How Data Quality Is Limiting Your Organisation’s AI Success
Posted: Monday, Apr 15

i 3 Table of Contents

How Data Quality Is Limiting Your Organisation’s AI Success

Too often, having strong data processes can be seen as a ‘nice to have’. In our day-to-day work, it can feel like the job is done when the file is neatly labelled and saved in the right project folder. And yet, getting Information Management (IM) right is so much more than that. We’re now living in a world where holding onto too much data could get you into hot water in the event of a breach, at the same time, not keeping the right data can cause a major regulatory headache.

Now there is an added complication: AI. Your AI tools can only be as effective as your information management processes allow them to be. Take Microsoft Co-pilot for example. Whilst Copilot will change the way we work, if an organisation’s data isn’t ready, it’s going to fail.

AvePoint’s inaugural AI and Information Management Report, which surveyed over 750 digital workplace leaders across the world (of which 20% were from APAC), found nearly every organisation experiences challenges during artificial intelligence (AI) implementation (95%), especially when we’re talking about the quality of an organisation of internal data. In fact, organisations with more mature information management strategies are 1.5x more likely to see benefits from AI than those with less mature strategies.

The State of Our Data

This year, organisations are significantly increasing their investments in AI, with 83% planning to increase their AI spending, 79% investing in licensed AI like Copilot for Microsoft 365, and 60% allocating at least a quarter of their technology budget to AI in the next 5 years. Despite these investments, the survey shows a limited understanding of just how much poor IM can impact the successful implementation of AI in an organisation.

Despite organisations with more mature information management strategies being 1.5x more likely to see benefits from AI, data strategies are not the highest priority for organisations, with only 17% of organisations today believing a robust strategy is the most effective way to ensure ROI on their AI investments.

The survey also found that there is a gap between how effective organisations believe their IM strategies to be and how effective they actually are. For example, despite the growing volume of data, just 29% use automation in most aspects of their IM strategy today. This gap is the largest hurdle to safe and successful AI implementation.

Interestingly, 80% of organisations believe their data is ready for AI, but 52% faced challenges with internal data quality and organisation during implementation, whilst 77% acknowledge they must implement new IM measures to ensure the accuracy of AI input and output, especially as data output increases. It’s a state of ‘data delusion’ where organisations believe they are ready for AI, but their data and information is telling a different story.

Organisations might not realise it yet, but data unpreparedness is going to hold back innovation and growth.

AI Paralysis Causes Delays

Beyond data readiness, organisations are fearful about the implications of AI. The survey found less than half of organisations (47%) lack confidence they can use AI safely. Further to this, 71% are concerned about data privacy and security when implementing AI.

It’s interesting to note that despite these fears, less than half have an AI Acceptable Use Policy, despite widespread use of public generative AI tools (65% of organisations use ChatGPT and 40% use Google Gemini today). In addition, less than half of organisations (46%) offer AI-specific training, hindering their employees from safely using and optimising this technology.

Worryingly, 45% of organisations encountered unintended data exposure while implementing AI. AI has seemingly put many organisations into a state of confusion, where they are scared of the consequences yet not proactively taking the right steps to mitigate risk when it comes to their data.

This deer stuck-in-the-headlights approach to IM is slowing down the successful adoption of AI and exposing organisations to risks that could be easily mitigated if they knew how to better protect and govern their data, and give their workforce the right tools to support new policies.

Commenting on the research, Tori Miller Liu, President and CEO, Association for Intelligent Information Management (AIIM) said: “The results of this study confirm what we’ve known to be true, that without proper IM strategies you can’t possibly succeed with AI.”

Tori is right. There is no path toward successful AI implementation without robust IM, and failing to get it right could have a detrimental impact on the public’s perception of AI. All we need is one major security incident to set the entire industry back. The denial as to how important IM is to successful AI implementation needs to be replaced with strong guardrails, employee training and a better understanding of effective IM strategies.

Alyssa Blackburn
Alyssa Blackburn is the Director of Information Management at AvePoint, where she helps organisations achieve value from their information and records. With nearly 20 years of experience in the information management industry, Alyssa has worked with both public and private sector organisations to deliver guidance for information management success in the digital age. A passionate information management professional, Alyssa is actively involved in the industry and is an in-demand speaker at conferences and industry events worldwide. She frequently contributes to industry publications and will happily talk for hours about how to modernise a retention and disposal schedule and classification scheme.
Share This