Just like humans need a good diet to thrive, AI needs good data. The wrong kind of data and poor hygiene can derail enterprise AI ambitions before a project even starts.
As CIOs and their executive peers feel the pressure to push initiatives forward, they first need to assess the resiliency and strength of the data strategy.
“You can use the vanilla version of generative AI or you can train your large language model,” Roxane Edjlali, senior director analyst at Gartner, told CIO Dive. “But in all situations, the only differentiation you have is the data. For you to deliver the most on the differentiation, you need to make sure that the data is ready.”
Technology leaders and market analysts suggest CIOs and their enterprises ask themselves a battery of questions to make sure strategies produce the desired results:
- What do you want to accomplish and how does that align with business objectives?
- How well do you understand where your data is?
- How open does the enterprise want to be with its data?
- How should data be accessed?
- How is data currently organized?
- Where is the data coming from?
- How do we ensure privacy and security?
- How is data quality measured?
- Who will work on data clean-up?
- How will success be monitored?
The stakes are high for organizations to get data strategies up to par.
Poor data frameworks lead to poor performance and outcomes. Around two-thirds of decision-makers believe no one within their organization understands data collection or access, according to a SoftServe report. Nearly 3 in 5 business leaders say inaccurate or inconsistent data informs key decisions.
Not only is poor data management hampering ambitions, these deficient practices are eroding trust and increasing costs, a Semarchy report found. Organizations that advance without first addressing data woes are only prolonging their pain.
Vendors are positioning themselves to help enterprises clean up their data strategies. Tim Guleri, managing partner at early-stage venture capital firm Sierra Ventures, said established vendors are looking to partner with AI startups to entice enterprise customers.
Large vendors have distribution mechanisms and a built-in customer base, but they often fall behind startups when it comes to innovation. Meanwhile, startups are typically ahead of the innovation curve but must work harder to gain enterprise buy-in. Alliances between the two are likely to become more common.
“That’s the risk mitigation that large companies are looking for,” Guleri said.
Still, getting data strategies on the right track is hard work, no matter who’s helping out.
“There is almost no shortcut in taking care of your data quality, foundation and connectivities,” Shiyi Pickrell, SVP of data and AI at Expedia Group, told CIO Dive. The travel company, which operates Vrbo, Orbitz and Travelocity, put considerable effort into merging the different brands’ data lakes and assets.
“We went through a very hard work phase to bring the data together,” Pickrell said. “We truly did the hard work, the less sexy work, first to get our data connected and then get our data quality up and our data availability up so we can truly leverage the newest things like generative AI [and] LLMs.”
