インタラクティブなデータを Web サイトにパブリッシュ
Tableau Desktop ファイルを見ることができる無料ツール
Tableau のプロが講師を務める 1 時間のセッションです。
他の Tableau ユーザーと知識やアイデアを共有しましょう。
クラウドでの Tableau 利用に役立つリソース
スムーズな Tableau 導入やダッシュボード作成をサポートします。
After a 30-year history of the BI industry, best practices are finally emerging. And for the most part, those best practices involve throwing out the first 25 years of the industry. Perhaps “throwing out” is too strong: the early days of BI gave us concepts and tools to build on. But a BI implementation was typically complicated, heavy, expensive and slow. Companies spent millions of dollars and thousands of hours but never attained the promise of intelligence through data—never mind intelligence when and where you need it.
Some companies are becoming more competitive by using analytics strategically. These companies follow best practices that build on the legacy of BI, then throw it out.
We've also pulled out the first several pages of the whitepaper for you to read. Download the PDF on the right to read the rest.
It seems simple: let people answer their own questions. But legacy business intelligence systems forced people to answer basic questions about their data by submitting requests to developers who would then prioritize them, write code, and eventually deliver an answer. Any changes went through the same process. Hence the famous BI “queue” which subjected business users to long wait times and frustration. Best practice? Get out of the queue.
The people who can best answer questions are the very people asking them. The promise of self-service business intelligence is to let people create ad-hoc analytics to communicate a result, answer a question or just satisfy their own curiosity. Because the pace of business today is such that anyone standing in a queue gets left behind.
Legacy business intelligence systems meant spending months creating hard-to-use dashboards and analytics tools. People had to go to training to understand how to use even the tools for end-users, never mind developers. This created an inventory of dashboards and reports that people didn't use. The consumerization of software has taught us that people will use tools that are attractive and usable. People readily adopt new apps on their phones and new online tools if they find them useful. Why should BI be any different?
Best-in-class BI includes a focus on interactive, beautiful, and easily accessible analytics. New BI systems deliver dashboards and reports right in a web browser so users don’t have to jump through hoops simply to see their data. They support the same dashboards on mobile devices so the information gets to the places where decisions are being made. New tools allow people to design analytics that feel delightful rather than oppressive. As a result, people use those analytics—and the business gets smarter.
Legacy business intelligence had a concept of a single data architecture that was fast, scalable and held only clean data. It’s a wonderful idea. The problem is that no business exists with this ideal data architecture. Real businesses have multiple data sources of varying types and capabilities. The data source that is perfect and complete is most likely also obsolete.
The best practice for the next generation of BI tools is to let you work with all of your data, from spreadsheets to the most sophisticated databases and data warehouses. It also means embracing new technologies like Hadoop that bring new capabilities to the enterprise. And it means bringing any new data sources into compliance with your data governance practices, so that data can be maintained properly and shared across the organization.
What if you could blend in new data sources on the fly? The answer to a question is rarely well-behaved enough to sit obediently in a single database. Blending sales data with operational and finance data yields answers that are much richer and more useful than analyzing single data silos. Publishing that blended data brings it into view of everyone with the proper credentials, and multiplies its usefulness because now others can run analyses against the combined data.