Skip to main content

More than ever before, COVID-19 has ensured the rapid adoption of digital technologies such as Artificial Intelligence (AI) and analytics as key to business transformation and indeed their survival. Analytics and AI are the No.1 and No.2 top game-changing technologies that emerged stronger from the COVID-19 crisis, according to a recent Board of Directors survey by Gartner. About 7 in 10 board directors said they had accelerated digital business initiatives due to COVID-19.

One of the key takeaways from the recent National Retail Federation 2022 show was that retailers are looking at technology to support change. AI helps businesses better understand and serve their customers, empower their employees, improve security, and make business more efficient. According to industry reports, AI will create over $3 trillion in business value.

Yet, about 85% of AI projects fail to deliver on their intended promises to business, according to a white paper released by Pactera Technologies. Just one out of ten data science projects actually make it to production. Here we list the top three reasons why AI projects fail and what you can do about it in your enterprise.

1. Lack of Integration

Businesses need to align their automated data capture process with AI and machine learning to provide context and improve interactions between customers and the company. These insights, however, have to be available in real-time to help users make purchase decisions when on a website or in-store.

Consumers have often turned away after researching a product because while they intend to purchase, the products are not displayed according to their preferences or are not happy with the information or service. Identifying these in real-time and using AI to build the back-end will be essential to emerging as a business leader.

Companies should also evaluate the required infrastructure to deploy AI at scale, considering that these are aligned with long-term business goals. The divide between data and the decision-making process needs to be bridged. One way for businesses to overcome this barrier is to adopt a problem-first rather than a technology-first approach.

2. Lack of Business Orientation

One of the fundamental mistakes of AI integration is all too familiar humans’ fallibility of going for the next fancy thing. Often companies want to be the first in using technology but have no apparent use of it either for their customers or employees. Such technological orientation that did not start with a business problem will not help find a business-oriented solution.

Business orientation will also help bridge the communication gap between directors and IT or data science teams. Lack of leadership and support is often cited as a significant barrier to the failure of AI powering an enterprise.

Business leaders sometimes fail to understand that the AI implementation chain stretches across business functions. Enhancing one department’s capacity without aligning other departments creates data and decision gaps resulting in less than optimum outcomes apart from increased costs and delays. Entreprises should invest in upskilling their in-house talent or contract a trusted service provider on top of the changing technological prowess.

3. Lack of talent

A survey of 2,000 IT executives and professionals by International Data Corporation found that while a majority of IT leaders said that the AI infrastructure was the most critical decision, a majority of AI projects were reported to be still in infancy. Many businesses have either deployed or are in the process of mapping a data strategy. Still, more than half of large companies and more than three-fourths smaller companies say they don’t understand the data infrastructure necessary to deliver AI use cases.

Most businesses do not have enough employees that are adept at the know-how to ensure that AI deployment is successful and at scale. In some managers’ minds, AI is a plug-and-play, seamless and easy deployment. But ‘fail fast and fail early’ is as valid for AI integration as any business strategy.

Investing in upskilling in-house talent and appointing a leader for a multidisciplinary team will ensure you build an AI team with a long-term vision. A successful AI implementation team includes a data scientist, a data engineer, a developer to test, and a UI developer to present business insights. These insights then need to be implemented by managers. When consulting with vendors, ensure they add to the existing capabilities of your in-house team and business needs.

Business leaders should also note that even if all the resources are available, a lack of communication and coordination may result in teams not being aligned with business goals, resulting in increased costs but inferior outcomes.

So far, businesses have been competing for data. But the competition now is who can better integrate data into strategic decisions and implementation. AI works best in concert with other advanced analytics.

Enterprises are prioritizing AI and machine learning over other IT initiatives in 2021. Results from Algorithmia’s third annual survey, 2021 Enterprise Trends in Machine Learning, showed that 76% of enterprises prioritize AI and machine learning (ML) over other IT initiatives in 2021. Yet successful AI implementation relies as much on technology as it is on a company’s culture of innovation and change.