Google Analytics has today (July 18) lifted the lid on a voice-based search feature that helps users of the software suite to ask for and receive “plain English” questions and answers, via artificial intelligence (AI).
The feature, which uses the same natural language processing technology available across Google products like Android and Search (see video above), will become available in English to all Google Analytics users over the next few weeks.
Analytics Intelligence, a set of features in Google Analytics, uses machine learning to help analyze and interpret data queries, and is being rolled out to users of the software suite, the world’s most popular business analytics service.
Google maintains the introduction of this “consumer-level simplicity” will save businesses across the world an untold amount of man hours, as many software and business analysts’ resources are consumed answering relatively simple queries.
A blog post attributed to Anissa Alusi, Google Analytics, product manager, and Ajay Nainani, reads: “We've talked to web analysts who say they spend half their time answering basic analytics questions for other people in their organization. In fact, a recent report from Forrester found that 57% of marketers find it difficult to give their stakeholders in different functions access to their data and insights.
“Asking questions in Analytics Intelligence can help everyone get their answers directly in the product ― so team members get what they need faster, and analysts can spend their valuable time on deeper research and discovery.”
Finding answers to relatively simple queries such as ‘what are my best selling products’, is normally a relatively laborious process, given that it requires Google Analytics users to navigate a series of text-based interfaces using language that are not as intuitive as human speech, Babak Pahlavan, Google Analytics, senior director, measurement and analytics, and Anissa Alusi, Google Analytics, product manager, speaking to The Drum.
“The vision has been wouldn’t it be nice if we could just ask the actual question [and then receive the answer],” explained Pahlavan.
“So if you are not an expert and don’t know the terminology, now you can just click [on the relevant dropbox] and literally just ask the question… ,” said Alusi, adding that the machine learning element of the new software suite means it can infer a lot of the details as to the meaning and context of the query in order to provide the most productive information.
For instance, if a user asks ‘what product is doing best?’ over a period of time and continues to input a specific KPI (such as ‘conversion rates’, ‘sales’, or ‘customer retention’) it will begin to learn that users’ preferences. “So it personalizes based on the metrics you care about,” added Alusi.
“The more people use it, the more the system as a whole gets more precise,” said Pahlavan. “Right now it’s about answering the ‘what questions’ by looks at historical data… but soon we’ll be able to answer things like ‘why are sales down’… based on the datasets, it will try to find causes,” he added.
Eventually, Google’s ambition for the feature is to start providing Google Analytics users ‘how answers,' such as “how do I make more money," Pahlavan added. Over time, Google hopes that its software developments will allow analysts to focus on more forward-facing and strategic issues like these, while the more basic questions that AI can answer can be broadly accessible to the rest of a business, such as marketing departments, via way of the features being rolled out today, according to Pahlavan.
“It’s about allowing Google Analytics users to be able to ask human-style questions and to then get answers from the system, just like how you’d ask Google Search or Google Home. We’re taking this consumer-level simplicity to the business world,” he added.
This roll-out comes as a Google report entitled '2016-2017 Marketing Analytics Challenges and Goals' identified a widespread difficulty in the sharing of data-based analytics and insights across participating organizations.
The reports states that 61% of marketing decision makers claiming they struggled to access or integrate the data they needed in 2016. Meanwhile, the same amount of such participants said they expected similar struggles in accessing or integrating the data they needed in the following year.