Quarantine was a kind of litmus test for online service systems. Many systems, even government services, could not withstand the load (and distance education is generally a separate song, some families might not even survive them). Many systems were functionally unprepared for mass service. Conducting a retrospective, now on every corner they began to write about the importance of online directions for stores, that we had to think about this earlier. 2 months of active online orders and the gradualness of the subsequent release could well radically change the preferences and purchasing model of residents of large cities.
Now IT can go into business and require tens and hundreds of millions to implement or develop trendy online systems. But will it all be justified? Without building full-fledged customer outflow management (what was called 'customer churn prediction' in the telecom), the effectiveness of the money spent will be a big question and why.
There are two commonly used marketing points in CRM topics:
- The cost of attracting a new customer is 10 times higher than the cost of selling to an existing customer.
- A satisfied client leads two, and a dissatisfied leads ten.
If this is not immediately taken into account in projects, there is a big risk that as a result of all these actions, the business will pay IT a lot of money, the quality of the product will be mediocre (at different stages). We, now no longer as IT employees, but as consumers, will pay for this by increasing the cost of production and by the very fact of use we will agree with the ever-declining bar in service quality. I would not want to.
I bring a specific case confirming the problems. Cases in different stores have accumulated a lot. ' I take ' the most interesting, because it is a product of the two largest Russian IT companies (how they position themselves) - Yandex + Sberbank . And from him, an appropriate quality of study is expected.
We make sequential orders:
190**460 — CR2032. . . . 2 .
? 5 9. "" . . . - ? .
198**674 198**931. . , . -. , "" , " ", , - . . , "" .
? ? ? ? NLP? AI/ML? DS? R Python BigData ?
, , "- ", 90% - (Wikipedia)?
- , . "" .
? . . , . ! .
, , .
, , . , . . - "" .
, - , - , , , QR ..
"" .
!!! Data Science , "", , .
— R, .
, , . , , . . . .
P.S. , " " , . , . — , .
The code is not here because it is, in fact, the formulation of the problem. And then there are two options:
- To fix the significance of problems that have long been well known to real business, and to solve them using well-known methods and algorithms.
- Leave everything as it is and consistently and methodically slip into the degradation of the services provided to us all. We ourselves will consume what we have done with our own hands.
The second scenario is somehow not very happy.
Previous publication - "The use of R in calculations with increased accuracy . "