Why Data Strategy Is Key to Unlocking AI

If you’ve identified your data sources, determined where they’ll be used and which end users will consume the output, have you finished your data strategy? Not by a long shot. Your data strategy is only getting started.

Amazon Web Services ?states that data strategy is “a long-term plan that defines the technology, processes, people, and rules required to manage an organization’s information assets. […] A data strategy outlines an organization’s long-term vision for collecting, storing, sharing, and usage of its data.”

That’s a bit more than the basic data roadmap we described above. With all the rush to fully utilize AI and become data-driven, why should businesses take the time to design and implement a full data strategy? And what questions should they ask (and answer) along the way?

That’s what we’ll discuss in this article.

Why Data Needs Strategy

It’s no secret that AI relies on data. However, many organizations that are interested in using business AI don’t fully understand the impact that data quality and the data pipeline in general can have on the performance and accuracy of AI tools.

An HBR/Cloudera report titled?Data Strategy: The Missing Link in?Artificial Intelligence-Enabled Transformation ?shared some interesting statistics about AI and data strategy:

• 92% of surveyed organizations are working with AI.
• 61% do not have a data strategy in place.

The same report quoted Nitish Mittal, vice president of digital transformation for the Everest Group, as saying “Data or the lack of the right data strategy is the number one bottleneck to scaling or doing anything with AI. When clients come to us with what they think is an AI problem, it is almost always a data problem.”

If that’s not an argument for data strategy, I don’t know what is. Organizations plan out sales, marketing, and advertising campaigns with care; they painstakingly develop HR guidelines and financial processes. Should a tool that’s being used to streamline and improve decision-making get any less forethought?

How Data Strategy Impacts Business AI

Without the right kind of data, delivered in a timely manner and processed with tested and monitored AI models, the efficiency of any AI tool plummets.

However, when AI is supported by a well-thought-out data strategy, it can produce accurate and actionable results. Furthermore, it can be delivered promptly to the people who need it. This not only allows them to get on with their analyses (without waiting for assistance from IT or data support), it also alleviates some of the strain on overworked data analysts.

Having an easy-to-use, highly functional AI solution also spurs its adoption, as business users aren’t stuck trying to work with a tool that doesn’t meet their needs or produce reliable results.

In short, the right data strategy can supercharge AI. It can make it more relevant, more usable, and more reliable. This can also overcome some of the adoption and usage roadblocks associated with scaling AI across an enterprise. So, how can you create a robust data strategy?

Questions to Ask When Building a Data Strategy

Most organizations will rely on internal or external experts to develop their data strategy. But along the way, make sure that everyone involved knows the answer to questions like these:

• What use cases will our AI support? Who will be using the AI solution, and for what purposes?
• What use cases will be the most impactful? Which will show quick results and can be leveraged as pilot projects?
• What data sources can we use? Will they meet our needs?
• What quality of data is available? What standards and metrics will we use to determine data quality?
• Where is our data coming from? How does this affect its accuracy or bias?
• What compliance and legal regulations do we need to meet? What risks might we face by using this data (e.g. personal information, sensitive information, etc.)?
• What data skills do we have in our organization? Can we train or outsource additional needed skills?
• Are our current IT and data architectures able to support this initiative? If not, what changes need to be made?
• How will we store, clean, process, and transfer data? How will we deliver insights to end users?
• How will data be processed by our AI – i.e. which models will be used, and for what purpose?
• How will we ensure data governance and security? How will we control user access?

And this is not a complete list of topics; it’s more of a starting point. Your actual needs will depend on your industry, the data you process, and your AI application.

Finally, remember to review your data strategy regularly during and after your AI implementation. Some experts recommend a review every six months, while others opt for one whenever there’s a significant change to the data source, pipeline, usage, or solution.

If you want to succeed at business AI, data strategy can’t be left to chance. Like any other important aspect of your business, it must be planned with care. And it will certainly repay you in terms of the results of your AI initiative’s accuracy and usefulness.

Authored by:?Anil Kaul, ?CEO at Absolutdata, an Infogain Company and Chief AI Officer at Infogain and?Anil Joshi , Vice President at Absolutdata-an Infogain company (add LinkedIn account Links)