Late last year, global BI giant Tableau announced it was pulling out of China and integrating its business into Salesforce's partnership with Alibaba. Judging from the trend of technological development in recent years, BI seems to be on the decline. Is BI still an easy business?
A common question about BI is how business intelligence generates value, and why enterprise IT departments or business departments are ultimately reduced to making reports. The value reflected is only to save us the work of making reports, and that is all.
Highly dependent on systems, knowledge and environment, more questions than answers
Traditional BI is concerned about whether the data is beautiful or not, whether the form presented is attractive or not, whether the threshold is low enough, and whether the value feedback cycle to the business is relatively long.
Liu Chao, president of Sugrui Data, summarized and analyzed the problems in the current BI industry of "good responsibility for tools, relying on people to do the work, and difficult to precipitate knowledge", which were roughly summarized as "three dependencies" : dependence on the system, dependence on knowledge, and dependence on the environment.
At present, most of the data analyzed by BI comes from various business systems. When using these data, we will encounter many problems in quality, safety, standards and other aspects. Problems arise from machines, but need to be solved by people, which is not reasonable and produces a lot of unnecessary workload. Therefore, the next generation of BI should let machines do what machines should do and let people do what people should do.
Second, there is the dependence on knowledge. Today, with the development of human society, various disciplines and industries have accumulated their own knowledge. If this knowledge is applied to data analysis, it needs to be accumulated, absorbed and integrated in advance, which requires a lot of efforts. The second question is whether tools can take the place of people to digest these knowledge and lower the threshold of knowledge application.
Finally, there is dependence on the environment. We believe that the data accumulated to a certain scale, combined with the enterprise's knowledge history data, can realize intelligent knowledge sharing. However, the environment of knowledge application is complicated. How to share the knowledge formed based on the analysis of a system to another system quickly to produce value? If the knowledge is embedded in other business systems, it still needs to be integrated, which in essence still does not get rid of a lot of technical work. Only by solving the problem of hindered knowledge transmission can enterprises truly have the ability of dynamic learning and accumulation.
The industry believes that according to the "DIKW" model theory, BI should not stay in the data stage, do statistics and visualization, but focus on transforming data into knowledge. The relationship between BI and people in the next generation should be that BI tools summarize knowledge in data, and people combine knowledge with business to produce wisdom and create value.
The next Generation of BI's hidden giants: Enhanced analytics and machine learning algorithms
With the change of business environment, the dimension of data analysis and the requirements of users' usage habits are also constantly upgraded. How to design BI for the next decade? I think a good capability is "when the user doesn't feel it, it's everywhere." That is, when it works, you don't feel it, and when it doesn't work, you feel strongly about it.
At present, some domestic BI, including nextionBI, have caught up. Many functional components are packaged with AI algorithm, and there are more automation and intelligence in the invisible place. For example, data analysis model provides general algorithms such as timing analysis, prediction, classification and clustering. Users used to need some model training of machine learning if they wanted to make some prediction or classification, but now users only need to use the box to complete 3D/timing/graph analysis with quick one-click operation.
Another highlight of the data analyzer is the intelligent data interpretation based on NLG(natural language generation). The interpretation is also natural language, so that boring data can be spoken automatically, and friendly data interpretation AIDS can be provided for people who are not sensitive to data. When the amount and dimension of data are relatively large, intelligent data interpretation can quickly tell you what the distribution of data is like, what are the characteristics, and whether there is an invisible correlation between the characteristics? What is the degree of correlation? Is the overall data distribution free?
The appearance of business intelligence is the presentation of visual analysis report, but its essence is still business problem, management problem. Enhancing analytics is about minimizing the work of data engineers, freeing up the power of data analysts, and letting machines do a lot of the work for developers.
The migration from data to knowledge will drive the evolution of BI in the next decade
"Ten years from now, what we are doing now will not count as BI," said nextionBI at the press conference.
That's a lot of talk. How do you do that? Liu summarized the key capabilities of nextionBI as data fusion, enhanced analytics, and agility and ease of use. In terms of technology, it seems that there is a suspicion of old words and new words, the same concept also needs to see what to talk about, data-oriented and knowledge-oriented have essential differences.
If it's just data-oriented, data fusion might support larger amounts of data, enhance analytics with a few more statistical functions, and be agile and easy to use with API documentation and detailed user guidance. However, if it is knowledge-oriented, data fusion needs to pay attention to the ability to cover more data dimensions and types, and provide guidance for dimension selection in combination with correlation analysis. Enhancement analysis requires both explicit knowledge recognition and tacit knowledge mining capabilities, and combines technical capabilities with scenarios. Agility and ease of use need to go deep into industrial iot and digital twin, and integrate with digital applications across industries.
From data-oriented migration to knowledge-oriented migration, it is closer to the business side, so that data is not a burden and business personnel can focus more on the business itself. The author believes that digitalization must be realized by the people closest to the business. To some extent, digital transformation is a requirement for people. If everyone has the ability of digital innovation, this company is a company with strong digital innovation ability.
The accumulation, discovery, and dissemination of knowledge can help us to see the world more fully, by trusting what you can't see.
