"We are giving people the same kind of search capability they have on Amazon but for their travel data," said CWT VP and chief data scientist Eric Tyree.
The product in question is AnswerIQ, the travel management company's latest response to the need for an intuitive business intelligence platform for travel managers. The company has included AnalytIQs as part of its suite of products since 2015. A few months ago, it rolled out Travel Consolidator as an add-on product to integrate corporate card and expense data, though AnalyticsIQ was already bringing these in as separate data feeds. Now, the critical element has been how CWT enables clients to search their data.
"It sounds really trivial, but it's not," Tyree said. "We need to enable an intuitive search that delivers not just a list of possible answer like Google would do, but one answer, and it needs to be right."
To do that, the company has spent the last year not only working on additional cleansing and normalization processes for travel data but also mapping that data to different search terms and dialect variations. This way, when clients arrive at the AnswerIQ data portal, which is very nearly a blank screen with a search bar, they don't struggle to come up with a limited set of pre-established key words or search combinations that will produce results.
Underlying the system is a search platform called ThougthSpot that specializes in natural language. In addition to supporting the language mapping, it's also running machine learning algorithms that continuously pick up search terms across the user base. It builds a history of search terms for individual users as well. Both of these "memory" systems help prompt user searches as the queries are entered into the search bar.
"We tell new users to type, then breathe; type, then breathe," said Tyree, describing the moment or two the system may need to figure out what type of search the user wants and deliver the prompt. "It turns out, a lot of people want similar searches," he said.
To that end, the tool also offers a unique sharing function that allows users to share search parameters and results not only to other users across their own company, but also to different groups of users across the entire CWT platform. "Travel managers aren't hypercompetitive with one another; we find that they want to collaborate with user groups and share their information," said Tyree. In sharing a search to another platform user, the new user simply plugs in the search parameters and the result will render with the new user's data. All users can create and share their dashboards, as well, except AnswerIQ now calls them "pinboards" for a more consumer flair.
For now, AnswerIQ is only built over TMC data, but Tyree said the company is producing a version that will incorporate card and expense data as well. "You can put anything you want in here, but it's the integration work and translation piece that is hard," he said, noting that travel data is by far the hardest, so incorporating the other two will be easier than the first task. Tyree said the company plans to put the AnswerIQ tool on top of Travel Consolidator in the coming months and Tyree added that it will likely become the entry point for all CWT business intelligence.
"We've fallen in love with this [technology] so much that we are pretty much planning to scrap all the other BI tools because this is so much better," he said.
The ultimate mission for AnswerIQ is to put analytics power in the travel manager's hands and increase productivity with AI-assisted capabilities. It's been a tough nut to crack because, Tyree said, "travel data is head and shoulders by far the worst data I've ever seen," noting many CWT tech-industry clients that wanted to work on travel analytics on their own, only to come back to CWT empty handed. "You really need travel experts to be involved, and we believe that is a differentiator for us," he said.
By the second quarter of 2019, CWT is aiming for true natural language capabilities for AnswerIQ and voice activation will come along with it.