This past year has seen the technology of embedded analytics — the inclusion of data ingestion, analysis and visualization capabilities within business applications — begin to leverage other growing technologies to improve the accuracy and scope of reporting, as well as make it easier for developers, and even non-developers to start including such capabilities in their software.
According to Forrester’s Deep Learning: The Start Of An AI Revolution For Customer Insights Professionals report released in September, advancements in deep learning have improved the accuracy of speech, text and vision data ingestion, allowing for platforms that implement these technologies to “extract intent, topics, entities, and relationships.” All of this is important data for companies that would like to know what their customers’ needs are, and adjust properly.
But before businesses start to apply advanced features into their embedded analytics solutions, Brian Brinkmann, vice president of product marketing at Logi Analytics, explained they need to make sure they’re ready at a more basic level first.
“What we have seen is when people are building applications, they have a set of requirements, they go build out those requirements, and when they roll it out to their customers, their customers have an additional set of requirements that maybe they haven’t thought of,” Brinkmann said. “The application’s end-users say ‘I want to build something.’ or ‘I want to be able to enhance the application.’”
According to Brinkmann, finding ways to support and bring more people into the development process can be invaluable.
“If I can build an application 100 percent, I’ll probably never get there, but if I can build out an application, say, 70 percent and let the last 30 percent be done by other folks, or citizen developers who probably have a much better grip on [customer] needs and requirements, I’m not only going to get there much faster, I’m going to get there with a much better answer and solution,” Brinkmann said.
Forrester believes AI-based predictions are a good answer for how to begin to include more people, end users or people outside of a core development in the creation process, but Brinkmann explained there are other options.
“I think conversational AI is a way to do it if people aren’t familiar with building pieces out and you can speak to something, and they can satisfy and help build additional pieces, that’s one way that they can answer some of those questions,” Brinkmann said. But the other options offer more traditional approaches such as simplified experiences and wizards for surfacing reports, data and predictive analytics.
Because of the amount of computing power required and lack of familiarity with the capabilities of AI, Brinkmann said that Logi Analytics has seen more requests for the more traditional options.
“If you’re a massive organization, and you might be able to afford to go out to a third party that runs a large, cloud-based AI like IBM’s Watson, that’s great,” Brinkmann said. “For the 99 percent of everyone else, I’m not sure that’s completely applicable.”
In last year’s Predictions 2018: The Honeymoon For AI Is Over report, Forrester determined that:
- A quarter of firms will supplement point-and-click analytics with conversational UIs
- AI will make decisions and provide real-time instructions at 20 percent of firms
- AI will erase the boundaries between structured and unstructured data-based insights.
And Forrester encouraged developers of analytics software to start experimenting.
Five ways to differentiate analytic solutions
Business Intelligence and analytics company Logi Analytics echoed Forrester’s AI and machine learning findings in its 2018 State of Embedded Analytics report. The report determined that there were five places where EA companies should focus their attention in order to remain competitive and provide adequate features to their customer base. Those are:
- AI implementation
- Predictive analysis
- Natural language generation
- Workflow management
- Database writeback
“We do have people ask us what to do, and we turn that question around and ask them: What are the business challenges that their customers would have that they’re trying to solve for, and does the technology help you get to an answer?” Brian Brinkmann, vice president of product marketing at Logi Analytics explained. “We’re all kind of technologists at heart, so we like to tinker with and use the latest gadgets and gizmos, but I think when you step back and say ‘How am I adding real value to my application for the customer?’ That’s where you need to start”
But once that’s been hammered out, Brinkmann said, there is a good place to start expanding.
“I think the one that people will go to first, and probably should go to first, is predictive analytics, and that’s because they have a set of historical data, and when you can unleash the machine learning algorithms on that data, it is highly likely that there will be business problems that people can solve right off the bat,” Brinkmann said. “And there’s problems everyone has regardless of the industry you’re in. You know, people have customers that turn, people have machines that break down, people have payments that aren’t paid. There’s things that they can do that will have a real impact on the business.”