Under the favorable policies of promoting consumption and the maturity of new energy automobile products, the domestic automobile industry is growing continuously: as of the end of September 2022, the number of motor vehicles in the country will reach 412 million, of which 315 million are automobiles. The monthly new increase exceeds 2 million, and the monthly car production and sales also increase significantly year-on-year.
Automobile OEMs promote the digitalization process of automobile marketing and production through self-built research and development and integration of external technology suppliers. In this digitalization process, the underlying infrastructure is the control of the quality and efficiency of the R&D team by the technical executives of each enterprise, and the control of the internal R&D team and external technology suppliers from the perspective of R&D process control, R&D result evaluation, and R&D resource evaluation. Promote the digitalization process of OEMs.
The R&D challenges faced by automotive OEMs in the process of building digital transformation involve various technology combinations such as multi-terminal, multi-platform, software and hardware combination.
- On the marketing side, cooperate with advertising suppliers to connect with multi-platform technology to place advertisements, track the effect of advertising data in the center through self-built big data, verify the effectiveness of invitation leads with OEM call centers and store sales, and evaluate the value of advertising.
- On the sales side, multi-terminal APPs and WeChat mini-programs for online car owner services, brand promotion, and online sales communication require the R&D team of the OEM to integrate the dealer system and cooperate to complete the data exchange of the brand and sales links and the unification of UI and UE standards.
- On the production side, through the integration of supply chain management system, manufacturing process system, vehicle test data, model data and simulation data on the vehicle product design side, user test drive feedback on the marketing side, car purchase intention collection and other data to achieve demand, design and production , Verify the closed-loop digital production process.
In the process of digital transformation, Autohome has gradually deepened its layout, and its service chain covers car viewing, car buying, and car use. Its positioning has changed from vertical media to connecting C-end consumers, B-end OEMs, dealers and other automobile The main vehicle trading and service platform. I have the same understanding as OEMs and car companies—looking inwardly, scale and business expansion mean higher organizational complexity, higher information communication and collaboration costs; Scenarios and higher-frequency iterations keep users sticky, while new businesses need to start quickly with the support of existing platform data and technical capabilities, which puts higher demands on the ability of the R&D team to deliver product value quickly and efficiently.
Under the joint influence of internal and external factors, the Autohome R&D team began to actively seek organizational-level efficiency improvements, hoping to play an overall effect and provide reference for the entire industry.
The Evolution of Autohome's R&D Tools
In terms of the use of R&D tools, Autohome has gone through the following stages:
The first stage: the company's initial stage, the speed of business development is the most important. R&D The business R&D teams of each department adopt tools or open source solutions from different vendors according to their respective needs.
The second stage: As the company's business becomes more complex, it is hoped that part of the management logic will be implemented in the tool to reduce manual operations. To this end, Autohome has tried to develop some self-developed tools for specific scenarios, such as project management tools, Kanban tools, and OKR tools. But the tools exist independently of each other and are spread out in dots.
Problems at this stage:
- The R&D tools of each business unit are not unified
. Previously, the R&D tool chain was built by each business unit or a department team subordinate to a business unit.
This aspect has resulted in extremely high collaboration costs. Some teams still use relatively primitive manual, email, or meeting methods for task assignment and status tracking; if cross-department/team projects are involved, communication and collaboration will be more efficient. difficulty. On the other hand, it is also difficult for excellent R&D practices to go out of departments and promote at the organizational level. - Business-industry-research cooperation chain is long
Due to the nature of the business, there are often cases where the same business service BC ends or needs to be coordinated with other businesses in Autohome. There are many sources of demand, and the length of the feedback cycle varies. The R&D department often accepts the requirements passively, complaining about frequent changes in requirements, low quality, and unstable quantity; while the business side often feels that the support of production and research is insufficient, and the progress is difficult to perceive. - The status quo of performance is not clear.
The tools are not unified, which means that a large amount of R&D data is scattered in different tools, and the dimensions are inconsistent. Without a credible data basis for measurement, it is difficult for R&D managers to objectively judge the overall performance status and make improved decisions.
At the present stage, the Autohome performance platform team has started to focus on one-stop platform construction. While improving the overall process and efficiency of R&D, it has also effectively formed the connection and data association between R&D and business departments, which also provides a basis for subsequent performance data. The extraction and analysis of provides a consistency guarantee.
The overall construction idea of a one-stop collaborative efficiency platform
The construction of a one-stop collaborative efficiency platform is not a simple system integration. The construction of Autohome's performance platform revolves around the following three features: standardization, visualization, and measurability.
Standardization: Precipitate the consensus of each team to ensure that the tool platform is usable and easy to use
Since the performance platform will eventually be used by various business R&D teams, it cannot be closed and closed from the first day of construction. The Autohome efficiency platform team conducted research on various R&D teams, sorted out the current R&D process, and then reviewed and abstracted the standardized R&D process. This set of standardized R&D processes needs to be approved by the front-line R&D team in order to ensure the availability of the final delivered tool platform.
The Autohome efficiency platform team first deposited a set of R&D standard procedures covering business, product, R&D, and testing on the collaborative efficiency platform. Through process specifications and process templates, the business R&D team can quickly get started.
In order to promote the smooth migration of business teams to the R&D efficiency platform, Autohome has done the following two aspects:
- On the basis of the standard process, support the personalized needs of the business R&D team
The Autohome Collaborative Effectiveness Platform introduces frameworks such as traditional waterfall and agile Scrum to meet the needs of projects of different stages, attributes, and sizes. At the same time, each team is encouraged to further refine the process specification based on the actual situation. For example, the second-hand car team prioritizes the standardization of the completeness of review materials in the demand stage, the definition of task disassembly caliber and bug rate in the R&D stage.
During the use of various business R&D teams, the efficiency platform team also continuously collects feedback on usage, accumulates best practices, and refines common requirements to make the platform more standardized and easier to use.
- Step-by-step promotion, so that the business R&D team can quickly feel the value
Autohome takes the kanban tool that has been promoted and used at the organizational level as an entry point, and drives each team to gradually migrate requirements management, project management, kanban management, document management and other capabilities to the collaborative efficiency platform.
In order to allow the business R&D team to quickly feel the value, the collaborative efficiency platform is also connected to the A-one release platform, which facilitates the team to manage the entire life cycle from requirement entry to release, and improves delivery efficiency from a global perspective. With the help of a unified tool platform, excellent R&D practices can also go out of departments and business lines, spread and promote throughout the organization, and quickly land in other teams.
Visualization: Focusing on "things", allowing business, products, and R&D to see each other
In the construction of the collaborative efficiency platform, Autohome takes the standardized management of demand as the starting point, and connects the three parties of business, product and R&D centered on "things". From the establishment of OKR, to the entry of relevant requirements into R&D, to the interlinking of functions on-line, so that the value flow can be visualized, colleagues on the business side, project side, and product side can directly perceive the overall progress of R&D and the progress of key states.
This not only solves the problem of long business-industry-research collaboration chain and poor collaboration experience caused by opaque information, but also helps the production-research team intuitively understand what R&D resources are invested in, and rationally allocate limited R&D resources with lean thinking.
Business: OKR-related projects, production and research progress transparent
In order to improve the information transparency of the whole process from business to R&D delivery, Autohome's collaborative efficiency platform has opened up the project management system and OKR system. When analyzing O (objective), you can not only see the supporting KR (key result), but also which projects are supporting the achievement of organizational goals, how many resources have been invested in these projects, where the projects have progressed, and whether they have encountered problems. to the bottleneck. The perception of R&D progress on the business side is enhanced, and the collaboration experience is also improved accordingly.
In terms of products: the needs are gathered in one place, standardized and online
Regardless of scattered requirements or project requirements, user requirements or product planning requirements, all types of requirements are unified into the demand pool for unified scheduling, classification, integration, and distribution. And the business team can view the man-hours spent, resources invested, and the overall progress of the requirement through the requirement.
This mechanism can drive product personnel to actively optimize demand quality, invite R&D Leaders to communicate in advance, reduce changes and repeated clarifications, and improve the efficiency of production-research cooperation.
R&D: Statistical tasks and working hours, reasonably matching supply and demand
On the one hand, the task dismantling and time estimation caliber are uniformly agreed, and the demand level is quantified by the number of tasks and task size; on the other hand, the effective coding output time of R&D personnel is counted to quantify the R&D resources on the supply side.
Through the standardized measurement of both ends of the supply and demand, it is possible to accurately understand which needs R&D resources are invested in at the granularity of hours.
This brings three benefits: first, both the product and the R&D can use this to know whether the current focus of the R&D team meets expectations; second, the R&D team can comprehensively consider the R&D bandwidth and demand priority, and plan and allocate tasks reasonably; Third, scattered small tasks can be interspersed with schedules to improve the utilization of R&D resources.
Measurable: Measurement is not the goal, improvement is the goal
Based on a unified collaborative efficiency platform, Autohome collected process and result data with consistent dimensions and high reliability, and began to gradually establish a R&D efficiency measurement system to quantitatively evaluate and analyze the current status of efficiency. At present, Autohome mainly focuses on the delivery capability, efficiency and cost of R&D.
Metric building step by step
The first step is to open up the collaboration system, business system, support end, and basic system data chain, and bury points in the entire process of R&D tools to collect performance-related data.
The second step is to establish the Autohome R&D indicator library, classify and display according to different dimensions such as delivery efficiency, delivery quality, delivery cost, delivery capability, and continuous improvement, and make a visual display of the caliber and graphics of each indicator.
The third step is measurement analysis. First, analysis reports and optimization suggestions are made according to the dimensions of organization, project, and people. In addition, some teams want to see some more detailed and personalized indicators. Later, they will build custom reports, which can be quickly Find and put it on your own Kanban board, and perform statistical analysis by yourself. Present it to teams and business parties in the form of reports
. The fourth step is to avoid measurement for the sake of measurement and let measurement drive insight and improvement. After the R&D team discovers the indicators that need to be focused on, it will drill down to locate a single point with a large room for improvement, such as a certain team or a certain period of time, and organize a split analysis meeting to discuss optimization suggestions and improvement directions.
Index selection and practice
Focus on Indicator Portfolio
At this stage, Autohome pays more attention to indicators such as delivery efficiency, delivery process quality and delivery capacity, and demand throughput rate. In the design process of the measurement index system, it is necessary to pay attention to the combination of different types of indicators: select global result indicators for evaluating capabilities, local process indicators for guiding analysis and improvement, and lagging indicators for post-mortem review.
for example:
- Demand delivery cycle and demand throughput are global result indicators, which can be used for overall evaluation of delivery efficiency.
- Delivery time at each stage, requirement change rate, requirement review pass rate, and defect resolution time are local process indicators that can be used to guide analysis and improvement.
- Online defect density is a lagging indicator. Online defects have already occurred. All we can do is to review and analyze the root causes of defects, and strive to improve the quality and indicators in the next statistical cycle.
Through the review of R&D members and the accumulation of experience and knowledge, the measurement index system can be expanded in a targeted manner.
For example, we found through the analysis of the review that the low demand throughput often comes from too many WIPs (Work in Progress) being processed at the same time, and members need to switch contexts frequently, resulting in a decrease in efficiency; while too many WIPs are processed first The indicator is rising demand backlog. Therefore, when we try to solve the problem that the number of requirements actually completed by the iteration is less than planned, the quantity of work in process and the quantity of requirements backlog can also be included in the measurement.
Introducing code equivalents
In addition to routine data such as working hours, requirements, and tasks in the project management process, Autohome has introduced the industry's advanced coding link workload index—code equivalent—as a basic index through SMAY.
On the one hand, the code equivalent index can squeeze out the water in the code, avoiding large fluctuations in workload caused by code movement, addition and deletion, etc. density etc.
On the other hand, the code equivalent index directly analyzes the logical complexity and importance of the code, does not rely on a standardized and mature R&D process, and is not affected by incomplete, inaccurate or timely manual data recording, and is more robust. As Autohome is being built and migrated to a unified platform, R&D practices have been adjusted, and code equivalent indicators can be used to detect data deviations in a timely manner, attracting teams to pay attention to R&D process specifications, and ensuring credible and healthy performance metrics.
For example, when the code equivalent is significantly low or high, it is necessary to combine the required throughput and man-hours to drill down to analyze whether the coding workload is reasonable. Behind the unreasonable coding workload, it may be the quantity and quality of demand, insufficient effective coding man-hours, insufficient skill stack, or untimely submission and other R&D process specification issues.
After the pilot team drilled down and analyzed, it was found that there is room for optimization in the effective coding man-hours and demand quantities. By adding a round of communication between the product and the R&D Leader before all R&D staff participate in the demand review, and aligning information in advance, the pilot team reduced the per capita meeting time by 20+ hours per quarter, and increased the one-time pass rate for multiple product requirements, reducing production and R&D costs. Collaboration costs allow developers to focus more on coding.
The current research and development efficiency analysis of Autohome is still in its infancy. Performance index data is becoming more and more unified, enriched, and standardized, and it also supports data drill-down; through the collaborative efficiency platform, we can assist business teams to improve efficiency and reduce costs, and at the same time promote the continuous iteration of our tool ecology and R&D methods. It is our value.
write at the end
In order to cope with the increasingly severe internal and external situation, Autohome will continue to deepen the research and development of digital upgrades, arm the team with advanced research and development tools, build an efficiency management cockpit, drive data, reduce costs and improve efficiency, so as to win greater gains in the next competition initiative!