I have met the following opinions from the developers. Many people think that once the major players from various segments of the market (Zynga, Facebook, Airbnb) do your Analytics, it really is the best way. Its not so easy — it has pros, cons and pitfalls.
So, you have developed their own Analytics system and created a Department that works with her. Imagine that every employee is a marketer, PR specialist, game designer, and even producer will ask the analyst questions from the. “What yesterday was DAU?”, “How many in this DAU were users from Russia?” “How many new users came to the country”. Analytics is easy to answer — it is only one SQL query or a couple of clicks in the system.
But you are hindering the activities of the specialist (who certainly has more important tasks), and the employee who asked the question. Note that every time such issues more than one, and it certainly does not benefit the company. Much more convenient when each employee can log into the system and to find the answer to your question. This is what is called data-driven, all know what data they need, know how to get them and do not distract other people.
Product Manager Instacart Farid Musavat faced at the time with the same choice, said. “If you consider yourself a data-driven, but all requests go through Analytics, youre not data-driven”. These costs are not equal to the sum of the salary of the analyst and the costs of maintaining servers with the data. First, in this situation, the companys development is hampered by the small tasks, which are constantly falling on the head of analytical Department — costs that are difficult to consider and evaluate.
Second, not all can be taken into account at the design stage of the system, if you write it yourself. More complex queries will require new functions from the system, and hence new data structures, and hence pulling engineers from the development of the primary product in the development of the Analytics system. Imagine, for example, how much will have to change the system, if you decide to do A/B tests that are not included in the original project.
Over time, the system acquires new costs and requires more resources that are taken away from the development of the product. Farid Mosavat said. “Im more interested in addressing critical issues of product than in building a technical infrastructure for Analytics”. In one article on this topic says.
“It professionals dont have to do the work of librarians who are looking for data, and properly interpretiruya for each request. Besides, engineers will need time to create and maintain data structures”. When the third-party system analysts the company product development is easier and faster. Here are just some of the functions of the analytical services that will require significant forces in the development, although solve a simple problem.
“But what about customization?” you ask. After all, if we write the Analytics system from scratch for your project, it will take into account all of our features and be able to answer any question. Indeed, can. And this is the main advantage of self-created systems.
But third-party Analytics something offers. As a rule, it is possible to download data incsvxlsx and so forth. Using this data, you will be able to answer any question, save on content servers. You could give the impression that I insist on buying third-party Analytics system, but this is not true.
I tried to compare in-house and ready, but the decision to accept you. Below is a list of questions to ask yourself to select the best option. Finally, some specific recommendations on how to proceed.
In conclusion, a simple metaphor. Imagine you want a kitchen set. You can do it yourself, because a lot of people do. Like even get cheaper (fanerki Yes screws) and is perfect for your custom kitchen. But before you start construction, go shopping, perhaps I could actually enjoy it.
Read more: google.co.uk/blog/in-store-analytics/