As early as more than ten years ago, Clive Humby, a British mathematician, said, “data is the oil of the new era and contains great value”. But at the same time, he also stressed that “without refining and analysis, the data can not show their value”. We must process and “refine” data crude oil in order to fully extract and release the value of data and provide continuous power for business digital innovation.
As a data intelligence enterprise, relying on advanced big data algorithms, knowledge mining and deep learning technologies, daily interaction has continued to “treat numbers” and “refine numbers” for many years, and has accumulated profound experience in data processing and data value extraction.
While continuously consolidating the data capability and providing professional data intelligence solutions for vertical fields such as mobile Internet, digital marketing and smart city, he interacted and summarized his valuable experience in data extraction and data insight every day, and creatively put forward the “five steps of data intelligence”, which aims to fully tap Efficient use of data value to provide mature ideas and effective solutions.
Five steps of data intelligence to create a closed loop of “refining” data value
The five steps in the “five steps of data intelligence” are “result sampling”, “detailed annotation”, “similar expansion”, “practical application” and “feedback attribution”. Through these five steps, the daily interaction completes the processing and extraction of massive data resources, constructs a rich three-dimensional label system, interprets the humanistic meaning behind the data, and truly gives play to the driving role of data in business growth.
Step1 result sampling
The first step “result sampling” in the “five steps of data intelligence” refers to receiving data samples, collecting and analyzing data. The sample data here includes not only one party data from industry customers, namely “y value”, but also rich feature data and other third-party data, namely “x value”. Daily interaction believes that “y value” plays an irreplaceable role in guiding and leading us to obtain accurate and effective data analysis results. However, only one party’s data is not enough. Enterprises also need the blessing of larger volume and more dimensional data to carry out comprehensive and in-depth insight.
For example, in the field of digital marketing, if brand owners want to understand consumers in an all-round way, they not only need to rely on brand owned data such as member data, transaction data and advertising data, but also need to carefully depict the overall characteristics of consumers from multiple angles with the help of third-party data; In the field of mobile Internet, app also needs to fully combine users’ app end-to-end data with end-to-end interest and scene preference data, so as to establish a more complete user portrait system and support intelligent operation.
As a data intelligent enterprise, daily interaction builds thousands of portrait labels with massive online and offline data accumulation, which can provide rich third-party data support for industry customers to improve data assets and carry out in-depth and high-quality data analysis.
Step 2 annotation details
The second step of the “five steps of data intelligence” is to carefully examine and insight the data samples, summarize the correlation characteristics and internal laws from the complex data, interpret the deep-seated humanistic meaning behind the objective data, and provide scientific basis and effective support for intelligent decision-making. This requires both efficient technical means and professional insight from people.
For example, when daily interaction helps relevant government units analyze the urban population, on the one hand, it uses big data technology to connect and connect multi-dimensional data such as time, space and population, and analyze and predict the urban population portrait, population structure, population distribution and population migration trend by building algorithmic models such as population characteristics and population flow; On the other hand, urban planners are also required to conduct comprehensive research and judgment in combination with the current economic, social and industrial development of the city, mark and interpret the humanistic meaning of various data indicators from a professional perspective, so as to provide scientific decision-making reference for urban development and social management.
Similarly, in the fields of digital marketing and mobile Internet operation, daily interaction also emphasizes that data analysts should manually observe and continuously optimize the sample data and algorithm models in combination with the experience and knowledge of their industry, so as to ensure that they are not misled by abnormal and invalid data and truly see the essence through the data.
Step3 similar expansion
The third step “similarity expansion” in the “five steps of data intelligence” refers to finding more potential people with similar characteristics according to the characteristics and models obtained from the insight analysis of data samples in the second step, so as to realize the field needs of accurate customer expansion, similar people mining, directional promotion and so on.
Just as every living body has its specific biological genes, daily interaction believes that every user in the Internet era also has its internal data genes, that is, the attribute preferences and interest characteristics shown in users’ online and offline behaviors. In the second step, the problem we want to solve is to find the data genes of sample users, and extract and construct specific labels and models to represent them; The third step of “similarity expansion” is to accurately select groups with similar data genes to the sample users from a large number of users.
In this regard, the daily interaction has a large amount of terminal data, covering users of the whole network, and has established a rich three-dimensional label system. With the help of looklike and other expansion algorithms, it can effectively help industry customers find large-scale potential target groups in the wide user pool of the whole network through similar characteristics.
Step4 practical application
The fourth step “practical application” in the “five steps of data intelligence” refers to the application of the results of data analysis and the accurate potential customers found to the actual business scenarios such as brand marketing and user operation.
Step5 feedback attribution
It is worth mentioning that whether the brand owner carries out advertising or the app carries out user operation, it will produce various effect data such as advertising click, user retention, transaction transformation and so on. Daily interaction believes that these effect data are of great value. It is necessary for enterprises to return the effect data generated in practical application and carry out in-depth summary and attribution analysis to help evaluate and optimize the existing data, insight results and algorithm models.
That is, as emphasized in “feedback attribution” in the fifth step of the “five steps of data intelligence”, through full backtracking analysis of the data of the whole business process and link, a continuous iterative and evolutionary “refining” closed loop is constructed to fully stimulate and release the positive driving value of the data to the business.
Five steps of data intelligence, enabling industry digital upgrading
At present, daily interaction has deeply integrated the “five steps of data intelligence” into the company’s data intelligence solutions and products to improve efficiency and energy for digital innovation in vertical industries.
Based on the “five steps of data intelligence”, daily interaction in the field of brand marketing helps brand customers such as mothers and infants, beauty salons and FMCG integrate and open up global data assets, gain comprehensive insight into consumers from multiple dimensions, screen out high-quality media, generally increase the TA concentration of advertising by 30% – 50%, and help customers find accurate potential customers from the vast crowd, Achieve greater marketing effectiveness;
In the field of mobile Internet, daily interaction helps app improve the portrait label system, achieve high-quality new customers and refined operation through in-depth data insight and advanced algorithm model, and help customers achieve sustained growth and efficient realization of traffic value;
In the field of smart city, daily interaction helps governments at all levels in-depth governance data, helps relevant departments such as planning, transportation, culture, tourism and emergency response to conduct grid analysis and multi-dimensional insight into regional population, spatial flow and industrial development, and provides in-depth data support for urban intelligent management and scientific decision-making.
At this stage, daily interaction is gradually turning the “five steps of data intelligence” into products and tools, and landing in the company’s data middle platform product “daily data governance platform”, enabling the digital upgrading process of more industry customers.
The core of the daily data governance platform lies in data governance and application. It not only relies on the strong data and technical ability of daily interaction, but also integrates the company’s in-depth understanding and expert knowledge of the vertical industry. It is committed to helping customers solve the pain points such as data island, low data quality and difficult data application, so as to reduce the threshold of digital upgrading of enterprises and governments.
“Five steps of data intelligence” is the methodology of daily interaction on data application, which has been precipitated into the daily treatment platform as an important ability. Through the daily counting platform, customers can quickly complete the aggregation, connection and integration of multi-source heterogeneous data, and build labels and models very conveniently; At the same time, the annotation of insight results and the evaluation and optimization of the model can also be completed efficiently and intelligently on the daily treatment platform.
We believe that with the help of mature methodology and intelligent products, enterprises will be more relaxed and labor-saving in the process of extracting and refining data value. In the future, the daily interaction will continue to output practical experience of data intelligence, accelerate technological innovation, sharpen and polish products, and promote the digital innovation of industry customers and government departments.