How did we fail to remake the architecture of the company
Last year I began to tell how we were fat, and then refactored. Let me remind you, then we chose the path of tactical half-measures: we divided a large company into 4 encapsulated objects and, therefore, greatly reduced the chains of management. At the beginning of this year, it is time to finally do the right architecture.
This turned out to be a very nontrivial task, and we did not solve it to the end. But they discovered a lot of new useful things in the process. For example, we already understood that there should be two IT departments in a company: tactical and strategic. Tactical is helpdesk, iron ore, tracking of resources and licenses, monitoring and in general everything that repeats more than 2 times. Strategic - this is the implementation of major features, planning for 2-3 years ahead and finance.
Then it turned out that the same applies to the personnel department. Tactical personnel deals with all documentation, leave-inquiries and hiring line personnel, while strategic personnel, for example, hunt, and train employees who in six months can become managers.
But let's start with the methodology.
It is assumed that you need to first understand what your company is doing and why. In general, the main question is always “Why?”. As a tool fits a normal mission that will determine the principle of decision making. Strange as it may seem, the future structure is formed from the mission. Because it is a document that describes the desired state, acceptable and unacceptable ways to achieve it. This is me about the mission, which was done with my head, formalizing the essence of the company, and not about the document "so that I could hang on the site."
I will simplify greatly now, but we decided that the most important thing for us is to deliver his favorite games to any person. To solve this problem, you can carry the best games of the world, you can make great games in Russia, you can do retail and online commerce, you can completely give up your retail and trade through someone, etc. The main thing is that you personally have the coolest games. The second vector is to make an honest profitable business. From zero up to the present moment it turned out (well, profitable - with interruptions), it would be foolish to throw such a challenge. This partly makes our life a game of hardcore-complexity)
These two goals plus the existing developments (strong brand, strong retail, customer focus, unique marketing), sets a certain coordinate system for building the company's architecture.
It remains to determine the specific measurable goals for 5 years. They will be planning from above: they will be priorities for the placement of individual modules.
Next you need to take and develop the company into blocks that produce some functions. It usually happens a lot of controversy, but somehow it turned out that almost everyone knows how to determine tasks (that is, they have at least basic skills to algorithms), so it happened quickly.
In the methodology, we proceed from something like container virtualization (microservice structure).
There are three types of objects in the company:
• Generating profit.
• Producing something with an understandable unit price.
• Producing something shamanic.
First example: retail (like a big top-level abstraction) generates a profit. You load the goods into it at the entrance, it grinds it and gives money at the exit.
The second example . Suppose we need to make a game based on already prepared and tested layouts. The production of a game (printing, assembling, purchasing components, molding plastic, painting wood, etc.) costs money. In theory, it can be ordered inside a group of companies in-house, or outside in some external printing house or several.
It is very simple to compare: the price per unit of game of our production is known, the quality criteria and SLA are known. If someone makes it cheaper with the same quality and SLA - obviously, all production calls can be redirected to this other microservice. Just replacing the module in the architecture. The only point is that it is important to take into account in the assessment such factors as “they always have the priority of our products” and “they will not go anywhere in 3 years”.
Anything that can produce a product or service at a reasonable price and with a fixed SLA is simply a counterparty, which for some reason is located on the company's territory. As in the story about a single manufacturer of wooden windows: they faced the problem of lack of material in the season and made their sawmill. Two years later, their business realized that it was more profitable to sell boards to other companies, and they were faced with the problem of lack of materials, for which they launched another own sawmill ...
It looks like a factory canteen that can sell lunch and someone outside, not necessarily a worker . Or the accounting company, which first serves several different legal entities in the holding, and then begins to trade outside. We have at one time IT outsourcing and contextual advertising specialist stood out.
Third type- the production of something shamanic is, for example, PR. When there is no price per unit, and the object is made holistically and magically.
For example, the work of enikei is a second-class object, and the work of a CIO who thinks about the development of a company can already be third.
We connect the modules!
Then everything is simple: you need to understand the chain of production. We take each module and write out who needs it. The task is to close all this on the profit generation modules. They will pay for the work of the services of the second and third type. The structure of processes will turn out (more precisely, while mutual requirements and budgeting).
For example, retail consumes IT support, tactical marketing, production of printed materials, logistics, etc. (but does not consume strategic marketing). Logistics module, a symbiotic retail, you may need more and wholesale, etc.
Retail puts clear SLA and pays for all this within the company. Or the retail has agreed at the beginning of the year with the event department about how many events and what is necessary - the retail has given out a budget for this.
And then the magic: the head of the event department reports not to someone abstract above, but to his internal customer. And tries to do better to him. Or simply conducts corporate events himself (the event earns it). Or trying to do everything at once. This is how the variance of management, the variance of responsibility and very, very good decentralization. And it determines the rate of change of the company.
In general, the whole structure of such a plan is sharpened for three things:
- Handling thresholds: one person cannot manage more than 7 subordinates. That is, the CEO can not have more than 7 heads of departments in the company. If there are 10 of them, it means that a part must be clustered in the “every kind of garbage” division (it is often called “strategic development”) and put one director on the garbage. That is, for development.
- Flexibility and speed of change. It's damn important to change quickly. This design requirement defines the structure. Decentralization of decisions is put in the first place. Modern business does not change from release to release, but constantly mutates without stopping.
- Understanding that tactics always devour the function of development. This means that it is necessary to distribute the modules that ensure the maintenance of the work of something and the modules that move the company forward. The same PR is, in general, as a commercial tool, no one in the company needs anything because it works as a multiplier. Well, if good PR is easier to negotiate with partners. Easier to close vacancies. Easier to sell. But it is difficult to assess its shamanic effect, so it is placed at the level of subordination to the profit-generating CEO module (in our case) and sits there as a strategic resource. Therefore, many divisions are divided into tactical and strategic. You can’t simultaneously think about sales today and development, for example.
That's about it. We have not yet gone into the new structure, because the transformation will be long. But now we are thinking of portals, and we understand how to recommute everything in the company. Next year we will grow and re-do the process.
But we did not succeed because we invented a structure that will be in 3 years. Then the intermediate structure that is needed for transformation. Then they tried to think up a plan of changes, and realized that we would not pull. We ran a couple of times over the processes, and it became clear that we still do not fully know what we are doing exactly: the possibilities of the sea, but some of them need to be cut to focus on two or three really important things.
Right now we are sitting and thinking.