Petya Petrova - Fotolia

QuickSight Q a potential winner for Amazon BI platform

Since its introduction in 2016, Amazon QuickSight has offered solid but not top-of-market BI capabilities. The addition of a new natural language query tool could change that.

With its recent introduction of QuickSight Q, AWS added natural language query capabilities to its analytics platform.

It's a move that has the potential to add top-flight functionality to a suite that analysts say is useful but somewhat basic in comparison with those offered by other vendors.

The perception that QuickSight is basic, however, doesn't reflect the capabilities that have been added to the platform during the past couple of years, according to AWS, which offers few enterprise application software systems.

"I think it's fair to say that if you looked at QuickSight two years ago that you would have said it's useful but kind of lacking, but that is not what we hear from customers today," said Matt Wood, vice president of artificial intelligence at AWS.

As a result, more organizations are choosing QuickSight for their BI needs, he continued.

"Organizations do very robust evaluations, and in those robust evaluations, when they look at the capabilities of QuickSight, the features we've added over the past two years, they're choosing QuickSight," Wood said.

In addition to QuickSight Q, AWS has added about 75 other new features over the past year, according to Tracy Daugherty, director of product management at AWS. They include new visualizations, connections to added data sources, and access control and security capabilities.

"They go across the board," he said. "I could pick every part of the product, and we've added a bunch. We've changed so fast."

Question and answer

QuickSight Q was unveiled as the latest addition to the AWS QuickSight analytics platform in late September. It enables users to ask questions of their data using natural language and get answers with relevant visualizations that lead to insights and ultimately foster data-driven decision making.

Users need no knowledge of code to query data, and the new tool does not rely on previously built dashboards and reports.

QuickSight Q is able to generate original content based simply on the user's query in natural language, thereby removing the need to constantly update dashboards and reports when new data is available.

According to Wood, QuickSight Q uses machine learning to interpret natural language, translate it into SQL, run the query, translate the result back into natural language and return a response in seconds. In addition, it's programmed to provide auto-complete suggestions for key business terms and is able to spell check and understand acronyms and synonyms.

"We've done a lot with machine learning, but QuickSight Q is really the first time we've applied natural language understanding to data," he said. "We were able, through some machine learning algorithms under the hood … to train a bunch of natural language understanding models. The innovation came in the translation to SQL."

The tech giant's intent with the feature is to enable more people to work with data, Wood continued.

"It allows organizations to take their data and broadly distribute it across their entire organization, and they've never really been able to do that before," he said. "Before, if you wanted to take your data and broadly distribute it across your whole organization so that anybody could access it and use it to inform their decisions, it was very hard to do."

Natural language query (NLQ), meanwhile, is still a developing technology. Most analytics vendors offer some natural language processing (NLP) capabilities , but they remain mostly rudimentary.

If QuickSight Q proves to be as effective as AWS posits, analysts say it has the potential to be a significant tool for analytics consumers. The tool could save them time and effort and open analytics to a broad array of business users without formal training in statistics and computer science.

We need to see how [QuickSight Q] performs in production, but I know several vendors who will be looking at this with interest to take their own applications to the next level.
Donald FarmerFounder and principal, TreeHive Strategy

"Natural language query features have been in BI products for years, but they have been good for demos rather than use in the real world," said Donald Farmer, founder and principal of TreeHive Strategy. "We need to see how [QuickSight Q] performs in production, but I know several vendors who will be looking at this with interest to take their own applications to the next level."

Similarly, David Menninger, an analyst at Ventana Research, noted that other vendors have made advancements in NLQ but there's plenty of room for improvement.

"Lots of vendors are working on their natural language query, and some vendors have made good progress," he said.

In particular, Oracle, Tableau, ThoughtSpot and Yellowfin are among those who have made NLQ a priority, Menninger continued.

"Everyone is dabbling around the edges and is working on it, but no one has nailed it yet," Menninger said. "There's still a lot of work to be done in this segment in the market, and it's not yet used very widely in organizations."

If QuickSight Q is able to change that and make using analytics more widespread with natural language, it has the potential to alter the perception of the entire QuickSight platform.

A sample Amazon QuickSight dashboard
A sample Amazon QuickSight dashboard displays hospital financial data.

An uneven past

Amazon first unveiled QuickSight in November 2016 after more than a year of beta testing with 1,500 customers.

It featured BI capabilities built around dashboards, and since then has added machine learning capabilities and other augmented analytics features. But other vendors have also added those same capabilities.

Meanwhile, those other vendors -- among them Tableau, Qlik, Google with the acquisition of Looker, and Microsoft with Power BI -- offer analytics platforms with deeper capabilities, according to the analysts.

"It's a competent, simple analytics component," Farmer said of QuickSight.

He added that he mostly sees QuickSight as being embedded into applications by customers already using other AWS features, and rarely sees QuickSight as an organization's primary analytics platform.

Likewise, Menninger noted that QuickSight provides basic analytics capabilities but not the depth of functionality that organizations often require to fully understand their data and make informed data-driven decisions.

Specifically, he noted that QuickSight doesn't connect to as many data sources as other vendors. Its mobile capabilities trail those of other vendors. It lacks collaboration capabilities, can't assemble briefing books or stories, and, while its AI and machine learning capabilities are strong, it's weak on planning and optimization.

"There's a core set of capabilities that everybody knows about and everybody uses, and Amazon has been chipping away at those for years," Menninger said. "They do a lot of what's necessary. The product does a lot of things, but it doesn't do everything."

He added that QuickSight doesn't appear to be an attempt by AWS at establishing a top-of-market position compared with other vendors. Instead, the aim of the platform is to provide a functional BI tool for AWS users.

"The rest of the portfolio is about offering a service that is good and capable, and the fact that it's available on Amazon is what makes it attractive," Menninger said.

In addition, the fact that QuickSight is cloud-native and built with a serverless architecture makes it unique, according to Menninger. Farmer noted that its inexpensive pricing is a differentiator.

NLP's potential

QuickSight Q adds only one new capability. But as BI evolves with a goal of enabling more users to work with data and make data-driven decisions without having to go through an IT department or analytics team, NLP is seen as a capability with the potential to do just that.

If users can query data by simply speaking a question into a microphone or typing a question into a search bar and immediately get a response, and users can then ask follow-up questions and get rapid responses to those as well to better understand their data, it frees them to explore data and discover insights without having to learn code.

However, to date, NLP capabilities have struggled given the nuances of language.

There are more than 5,000 languages and hundreds of alphabets worldwide. Most NLP tools understand only a handful, if that. And within the languages they understand, there are words that are spelled the same but have different meanings and others that sound the same but carry different meanings.

So NLP tools often struggle to decipher the intended meaning of many queries.

Amazon, however, has more experience with language-based technology than most. It first introduced Alexa in 2013, and boasts other AI initiatives as well.

So if Amazon can make QuickSight Q a more functional NLP tool than Tableau's Ask Data or ThoughtSpot's search-based platform, among others, it has the potential to turn QuickSight into a critical capability for organizations looking to enable business users.

In most organizations, fewer than half of the employees use analytics, according to Menninger. Strong NLP capabilities could alter that drastically.

"If suddenly you could do it with QuickSight Q, maybe then 80% of the organization could use analytics," Menninger said.

And QuickSight has the significant advantage of being part of the Amazon ecosystem, he continued.

"If you think about Amazon's work … around Alexa, they probably have a leg up on other vendors in applying natural language to business intelligence," Menninger said. "It's early in the days of natural language for business intelligence, but because Amazon has a lot of experience, it could give them a leg up in a very interesting part of the market and they could potentially leapfrog the competition."

Similarly, Farmer cited Amazon's work with commercial voice recognition technology as a potential advantage for QuickSight.

"Amazon has some great experience in this space thanks to its work with Alexa and other AI initiatives," he said.

Wood, meanwhile, noted that QuickSight takes advantage of AWS' machine learning capabilities to improve the more it's used, learning from users' queries to respond with more relevant responses.

"You get this nice flywheel effect," he said. "The more customers ask questions into Q, the more that feeds back to the authors and the more authors can build new dashboards or new data sets. The sheer act of asking questions improves the data quality and analytics capabilities of the entire organization."


Looking beyond QuickSight Q, additions to QuickSight will include more machine learning capabilities to enable new kinds of analyses such as discovering the root causes of anomalies and changes in data, according to Wood.

"We're continuing to invest in machine learning," he said. "We're continuing to find new algorithms and new approaches to apply to data to allow customers to more easily understand it. We're super interested in helping customers do more through machine learning."

In addition, the QuickSight roadmap includes further investment in its mobile experience and more embedded analytics capabilities.

And with AWS re:Invent, the vendor's annual user conference, scheduled to begin Nov. 30, some of the new QuickSight capabilities will be unveiled soon.

As for whether QuickSight Q and other initiatives can alter the perception of QuickSight from a functional analytics platform to an innovative market leader, it depends, in part, on whether AWS aims to change what users view as most important, according to Menninger.

"It will depend on whether they want to move the goalposts or compete on the playing field that everyone else is on," he said. "The QuickSight Q announcement suggests a path toward where [QuickSight] could become a leader in the market."

Dig Deeper on Business intelligence technology

Data Management
Content Management