Published on the 18/10/2018 | Written by Jonathan Cotton

Gartner finds majority of AI projects are very successful – for e-commerce…

Outsourcing the hard stuff is key to success when it comes to AI-based digi-commerce projects according to new research from Gartner, which surveyed 307 digital commerce organisations currently using or piloting AI in their digital commerce work.

The paper, How to Increase Chance of Success for Digital Commerce AI Projects, finds 70 percent of the digital commerce organisations surveyed across Australia, New Zealand, the US, Canada, Brazil, France, Germany, the UK, India and China say their AI projects have been ‘very or extremely successful’. Furthermore, three-quarters of respondents said they are seeing double-digit improvements in customer satisfaction, revenue and cost reductions – at 19, 15 and 15 percent, respectively.

“It seems the commerce organisation is more advanced in leveraging emerging technologies.”

Too good to be true? After all, aren’t such AI applications still bleeding-edge tech?

Well yes, but Sandy Shen, research director at Gartner, says that, in this instance, it’s simply a case of horses for courses.

“Digital commerce is fertile ground for AI technologies thanks to an abundance of multidimensional data in both customer-facing and back-office operations,” Shen says.

“Digital commerce organisations see much higher adoption of AI than average organisations. Our survey shows that AI has been adopted by only four percent of all enterprises, while 34 percent of commerce organisations have deployed/piloted AI. So it seems the commerce organisation is more advanced in leveraging emerging technologies.”

That’s set to increase says Gartner, with the prediction that by 2020, AI will be used by at least 60 percent of digital commerce organisations and that 30 percent of digital commerce revenue growth will be attributable to AI technologies.

The most promising use cases? Customer segmentation, product categorisation and fraud detection.

The challenge however seems to be in the implementing. Using AI to solve business problems requires talent and technology expertise to identify the right algorithm and model features for the target problems and such skills are scarce assets in today’s AI-hungry world.

“Many organisations won’t have such skills in-house and will have to hire from outside or seek help from external partners,” says the report.

According to those surveyed, a lack of quality training data (29 percent) and a lack in-house skills (27 percent) are the top reported challenges in deploying AI in digital commerce.

Outsourcing the grunt work, in this instance at least, might be the order of the day: While, on average, 43 percent of respondents chose custom-built solutions developed in-house or by a service provider, 63 percent of the more successful organisations chose commercial AI solutions.

“When organisations realise the importance of AI in terms of the profound impact on how they organise and work, and the business outcomes, they are ready to invest in AI,” Shen tells iStart.

According to Shen, successful implementations are often the result of enterprises working on multiple use cases in parallel so they can leverage the lessons learned across them.

“They integrate the AI solution with the technology infrastructure and adapt business processes so their employees can use it on a daily basis. This helps you get the most of the investment.

“But the mindset is the most important in getting ready to embrace AI. Then you deal with the details – hiring the talent, identifying business problems, preparing the data, looking for the vendor etcetera”.

Fundamentally though, AI implementations are still complicated things says Shen, and success can often be a case of simply biting off only as much as you can chew.

“Organisations looking to implement AI in digital commerce need to start simple,” says Shen. “Many have high expectations for AI and set multiple business objectives for a single project, making it too complex to deliver high performance. Many also run AI projects for more than 12 months, meaning they are unable to quickly apply lessons learned from one project to another.”

“Solutions of proven performance can give you higher assurance as those have been tested in multiple deployments, and there is a dedicated team maintaining and improving the model”.