Sign up for our newsletter and get the latest big data news and analysis.

The Power of Search to Analyze Business Data

Scott Holden_ThoughtSpotIn this special guest feature, Scott Holden of ThoughtSpot discussed the power of search for extracting insights from business data. Scott is the VP of Marketing at ThoughtSpot where is responsible for product marketing, demand generation, content marketing, PR, AR, and events.  Prior to ThoughtSpot, Scott spent seven years at Salesforce overseeing a number of marketing teams, including leading marketing for the Salesforce1 Platform, the Sales Cloud, Chatter, Industry Marketing, Customer Marketing, and SMB Marketing. Scott has an MBA from Stanford University and a BA from Colgate University in economics and philosophy.

The ultimate solution to the failure of “self-service BI” lies in the same trend that spawned Bring-Your-Own-Device (BYOD): the consumerization of IT. With the rise of mobile, business users wanted the ease-of-use and familiarity of their personal smartphones, and enterprises found ways to make that work. Now business users want to be able to query their data with the ease-of-use and familiarity of search engines such as Google and Yahoo.

Despite promises that self-service BI would deliver on this desire, typical business users, such as marketers and sales managers, can still do little with the current crop of tools unless they go through costly and time-consuming training. As a result, very few organizations are realizing the full potential from their BI investments.

The solution to providing every business user with full access to up-to-date information from every corner of the enterprise lies in search technology. While this idea has been floated in the past, recent advances, including increased computing performance and scale out architectures, have now made powerful search-driven analytics products feasible.

Making Search-Driven Analytics Useful

As helpful as Google and Yahoo searches are for some tasks, typing in “sales for north america” into a search box and getting a long list of every mention of “sales” and “north america” found in every report from across the enterprise is not useful for business users. This is Business Intelligence after all: accuracy is paramount and required to build trust and adoption. “Close” only counts if you’re playing horseshoes.

Instead, a search-driven analytics solution should guide users to one accurate answer. For a search technology to be able to do this successfully, it must have the following characteristics:

1. Guided Suggestions – An interactive guided search experience should lead the user and the system to the right answer, similar to how Google’s type-ahead feature guides consumer queries. If you’re looking for sales data, suggestions should give you the choice of calculating sales data from the order table or the from the product table and ranking those two suggestions based on the rest of your query and what other users like you are doing. Analytics users don’t want to wade through a list of ranked pre-built reports or have to trust probabilistic results based on an error-prone programmable algorithm.

2. Search Intelligence – When it comes to analytics, nothing is more important than 100% accuracy. Natural Language processing (NLP) has received a lot of attention recently, but the world’s best algorithms still have 10-20% error rates.  Search-driven analytics should crawl your data to understand the relationships between data tables (e.g. “product” and “sales”), eliminating the need for a BI architect to create schemas and map tables together. This enables the system to perform calculations (“sales last year by product by region”) on the fly, including across multiple time grains like weekly, monthly, quarterly, etc. The solution should also be able to check spelling and look for synonyms (“sales” “revenue” “bookings,” etc.).  To be truly useful for business users, a search-driven analytics solution should also be able to automatically show a visual representation of the search results in the form of a chart or graph.

3. Speed at Scale – To provide answers in a timely manner, a search BI solution must leverage hardware advances, such as processing speed and greater in-memory processing capacity, to be able to easily handle billions of rows of data, responding to simple queries in less than a second and returning complex queries in minutes. As Google has proven, 40% of users abandon queries that take longer than 3 seconds to return results.

4. Automatic Modeling – a massive amount of time and expense on any BI deployment is spent on data modeling. Whether it’s building cubes, tuning hardware, or programming semantic search models – it’s expensive. A modern search-driven analytics tool should be schema aware, calculate time granularity on the fly, and require minimal data modeling. It’s now possible to reduce implementation times from months to hours.

5. Data Freedom – In keeping with the major expansion in the types of data sources considered valuable today (unstructured, semi-structured, etc.) the search-driven analytics solution should be able to handle any type of data, from text documents and spreadsheets, to social media streams, to data warehouses – on-premises and in the cloud.

6. Data Security – To ensure compliance with regulatory and business requirements, the search-driven analytics solution must be able to return only those results that the user is authorized to access. Access controls should be built into both the results and the search box itself.

Enterprises desperately need a better way to gain insight from the volumes of information they collect. Fortunately, new search strategies and technologies are finally in place for to help business users to get the exact information they need when they need it.

 

Sign up for the free insideBIGDATA newsletter.

Leave a Comment

*

Resource Links: