Barriers to retail AI adoption

The research also reveals that the majority (69%) of retailers plan to adopt these technologies within the next 12-24 months, signalling a growing recognition of AI's potential to revolutionise retail operations. However, retailers also share a range of significant barriers in adopting predictive AI/ML technologies, including:

· Data preparation challenges: Almost half (43%) of respondents highlighted the complexity of preparing data for AI/ML models. Retail data often resides across disparate systems not originally designed for the nuanced demands of AI training, complicating the data preparation process.

· Talent and knowledge gaps: 41% of retailers pinpoint a lack of in-house expertise in AI/ML as a significant barrier. Despite the emergence of new courses, open-source models and AI-first vendors, the pace of evolution in the AI space demands continuous learning and adaptation.

· Lack of executive support: Over a third (35%) of respondents cite the absence of executive backing as a critical challenge. However, this is expected to change as success stories proliferate and confidence in AI technologies grows.

In addition, the research also highlights a gap in historical data, crucial for training predictive AI models. Only 40% of retailers have records of stock quantities at each location when orders were placed, while only 36% track stock allocations across sales channels. This lack of historical insight into inventory levels, combined with gaps in order processing and delivery data, poses significant challenges for AI implementation.

When looking at fulfilment and delivery challenges, over two-thirds (68%) of respondents do not have location status data, such as whether a location was available to fulfil orders or not (due to weather, power outage, maintenance, etc.). Furthermore, 75% do not have historical data when it comes to average order processing time, which is a critical factor for optimising fulfilment and on-time delivery.

Improving customer experience is hindered by a lack of historical shipping data, with most retailers missing crucial information on order fulfilment attempts, delivery times, costs and carriers. This deficiency in data prevents accurate delivery predictions and sourcing optimisation, impacting three-quarters of retailers who lack insights that can help optimise their use of AI and ML applications.

Despite these challenges, the potential benefits of AI for retailers are immense. The survey highlights increased inventory returns, improved first-attempt delivery success rates and reduced markdowns as the top advantages. Specifically, over 62% of fashion and apparel retailers identified 'reducing out of stocks' as a key benefit, underlining the impact of AI on operational efficiency and customer satisfaction.

"Utilising Predictive AI in retail, especially for enhancing backend operations, poses challenges but promises significant rewards,” commented Nicola Kinsella, SVP of Global Marketing at Fluent Commerce. “Retailers aiming to capitalise on AI must identify their target business issues and gather a robust dataset, ideally spanning 2-3 years, to effectively train AI models. By carefully selecting or developing AI models that align with their strategic goals, retailers can achieve a competitive advantage that translates into operational efficiency and improved customer satisfaction. The process requires effort, but the benefits – ranging from better inventory management to enhanced delivery success – are potentially highly significant."