Data analytics has long been a staple of the business world, but now AI is making it better and stronger. Non-technical stakeholders are growing accustomed to using generative AI chatbots to ask questions about business data. Advanced analytics engines that can handle enormous datasets and deliver faster insights are becoming table stakes, with scalable, intuitive solutions bringing these capabilities to companies of every size.

The next frontier here is in embedded analytics, says Avi Perez, cofounder and CTO at Pyramid Analytics, whereby people can access advanced decision intelligence capabilities within any app they’re using. If that sounds familiar, it’s because the concept of embedded analytics has been around for a while. However, as Perez explains it, the new generation of embedded analytics is an entirely different proposition.

“In the old days, embedded analytics was typically a watered-down, web-based implementation of the full-blown analytics tool,” he says. “The newer version of embedded analytics is done through what we call injection, which delivers a super-fast, super light, very effective set of capabilities that are pivotal to the way the hosting page operates.”

Once you include generative AI functionalities, it becomes possible to submit queries using natural language, and open up access to users who aren’t familiar with data science or particularly expert at analytics.

But Perez is quick to point out that the new generation of embedded analytics is in its infancy. Many benefits are still on the way, and several challenges need to be overcome before embedded analytics can fulfil its potential.

Embedded analytics are accessible analytics

The idea behind embedded analytics is to negate a great deal of the friction around data insights. In theory, line-of-business users have been able to view relevant insights for a long time, by allowing them to import data into the self-service business intelligence (SSBI) tool of their choice.

In practice, this disrupts their workflow and interrupts their chain of thought, so a lot of people choose not to make that switch. They’re even less likely to do so if they have to manually export and migrate the data to a different tool. That means they’re missing out on data insights, just when they could be the most valuable for their decisions.

Embedded analytics delivers all the charts and insights alongside whatever the user is working on at the time – be it an accounting app, a CRM, a social media management platform or whatever else – which is far more useful. “It’s a lot more intuitive, a lot more functional if it’s in the same place,” says Perez.

“Also, generally speaking, the people who use these types of business apps are non-technical, and so the more complicated you make it for them to get to the analysis, the less of it they’ll do.”

Expectations are high

So far, so impressive. But Perez emphasises that there are a number of barriers to embedded analytics utopia. Businesses need to bear these in mind as they seek to develop their own solutions or find providers who can deliver them.

First of all, the technical requirements are high. To fit today’s suite of business tools, embedded analytics have to be extremely fast, lightweight, and very scalable, otherwise they risk dragging down the performance of the entire app.

“As development and the web moves to single-page apps using frameworks like Angular and React, it becomes more and more critical that the embedded objects are lightweight, efficient, and scalable. In terms of embedded implementations for the developer, that’s probably one of the biggest things to look out for,” advises Perez.

On top of that, there’s security, which is “another gigantic problem and headache for everybody,” observes Perez. “Usually, the user logs into the hosting app and then they need to query data relevant to them, and that involves a security layer.” Balancing the need for fast access to relevant data against the needs for compliance with data privacy regulations and security for your own proprietary information can be a complex juggling act.

Integration is challenging

Additionally, the main benefit of embedded analytics is that it makes insights easily accessible to line-of-business users. “It should be very easy to use, with no prior training requirements, it should accept and understand all kinds of requests, and more importantly, it needs to seamlessly work on the company’s internal data,” says Perez.

This usually means integrating GenAI, so that non-techie users can ask queries in their natural language and receive responses that they can understand independently. There’s no question that this expands accessibility enormously, but it also ramps up complexity, says Perez.

“It can be a tremendously complicated exercise, because not only the functionality needs to be surfaced, but whatever is requested by the user, through a chatbot for example, needs to be piped through the embedded analytical engine sent to the underlying platform, which has to resolve it and then come back with a result into the hosting app,” he explains.

“So in other words, it brings in a very complex set of functionalities that need to be exposed in the host application through the embedded app, and that is a tremendously complex process if it needs to be light, fast and efficient.”

Customer-facing apps take it all up a few notches

Perez goes on to explain that security, access, and performance are all easier propositions when the embedded analytics solution is intended for in-house use only.

In these cases, your business owns all the domains, there’s no exposure to the wider internet, and you can control the client environment. “It’s actually a lot more achievable to take shortcuts in how you embed, so you’re less stressed about it,” Perez says.

However, the whole endeavour is more difficult when you’re building a customer-facing app, which is the prevailing trend. “Now it’s got to look good, work with all this magical functionality, with very secure, very high performance,” he explains. 

“And the entire stack needs to be bundled up and be fully operational, very efficient, very fast, and magically appear on someone else’s website without anyone lifting a finger and without requiring a two-year development cycle. It’s a tremendously high bar of functionality that needs to be met.”

The future beckons for embedded analytics

GenAI offers a whole new set of capabilities and functionalities for embedded analytics. Enterprises can look forward to a more democratic data world, where relevant insights are an inherent part of every business user’s workflow. There are hurdles along the way, but with the right partners and careful planning, intuitive embedded analytics can become the norm.

Related articles

What generative AI means for business analytics – Jim Goodnight, founder and CEO of SAS Institute, tells Information Age his thoughts on the impact generative AI will have on business analytics

Tony McCandless – The role of generative AI in intelligent automation – Tony McCandless, UK, Ireland and Benelux managing director at SS&C Blue Prism, spoke to Information Age about the keys to intelligent automation leadership, and the current generative AI trend