When is the time to think about implementing a BI system?

    In this article I want to share my personal observations of this process. How companies go from the item “We need enough standard reports in the corporate accounting system” to “Preparing reports requires a lot of time and resources. It's time to automate everything! ” I hope that the foregoing will help someone avoid some errors and choose the right Business Intelligence solution (BI platform).

    Stage One . Prelude.

    The company's management has a need for regular reporting (sales, plan fact, financial condition, etc.). Reporting is prepared by specialists of the relevant business units (financial unit, commercial service, logistics, etc.). In these units, people are forced to combine their main responsibilities (keeping records, maintaining contracts, etc.) and preparing reports. Of the tools they have standard reports in the accounting system and Excel. Moreover, using Excel, they usually do not have everything in order.

    Stage Two . Excitation.

    Then the managers' appetites grow (the company grows, the number of managers grows, managers come with a new outlook on business, etc.), and they begin to request more and more one-time (ad-hoc) reports to look at the business from different angles. With the growth of the company, there are more and more such reports, some of them become regular ones, and business specialists have problems with preparing all this variety on time. In search of salvation, they begin to demand from IT to take part of the work on themselves, namely, they ask for various uploads from the database of the accounting system (CSS) and are increasingly turning to requirements to develop new reports in the CSS.

    Stage Three . Conception (Embryo).

    After all these processes, the company begins to form a line called the analytics business. So, a SQL developer may appear in the company (from this I started the way) and a specialist / you who own Excel and other Office programs. Development at this stage can take place in different ways. I personally saw that over time, the number of analysts can become quite large (each unit acquires 1-2 specialists or a separate unit is formed in the company). Incidentally, I could be in this role, but I was lucky that I was not taught Excel at the university and at the first interview my aunt from the HR department said “ah-ah-ah”. My assurances of her that I quickly (in a couple of weeks) will master this product, most likely, gave rise to the thought in her: “Obviously in front of me is a confident cock, for here other aunties here have been working on the computer for years and are still afraid of this beast, and this impudent shket declares this. ” Personally, I was quickly bored with creating various unloadings, and I began to wonder what was suitable in the market. But in 2006, I did not even know the term BI, so the search was short-lived. As a result, I stopped at OLAP technology.

    Stage Four . Getting rid of IT from ad-hoc or the birth of OLAP. Beginning of the BI project.

    It seems to me that OLAP is already a fairly common thing, and with a high probability there are people in the company who work with OLAP cubes as users or developers. They sow the idea that the introduction of cubes will be a deliverance from many problems and make life easier for a large number of employees. Or this person had experience with a certain BI system. Since now it’s more likely not about the largest companies, people offering something from SAP Business Objects, IBM Cognos, Oracle BI will stop listening when they see the price tag. In large companies, for a long time there is something, at least Microsoft BI (SQL Analysis Services and Reporting Services). “At a minimum” is not a neglect, it’s just a fairly common solution, as it comes bundled with the database server, which often leads to the choice of this particular platform.

    Since we all want our project to grow good, kind, strong and healthy, it is important not to make a number of typical mistakes.

    First mistake . Spontaneous system selection.

    How does this happen? Since the same person can be either a high-ranking person or a very pro-active IT specialist who prefers a certain product, there is no time to choose a system, but just buy what these people lobbied for. This can lead to the fact that the selected product does not really take root in the company and, having solved some problems, replaces them with others.
    I think that before the choice they should answer at least the following few questions (I wrote more about this in the last article ):
    • Who is the system for (who is the main consumer)?
    • What are the company's preferences for ways to deliver data to consumers?
    • What amounts of data will have to deal with?
    • How much is the company ready to pay for all this and spend on further development and maintenance?

    I would also like to add here to add a question that I had previously silent about:
    • How much functionality can be divided in the system between IT and the analysts responsible for preparing presentations, analytical notes, etc.

    I will explain the essence of the issue. In some systems, you can quite simply divide the functions as follows: IT is responsible for the development of data models, and already analysts are engaged in dashboards, reporting forms - what end-users see. Why is this? Requirements for the appearance of reports often change most often. Plus, the same report (data set) can exist in several variations for different customers. In addition, aesthetics are still important in the design, to which there is not always a case for which IT does not have enough time, their job is to make it all work. This is especially important if you have good developers with a BIG salary.
    In order to avoid the problem, therapy must be started in a timely manner. Well, if at the 2nd-3rd stages, IT systematizes the needs, answering the first question and pushes the business to the question of introducing BI. At these stages, customers are enthusiastic and ready to discuss different options, and most importantly - to formulate their requirements, which, in my opinion, is the main task of business users in this project (again, see the previous article).

    The second mistake . DWH or its absence.

    It happens that before the implementation of the automated reporting system itself, no one bothered that it would still be nice to have a data warehouse (DWH). As a result, this leads to the fact that OLAP or reports refer to a large number of data sources, various storefronts that were formed at the 3rd stage. From this comes chaos. Developing and maintaining the system after this becomes extremely difficult, and it may turn out that everything will have to be rebuilt.
    Choosing a system for DWH also needs to be given separate time, but often it is the same DBMS on which the accounting system (CSS) works. This approach is justified, as the company already has product specialists and can save on licenses. Of course, this does not apply to the case when a lot of data is accumulated in the USB (a large number of transactions and the DBMS were chosen for this) and it is better to look for another DBMS that is more suitable for analytic tasks (there are column storage mechanisms, placing tables in RAM, etc. .P.).
    Another option is the integration of the accounting system with a certain BI tool (I saw the offers 1C + QlikView, from some integrators), but here I am not quite aware of how everything works. I would be glad if someone writes about this in the comments or in PM.

    The third mistake . Ignoring the fact that the project had long had to start.

    Often companies get stuck for a long time in the second stage and this leads to the fact that there are many different data marts, heaps of Excel files with macros for data processing and, of course, no one systematizes and describes this in any way. Starting as a result of the BI project, you will spend much more time systematizing the requirements, and you will encounter some resistance from users who have long been accustomed to what is. For example, they don’t want to work with cubes on their own, because Vasya from IT always made the necessary unloadings for them himself, thereby spoiling them greatly. Everything must be on time!


    In conclusion, I would like to say that it’s good when you can spend time meeting with integrators to get acquainted with various systems, but I saw this only in a very large company, in which the cost of the DWH + BI project was estimated at hundreds of thousands of dollars. Here was the interest of management (responsibility for the budget), and the interest of implementers (large order). And it lasted all this year. In small companies, this is simply not possible. Unfortunately, I have not met any integrators (perhaps I’ve looked poorly) providing extensive expertise. Usually they concentrate on 1-2 products. He personally communicated with 10 integrator companies in approximately the following way: 2 engaged in Microsoft BI, 2 - QlikView, 3 - Cognos BI, 1 - Tableau, 1 - Cognos and Microsoft, 1 - QlikView and Tableau.

    In general, be prepared for the fact that the responsibility for the choice and the fate of the project will rest with you - the readers of this article, which, I hope, will help you at least somehow.

    Also popular now: