Editor's introduction: Many companies' self-developed data products have a lot of input costs and have been iterated for more than a year, but there are still many deficiencies in the business. With so much investment and no results, why not buy an external business data platform product?
Today casually talk about a topic, self-developed data product iteration for more than a year, why not buy a third-party commercial data platform product?
First, the status quo of big data product research and development
Multi-enterprise in the construction of big data will encounter a problem, put into production and research after more than a year of iteration, made some data products, big data team in the construction will combine the needs of the tools and business on the market, iteration for a period of time is able to meet the company's own business development on the data demands.
With the development of the company, it is often affected by changes in the organizational structure of the enterprise, segmentation, team work intersection and repetitive work, etc., or the resources are called, and the enterprise big data products are affected by the ability of the data product team and the big data capabilities of the technology (such as front-end resources, back-end resources, server resources) in the long-term iteration, and the decline in data products is getting slower and slower.
Some features in the data platform are lost more over time. At this time, in the inventory, it was found that there was no degree of investment in so much manpower and material resources why these platform businesses wanted nothing (let's not talk about whether the business was reasonable). At this time, everyone began to talk about whether to buy external products.
Second, the comparison of external data products
In recent years, there have been some professional big data tool platform companies developing their own data platform tools, such as metadata tools, report tools, and APP log analysis tools.
These tools will be more professional and more versatile, but the disadvantages will also be obvious in the fit of some enterprise business blockage is quite lacking.
For example, the BI function in the third party is very comprehensive and complete, the chart library is rich, the display is very beautiful, some featured permissions are difficult to match the company's organizational structure, and some business chains are particularly long, but only bought log analysis tools, traffic data and transaction data will cross-analyze and other scenarios, these third-party tools will deviate a little.
There are also a large number of middle layers to build a large number of middle layers for various types of cross-analysis of the business, and some tools support data import functions, but they are also difficult to use.
Self-service research and development of data products of data products, the disadvantage is that the data dashboard function, report function, UI aesthetic function and icon richness are worse than the third party, in the case of resource shortage, a row in the report may be scheduled to several Q after, the business team often complains about this function is poor, that function has bugs. The advantage is that it is in line with the personalized demands of the business.
Third, self-research and procurement of ROI
A third-party data product platform is only a few hundred thousand, and the people who want to do this thing may be millions a year. Or buying a third party?
In essence, this is a conflict between self-research and procurement, which is something that needs to be considered in decision-making, and the decision point is the comparison of short, medium, and long-term ROIs.
In terms of enterprise size:
- In general, small and medium-sized enterprises are recommended to buy (whisper: there are quite a few third-party platforms have a common problem, the renewal rate is very low, the customer churn rate is very high, we will not explore this problem).
- Medium and large enterprises to see whether the enterprise in the self-research investment cost to do, in general, the organizational structure ReOrg, after the integration of resources will give priority to do things of greater business value, the general data product team in this tool can not get the barrier output will give priority to make adjustments. If there is a surplus of resources, the big data team can do some value data products.
This needs to be judged from the synthesis of short, medium, and long-term ROI implemented.
- Self-research: advantages Close to the business, most of the technical implementation is empty, compatible, and the business coincidence is 100%. Internal important non-urgent tool construction, long construction cycle, affected by front-end resources, back-end resources. The input-output ratio is moderate.
- Buying: Advantages, short cycle, fast results. It is rich in functions and can be used quickly to meet the needs of rapid business development, and the business support coincidence reaches a maximum of 50%-70%.
Fourth, implement some potential possibilities of third parties
The introduction of a new system will cause the overall system to become complex for a period of time (such as self-developed system A, purchase system B will coexist, which is not good for the team or the system level). The implementation is short-term and medium-term (1-3 months is more painful and the work is large), and the medium-term advantages are obvious. There is a long-term problem of customizing personalized functions, and it is very likely to do secondary development.
In the general short term (within 3 Q), the cost of self-research is greater than the cost of procurement and implementation: the data team is required to do a good job of taking out the self-developed Plan within a certain stage, and can only be able to build a good alternative solution under the input of labor cost.
Medium and long term (more than 12 months), the business development is very rapid brought about by the demand for data will also be changeable, the collection of data, the demand for the middle layer of data is to require rapid implementation to support the demand for business, the semi-closed nature of the third-party platform and its own storage format lead to the expansion of the middle layer will become unable to keep up with the rhythm of business demand, and the contradiction will break out over time. There needs to be a corresponding scheme to keep up.
Fifth, the establishment of some advantages of data products
Whether data products and technologies have established a barrier to third-party platforms in the support and implementation of business, of course, the barrier advantages in content construction will be greater than the advantages in tool functions.
Internet companies in the construction of big data generally follow a short and fast way, and even integrate the tables of the business system into several warehouses, rarely do the design or do a simple integration according to the needs, and export wide tables or summarize data for business use.
Companies in the Internet industry rarely publish some data models that have precipitated for the industry. I know that in the field of big data in recent years, only Alibaba released a standard data model for e-commerce transactions last year.
The construction of the data middle office, it can be said that if the Internet enterprise data middle office wants to enter different fields in digital transformation, it needs to precipitate a large number of industry models to enter the data integration of some heavy industries, otherwise it is difficult to produce a good data model navigation chart. For example, data governance is now very important, but it is a bit reluctant to land, data governance is very important, but many things are a bit inverted.
Author:Matsuko (Li Boyuan)
This article was originally published by @Matsuko and is not reproduced without permission
The title image is from Unsplash, based on the CC0 protocol