend those models. You also need to determine the performance of model training and model optimization activities, matching the model performance against business requirements, as well as selecting the appropriate infrastructure for model training. Key activities in this phase include model technique selection and application, model training, model hyperparameter setting and adjustment, model validation, ensemble model development and testing, algorithm selection, and model optimization. The first iterations of models should be quick and short enough so that a model is produced within the first week or two of each project iteration. This step-wise approach to AI takes an agile approach: you want to be able to do things in short, iterative sprints. Long iterations of AI projects that follow non-agile approaches are notorious for failing by spending significant time and effort in data preparation and model training only to realize that business requirements have changed, data sources have changed, or priorities have shifted because it took too long for your team to get anything out.Fifth Step – Phase V: Model Evaluation
Just like quality assurance and testing in the non-AI world, model evaluation is crucial to making sure that the AI solution meets the business needs. Without knowing if the AI solution adequately meets the needs setup in the first phase of the iteration, your AI project could fail. Once a model has been created to meet the business needs, it needs to be evaluated to make sure it performs according to the business requirements and other factors set in the previous steps of your AI project lifecycle. In the fifth step of this step by step approach to AI, your AI project you should be evaluating and testing your model, evaluating model performance measurement and improvement, and determining needs for ongoing model iteration. You should be determining if the model meets requirements for accuracy, precision, and other metrics, evaluating concerns on overfit and underfit of models, evaluating your models against business Key Performance Indicators (KPIs) as well ad determine means for model monitoring, iteration and versioning. Most importantly, the key consideration is determining if the AI solution you built is actually achieving the goals you set for yourself in the first phase. If not, you have the ability to iterate back to any previous step to figure out where things might have gone wrong before proceeding to the next step.Sixth Step – Phase VI: Model Operationalization
Once you know that an AI solution is indeed working and meeting the needs identified in the first phase of this step-wise approach to AI, you can put the AI solution into operation. However, in the world of data-driven projects, especially AI projects, there’s no such thing as “set it and forget it”. Real-world data continues to change which means your model performance will change over time as well. Organizations that haven’t budgeted time or resources for ongoing AI project maintenance realize quickly that their expectations of how the model will perform in the real world don’t meet the real world realities. The key needs to address during the Model Operationalization step include model deployment, model management, and model governance. Ask how this model be used in production / operational environments, determine the requirements for data flow for a model to be useful, set requirements for performance, and determine ongoing iteration requirements. This step also requires making sure to address model versioning and iteration, model deployment, model monitoring, model staging in development and production environments, and other aspects of getting the model in a position to provide value to meet the stated purpose.Implementing A Step by Step approach to AI
The step by step approach to AI outlined above is part of the well-established Cognitive Project Management for AI (CPMAI) methodology for AI project management. Individuals that have leveraged CPMAI training and certification not only have received the know-how on how to run AI projects using the proven step-wise approach detailed above, but also get a certification that shows their ability to master these steps.Despite decades of experience running major technology projects and with billions of dollars invested in emerging technology, organizations are still experiencing a high rate of failure for their AI projects. Some statistics show failure rates as high as 80%. It doesn’t have to be this way. Invest in yourself, invest in your team, invest in your project success and take a proven AI project management course. By adopting a robust, iterative step by step AI project plan approach as outlined above by which to reliably run AI projects and following a project plan for all your projects will automatically increase your chances of success. Download our CPMAI project checklist to give you guidance on the steps for AI project success.