Companies are gathering more and more information about users. But it is not enough to get it, it is important to understand exactly which data will be used and how. We invite product and data analysts at Avito Product Analytics Meetup. Let's talk about how to work with the data warehouse and scale it for high loads, about site customization for users based on customer journey map and measuring user satisfaction. Under the cut - abstracts of reports, links to registration and live broadcast.

Reports

Analytical storage Avito.ru - from large to very large data. Artem Danilov (Avito)

At first glance, designing a data warehouse is a typical task: to collect requirements, build a data model and implement ETL. However, the avalanche-like growth of data volume and their complexity can easily disrupt the storage operability. The report shows:

- how, following the expansion of business, Avito’s storage has grown to 76 TBytes over several years;

- where does the company have big data from;

- what are the main principles of analytics;

- how the Avito.ru repository evolved.

Particular attention is paid to the answer to the question of why following the sixth normal form constraints (Anchor Modeling) helps to quickly expand storage, effectively scale the load in the MPP DBMS environment and develop in terms of microservices.

Personalization of the site for the user based on the customer journey map. Alexey Chernobrovov

From the report you will learn about what is a customer journey map (CJM). The main approaches to the automated allocation of patterns of user behavior using modern machine learning methods will be described. Alexey will talk about building personalization based on CJM and choosing the right metrics for evaluating product improvements. As examples, the application of a large bank, the site of a well-known car dealer and a large online store will be used.

Avito NPS: the voice of users in product development. Danila Lenkov (Avito)

In any large company they know that the secret of a successful business and a bright future is the happiness of customers. But measuring it and learning how to work to increase the level of happiness is not an easy task. Worldwide, it is solved through customer surveys. The most successful method of such surveys is called Net Promoter Score, or NPS.

Avito is an extensive business with a lot of business verticals. Therefore, the task of building a survey system is even more difficult than usual. The report will tell the success-story about the implementation of NPS in all parts of the product and show how a large flow of votes from our users helps us make effective decisions and constantly improve.

Avito polling system features:

- Thousands of users per week from all platforms: site, mobile site and applications;

- the decision to show the desired survey option is personal for each user and is made directly online;

- automated pipeline for collecting results and delivering them to reports;

- classification of text feedback for prioritizing solutions.

How to increase conversion with smart sorting of issue. Case Ostrovok.ru. Evgeny Shneiderman (Ostrovok.ru)

The problem of sorting goods in the list of issue is a classic task of modern internet business. In Ostrovok.ru, too, they have long been concerned about the issue of issuing optimization and have done a lot of work along the way. In the report, Evgeny Schneiderman will tell about the main stages in the development of the hotel sorting logic, describe the problems and mistakes that the team faced, as well as the conclusions that came out to be drawn.

• Why is the issue order optimization the most powerful growth lever for conversion?
• What can be approaches to the logic of sorting, what are their pros and cons?
• Why are such solutions difficult to provide without a high-quality Big Data infrastructure?
• How to structure the project team to ensure fast iterations and improvement of product data science?

All this will tell on Avito Product Analytics Meetup.

Cross-product effects, or how we evaluated the contribution of "Electric". Ekaterina Loseva (Tutu.ru)

Many companies offer several products to the market at once. A special place is occupied by the case when there is a product in the portfolio that does not bring revenue directly. How to measure its impact on sales? How does the company benefit from the existence of this product? In her report, Ekaterina Loseva, an analyst at Tutu.ru, will talk about the evolution of the company's approaches to calculating cross-product effects metrics using the example of the “Trains” section.

Passwords and attendances

Participation in the event is free, registration is required .

Collection of participants: 12:00

Reports start: 12:30

Address: Avito office, Lesnaya 7 .

The broadcast will be organized on the day of the event on our youtube-channel AvitoTech . Subscribe to not miss anything!

See you!