This article is more than 1 year old

Before you sprinkle AI on all your analytics, check data quality

Gartner asks for some fundamentals in a slew of AI-in-analytics announcements

Following a fortnight festooned with analytics and data management announcements, Gartner has warned that users are not keeping pace with analytics vendors' fashion-following desire to inject just about anything they get their hands on with some AI serum.

Speaking as the curtain fell on global analyst company’s Data & Analytics Summit in London this week, Jason Medd, Director Analyst at Gartner, said despite the slew of announcements from Microsoft, SAP and Google, many customers still had to catch up in terms of data quality.

“Any shiny new thing that comes along, very often data quality gets dismissed, they try and implement it, they try and get some value, and… the quality issues start to creep in. There are so many ways bad data can creep into a system. People start to lose track of it as they keep chasing that shiny new thing,” he told The Reg.

Gartner analysts estimate that through 2024, half of organizations will adopt modern data quality technology to better support their digital business initiatives.

Earlier this week, Microsoft relaunched its analytics platform under the new name Microsoft Fabric, which encompasses a data lake called OneLake, Data Science, Data Warehousing, and Power BI. Microsoft promised that Copilot would allow users to create dataflows and data pipelines, generate code and entire functions, build machine learning models, or visualize results using conversational language.

Meanwhile, Google formed a partnership with enterprise software company SAP to bring together its Datasphere analytics tools with Google Cloud’s data and analytics technology, including its AI and machine learning (ML) models.

However, Medd said users were still struggling to find the right business case for analytics rollouts.

“The challenge is connecting the business use case [to the technology]. One of the things we did see a lot at the conference was people asking how to get value out of data. To a certain degree, it doesn't necessarily matter how fast the technology runs until you get that business case right and until you understand how you're going to get value from data,” he said.

It was also a challenge getting the right culture and awareness of business data as an asset, as much as the technology, Medd said.

Gartner has a four-step process it recommends for data quality. It starts with understanding which data influences business outcomes most, introducing data quality accountability, then validating data quality, and finally, integrating data quality into corporate culture. ®

More about


Send us news

Other stories you might like