CIOs and IT leaders should follow three best practices to increase their success rate with digital commerce AI projects.
The age of artificial intelligence (AI) has already begun — at least in digital commerce. Gartner predicts that by 2020, AI will be used by at least 60% of digital commerce organizations. The abundance of customer-facing applications and back-end management tasks in digital commerce makes it fertile ground for AI to prove its business value.
We surveyed hundreds of digital commerce organizations worldwide — about 70% reported that their AI projects are very successful or extremely so, and three-quarters saw double-digit improvements in their measured outcomes.
AI technology thrives in digital commerce, because there is a huge amount of data available and a wide range of use cases that support critical business goals such as revenue increase, cost reduction and customer experience improvement. The most popular playing fields for AI projects are customer segmentation, product categorization and fraud detection, but there are many more.
Set up for success: Time and money management
Despite our surveyed success rate, AI is not a sure-fire success. CIOs and IT leaders must set a strict time frame — and make sure the project can be completed in under 12 months to showcase quick wins and secure C-level support for further implementation.
The key to reaching shorter time frames is to make each project small and simple, so it can either be completed fast or abandoned even faster without too many resources wasted. The more successful organizations in our survey did multiple projects of this type.
Alongside a time frame is budget attribution. Survey respondents spent on average $1.3 million in development for an AI project. While controlling development costs helps limit the scope of the project, getting the technology to deliver the best result should be the primary goal, and should not be compromised by the desire to contain costs. Hiring the right talent, obtaining powerful data management and processing tools, and integrating the AI application into the existing infrastructure — these should account for the majority of the development costs.
Overcome challenges: Build your data and skills base
CIOs and IT leaders in digital commerce struggle with two major challenges when it comes to deploying AI projects: lack of quality training data and lack of in-house skills.
While a mindful budget allocation provides relief, organizations should also look to outside vendors for support. Assess whether there is sufficient AI talent inside your organization to develop a high-performance solution yourself. If not, consider looking at commercial AI solutions. Those have already been tested and the vendors have techniques to deal with data issues.
A commercial solution does not mean that you can skip the search for talent. Integration into your existing application infrastructure — the third big challenge — is a task for a skilled in-house team. There may not be ready-built connectors to the commercial AI solution, which would complicate the integration process. Leaving this too late could mean your initiative stalls at the pilot phase.
Take the lead, but don’t rush ahead
While AI technology is becoming more and more mainstream, there is still a lot of skepticism and sometimes aversion toward it among employees. CIOs and IT leaders cannot and should not ignore those emotions, as organization wide support is vital for shaping AI-friendly business processes.
Employees need time to learn to trust AI and learn how it can augment their work. Explain, where possible, why the solution makes certain suggestions and show the effects of AI-made decisions on key performance indicators. Provide training for employees and encourage them to add their expertise to the continuous improvement of models.