Many companies have set goals to achieve data analytics at scale. They’re understandably excited about equipping their employees to pull more customized insights from their data. So why do few companies ever get there? 

The first issue is that analytics at scale can mean a few things. In large enterprises, the goal might be to scale analytics into the hands of every person in the business, which could total hundreds of thousands of people. At this scale data catalogs become a key consideration so that people can find the data they need and get analytics for it. Without robust data catalogs, companies can find it difficult to discover, access, share, and secure their data.

If the analytics are coming from a central team, one of the biggest challenges in scaling data is how you get the right analytics to the right people. Even in a company with only two dozen employees, one dashboard will not meet everyone’s needs, and customization can take weeks or months. On the other hand, giving everyone the same kitchen sink of dashboards in a company with hundreds or thousands of people is naïve and less than ideal. 

AI in Data Analytics

Although it’s popular right now to look reflexively to AI to manage analytics at scale, data leaders shouldn’t overlook complementary approaches, such as data mesh architecture. In a nutshell, this means distributing your analytics team across the organization so different members can focus on different parts of the business. Although data mesh does help to get analytics to the right people at the right time, however, it often leaves out customizing the data to the level end users may want.

It's in these scenarios that generative AI can be useful. When it’s built on the right data sources, generative AI can understand whatever you ask it, even if it’s to fill out a Jira ticket to customize a data set. Multiply the number of customization requests by the number of employees in a large enterprise and generative AI starts to make intuitive sense. 

Small scale to large scale

When scaling analytics, as with nearly any IT program, starting small with a pilot execution makes logical sense for generative AI. Flashy demos can create an inaccurate perception that generative AI is naturally super intelligent and any chatbot can answer a user’s sophisticated inquiries about their business data right out of the box. Not to mention that people often ask bad questions.  

What companies tend to discover is that natural language query for data is one of the toughest challenges in AI, so handing a data leader a few tools and a GPT to figure it all out is not a formula for success. 

Taking a business use case approach tends to be far more successful. In this regard, users are smarter about generative AI and how it can be deployed for analytics than they were even six months ago. Here are a few examples of popular use cases for analytics at scale.

Marketing. Marketing is a very data-heavy space, which often means teams of non-technical users who have lots of new questions that they need answered for their research; for example, identifying a new audience based on online intent data.

Supply Chain. With millions of IoT sensors sending data from machinery and shipments, the need to analyze and optimize supply chain processes is great. Analytics also can be helpful to better predict inventory needs and suggest improvements in operational processes.

Cross-Business Unit (Cross-BU). Some enterprises are looking to scale their finance office, their procurement office, and the office that’s in charge of maintaining inventory in order to build something that can scale to multiple teams. Data governance becomes a particularly important factor in this example. 

This last example is particularly exciting because it is where analytics at scale becomes collaboration at scale. If we can find ways to join different data sets together or create governed workspaces where the departments can safely work together with some shared data, AI can play a central role.  

Analytics at scale: how to get started

If I were to advise a data leader on how to ramp up analytics in their organization, I’d suggest a few movements.

  1. Figure out the foundation. How do you want to use generative AI? Some camps of companies want to make their AI data ready before they even touch generative AI, while others don’t think it has to be a precursor. Depending on the direction you want to go, there are different ways to implement generative AI. But start by figuring out what kind of foundation you want to build. 
  2. Get AI educated. Some companies have a handful of resources well versed in generative AI, though what’s even better is a Center of Excellence for AI. This approach gathers the research and shares the knowledge across users, so more people stay informed about how generative AI works.
  3. Walk don’t run. Eagerness to scale up analytics is fine, but that shouldn’t mean blindly rushing into it. Everyone does have FOMO (not to mention FUD) about AI, but that's not an excuse to roll it out recklessly. See how it works with a small group, then learn from that as you roll it out to larger ones. 

Analytics at scale is a very powerful idea and one of the brightest horizons for generative AI. With the right approach and a touch of realism, there’s no reason that most companies can’t get there.