Estimating the cost of a company's IT systems
Many, especially in large companies, ask themselves: “How much is spent on this or that IT system?”. The question is quite interesting, and the answer to it is not always simple. Complexity is caused by both the complex cost structure and the management of complex changes that affect many IT systems.
We also set out to calculate exactly how much money is spent on which IT system.
Formulation of the problem
Our main task was to determine how much money a company spends on each IT system. In other words, we tried to calculate the TCO (total cost of ownership) by correlating all of the company's IT costs with IT systems (which we understand as IT systems can be found in this article ). Only the expenses of the Department of Information Technologies took part in the calculations, and the costs of AHD, rent and payroll were not taken into account in the calculations (there is a desire to include them, but next time).
Analysis of the current state
The first stage is an analysis of the current state: how is the payment of IT costs, what conclusions can we draw from this and how to use them in the future.
The analysis showed that the payment documents do not contain accurate data on which system the payment is being processed. According to some indirect signs, individual payments could be equated with a certain system, but a single approach could not be built. It was not possible at all to correlate payments with business tasks. In most cases, payment was made for the work of the team over a period of time. As an example, for many systems that use release delivery, the final payment documents are issued for the entire release, and not for each release task.
The second, no less important question: is it possible to build statistics on the life cycles of a task or project? Unfortunately, the adopted budgeting system did not imply accurate management of life cycle costs. Those. At the first stage, this idea was also abandoned.
All of the above related to consulting costs. But the bulk of the costs associated with the infrastructure, it was not at all possible to spread on IT systems. The thing is that most of the activities related to the purchase of equipment and work in the data center were carried out in a single amount. Those. colleagues from the Department of Infrastructure Operations gathered all the needs into one big activity and spent their expenses within the framework of infrastructure activities.
Company Budgeting System
Now we need to talk a bit about the budget structure of the company.
The company's budget is divided into the following categories:
- Directorate to which the budget belongs.
- The area of activity to which the costs relate is the financial position (for example, project consulting, telephone subscription fees, rental of main communication channels, etc.).
- A project is a combination of costs into some kind of big activity.
In the sections described above, the initial budget is allocated, which is issued to the directorate. Then, documents with payments to contractors begin to be carried out - they list the materials supplied and the services rendered, the numbers of contracts and the numbers of contractors.
In general, we have a planned budget in the context of: management, financial position and project. The actual budget in the context: direction, financial position, project, supplier, contract and material.
Formation of an approach to costing IT systems
To solve all the problems described above, we have developed a system for assigning payments to IT systems. It is based on the methodology described below.
At the first stage, we decided to divide the allocation of costs to IT systems in accordance with the direction of activity. To do this, all areas were divided into groups:
- IT Systems Consulting Costs: Costs associated with maintaining or making changes to IT systems.
- Infrastructure Costs.
- Other expenses.
IT systems consulting costs
Here we took this approach:
- At the old expense, i.e. last year’s expense, we analyzed each payment separately and determined the percentage of how much it relates to a particular IT system of the company. The work was not so big, it was necessary to look at only a few hundred payments, it took several weeks.
- For new expenses, we obligated each manager to indicate the IT system to which the payment relates when paying. If the payment refers to several systems, you must either divide the amount by these systems, or indicate the share ratio of each IT system in the total payment amount.
Since, in most cases, infrastructure costs are taken into account within the framework of large activities, we did not try to correlate each payment with the IT system. Instead, all infrastructure costs were attributed to infrastructure services (we have 26). Next, we came up with a methodology for allocating the collected costs to IT systems in accordance with some metrics. For example, we assign server costs in accordance with the kernels used by IT systems. Those. we consider how many total cores all IT systems currently use, then we calculate how much we spent per year on the purchase of all servers, and calculate the cost of one core. Next, we distribute all these costs in IT systems.
Approximately the same method was introduced for all other infrastructure services.
I agree that the model is very rough, but taking into account the use of virtualization and frequent changes in infrastructure, it is very difficult to come up with something more accurate and not requiring a lot of effort to account for.
The model is implemented in the form of an Excel file, all parameters are written into it, and costs related to infrastructure services are distributed among IT systems. In the open spaces of Habr, I was surprised to find an article from a contractor who helped us with this methodology, which describes their view on our task.
In the IT department’s cost structure, expenses were found that are not directly related to IT systems, for example, team-building events. After some deliberation, these costs were excluded from the analysis. And the amounts there are small.
It takes a lot of time and effort to manually correlate costs and reformat reports. Since our company has a budget reporting system, we implemented distribution rules in it. And as soon as payment data arrives from the ERP in the reporting system, its data is automatically distributed across IT systems in accordance with the prescribed rules.
A few results. After implementing all of the above, we were able to distribute 94% of payments by IT systems and infrastructure services for 2018, and about 90% of payments for 2017.
Since the cost structure is more or less constant, it is not often necessary to establish new rules, and it takes a little effort to maintain the system. But we got not only the expected statistics, on which systems how much money we spend, which contractors for which systems work and which systems were affected in different projects. There were interesting “discoveries”.
After introducing the described methodology, we were able to distribute 94% of payments for 2018 and about 90% of payments for 2017 among IT systems and infrastructure services.
Since the cost structure is more or less constant, it is not often necessary to establish new rules, and it takes a little effort to maintain the system.
A reasonable question arises, why do we need this data? Thanks to them, we started the following activities:
- We started work on category strategies for each IT system and provider. That is, having found out the cost structure, we can more accurately understand the distribution of costs for each supplier. This, in turn, made it possible to single out the total volume of work of each supplier in each system, and to begin negotiations with suppliers on planned volumes for a year in order to obtain more favorable conditions. Since previously the total historical volume was not always clear, we held separate negotiations on each system, and each project could have its own conditions.
- Having data on the costs of systems, we began to introduce the life cycles of systems and develop strategies for maintaining these cycles. For example, some of the systems in our IT landscape are already considered obsolete and are planned for decommissioning. But the model shows that we are still spending quite a lot of money on them. For some systems, this was considered unreasonable and redistributed work to other projects, and in some cases we decided to accelerate migration projects.
- A separate issue is the cost of support. Based on the model data, we found out the total costs of supporting systems: licensing fees and the cost of non-vendor support. Now we are developing strategies for optimizing the cost of supporting part of the systems.
Another, albeit indirect, but still very interesting result was the creation of a budget report with a plan-fact analysis. Before the creation of the model, such reports existed, but they were fragmented, and the set of analysts in them was very small. When automating the model in the reporting system, we created a report for the plan-fact analysis, which included the maximum set of analysts. Now managers can build the reports they need in a more convenient tool (this could have been done before, but the task was postponed every time as not a priority).
At the moment, we have identified two areas of development:
- Creating a driver budget model for IT .
In the framework of this area, we are creating a budget model of DIT, which will allow you to get a planned annual budget based on the input of IT drivers. We can also protect and adjust the budget based on changes and viewing drivers, and not just based on expenses. The project is now in an active stage, and I plan to write an article based on its results.
- Creation of a unified methodology for maintaining IT assets .
One of the main problems in correlating the costs of IT infrastructure with IT systems in our company is the lack of a single database of IT assets and a complete map of the relationships between IT assets and IT systems (in addition, there are methodological discrepancies, which is understood as IT assets) . Now we are planning to start a large project, within the framework of which we will carry out a complete inventory of the infrastructure, as well as work out methodological issues of maintaining and accounting for IT assets. I would like to talk about this project, but there are doubts that this will be of interest to many, because the topic is very highly specialized.