6 Tips by OCTAVE on Delivering Advanced Analytics Projects Successfully
According to Gartner, 85% of advanced analytics and data science projects fail to deliver commercial value or see no business adoption as expected. Being successful at data science in a business setting is a complex challenge. This series of articles will cover ‘how to deliver an advanced analytics and data science project successfully’. We will also evaluate the key elements of a successful data science project across multiple domains and look at what constitutes a ‘successful’ data science project.
The purpose of a data science project in a business must be to solve a business problem, and not ‘find an insight’ from a dataset. How do we do that? The best way to identify the business problems are from the subject matter experts (SMEs) who handle these functions. In other words, ‘Functional Experts’. In future articles we also intend to cover steps involved in ideation to independence and deep dive into each stage to understand the critical success factors.
1. Don’t start without understanding the 3W’s (What, Why and Who)
Be firm upfront and set yourself up for success before investing time in a technical solution. Gather information on the 3W’s and brainstorm with the SME’s to understand better insights. By doing this you will be able to approach a Data Science project to foster ROI and better adopt it within the business.
2. What are you trying to solve?
It sounds obvious, but be certain you have a clearly defined problem statement. Arriving at this would be a responsibility of the data science team and SME’s. The SME’s identify the primary challenge and the analytics function act in a consultative manner to brainstorm and help define the problem within it which a data science team can find a solution. Identifying problems that data science can resolve is the primary objective to foster maximum ROI.
3. Why solve it?
It is always better to define the desired outcomes and quantify the benefits by solving the problem.
A helpful method here is to estimate the cost of getting it wrong versus the benefit of getting it right. By doing this, we can maximize value creation and make the data science team drive outcomes for better incremental value creation.
Setting clear goals upfront also provides an opportunity to manage the expectation of business users on the outcomes the project can deliver, which should increase user adoption in the long run and help avoid disappointment when unreasonable expectations are not met.
4. Who is your Sponsor/ Champion?
Having an executive business sponsor who understands the business problem and wants to improve it through analytics driven insights is vital. Without a champion who is an SME, it is difficult to develop a relevant solution with the required time and financial investment, thus making the project (and subsequent adoption) an uphill battle from the outset!
It is also important to agree to a process for measuring benefit with the project owner. This provides a clear method of evaluation within the project and gives them a mechanism of reporting success up the value chain. Having a business sponsor that promotes the data science use case, especially in monetary terms, has a catalytic effect on inducing adoption within the wider business.
5. Iterate, Collaborate, Appraise
Deliver agile. You need to deliver fast, fail fast, and thrive faster.
Hold regular problem-solving sessions with the SME’s and functional experts throughout the delivery process. Their expertise is crucial in giving direction to data interrogation and it keeps engagement and understanding high and increases the likelihood of a sustained long term adoption of a data science project.
Iterate continuously and evaluate the data science model’s effectiveness, both in statistical terms as well as using historic data (control and test sets) to compare predictive accuracy against the target outcome defined upfront. Gauge it against the expected business value realization.
Pilot, Pilot and Pilot. There is no substitute for test, control and learning iteratively. Piloting enactments of the model, using things such as A/B tests, to understand real-world effectiveness and attribute expected ROI. This allows you to understand whether you are getting the maximum out of the expected deliverables and makes sure the project sponsor’s expected value creation is achieved.
6. Model Management. It does not end post-delivery!
Part of the delivery is to understand the change management required and it is a crucial step in entrenching data science solutions within a business and driving long term adoption for maximum ROI and value creation.
Make sure you iterate the model training on regular intervals. Set up triggers to gauge model performance. Keep in mind that lack of support when solutions are implemented will simply result in a lack of adoption, compromising maximum value realization. This also provides a forum for feedback and nurtures an environment for collaboration between the business and data science functions.
Follow this series for more interesting tips and facts on streamlining a productive and functional delivery of data and advanced analytics projects. The second article in this series will be published next month.
Written by Duminda Jayathilake, Analytics Delivery Lead at OCTAVE.
Further Reading: https://www.techrepublic.com/article/85-of-big-data-projects-fail-but-your-developers-can-help-yours-succeed/
