New Customer Segments Help BUTIK. Improve LTV

About BUTIK.

BUTIK. is an online store for clothes, footwear, and accessories. It’s also a huge mall in the center of Moscow, covering over 45,000 square feet and providing 350 brands for customers’ choice. The range of products online matches the product selection in the offline store. This allows any customer to order products online and collect them in the mall an hour later, as well as choose a courier or pick up delivery to any city in Russia. BUTIK. is an alternative to big shopping malls that require a lot of time for choosing the clothes you’d love to wear. The company has become the first to combine online and offline stores in a single servicу, transforming into a truly omnichannel business.


To reduce ad costs, increase the customer lifespan and the LTV of the whole customer base, BUTIK. decided to:

  • Segment customers based on their buying activity.
  • Personalize communication with obtained customer segments, using digital communication and direct marketing channels (email, sms, call centers).

The company uses Google Analytics to collect and store user behavior data, and utilizes CRM (Microsoft Dynamics) for the information about completed orders. To segment customers, BUTIK. needs to combine the aforementioned data in a single system, based on customers’ buying frequency. The company can send the obtained segments to ad services and use them for displaying relevant ads and personalizing direct marketing communication.


We started working with the OWOX BI team about a year ago. Our goal was pretty standard way back then: our company needed to build custom attribution models. We asked the OWOX BI specialists to consider the fact that BUTIK. is an omnichannel business with multiple ways to communicate with customers. This means that we had to evaluate the contribution of each channel by analyzing its input into sales and CPO with the account of the ROPO effect.

Having achieved our goal, we had all our data brought together: the information from Google Analytics, ad platforms, and the CRM system. This gave us an opportunity to accomplish even more, and to segment our customer base. The OWOX BI team became a part of this exciting project and helped us find solutions to off standard tasks, allowing our company to successfully implement our targets.

Yana Parshutina,
Web Analyst at
Yana Parshutina


To achieve the goal, BUTIK.’s analysts combined all the information in Google BigQuery. Next, the OWOX BI team helped them process data and build reports. Let’s have a closer look at how the company achieved its goals.

Step 1. Combine data

The company chose Google BigQuery as a cloud warehouse for combining the data, because of the high security standards and simple integrations with other services.

BUTIK.’s analysts use OWOX BI Pipeline to send raw unsampled data about user behavior to GBQ, in near real time.

With the help of the API and Client Libraries, the company imports the following data from CRM to Google BigQuery :

  • Information about all orders, including the completed orders (online, offline, and through call centers).
  • Customers’ UserIDs, along with their personal data (name, gender, birthday, email address, phone number, registration date, loyalty program status, email and sms subscription, etc.), as well as the information about customers’ buying activity (a number of orders made by a customer).

Below is the BUTIK.’s data flow chart:

BUTIK.'s data flow chart

Step 2. Process data

After combining the data, analysts from BUTIK. and OWOX BI started building segments. As BUTIK. is an omnichannel fashion retail business, the analysts decided to create their own segments with additional custom parameters, instead of RFM segments.

To calculate the time period for segmentation, analysts used the consumption cycle time of the customer base, which is 1.5 months ± 2 days. This value is a median number of days between the two neighboring orders. To check this median number, the analysts calculated the number of days between the online orders, then the number of days between the offline orders, and got the weighted mean value for both types of orders.

Next, the analysts identified the main segment types, based on the calculated time period for segments:

  • New Members — new registered users who made no purchases.
  • Old Members — old registered users who made no purchases.
  • New Buyers — customers who made their first purchase.
  • Good Buyers — customers who made 3 or more purchases within the last 6 time periods.
  • Very Good Buyers — customers who made the most purchases within the last 6 time periods. As the upper threshold for this segment, OWOX BI analysts used the Transformation Rate (the percentage of customers who made a purchase within the reporting period). For example, a customer who bought something in each time period or in 4-5 of the 6 last time periods.
  • Casual Buyers — customers who made a purchase in 1-2 of the last 6 time periods.
  • Sleep — people who haven’t made a purchase within the last 6 time periods.
  • Inactive — people who haven’t made a purchase within the last 12 time periods.

Having specified the conditions for segmentation, the BUTIK.’s analysts created a schema of the possible user transitions between customer groups. This is necessary to see the user migration from one segment to another within the analyzed time period and after communicating with customers through the digital and direct marketing channels.

Schema of main user segments for CRM analysis

The schema above demonstrates the percentage of users who switch to more active segments within a reporting period. The transition to more active segments is a positive tendency and is shown with green, while the transition to passive segments is a negative tendency and is shown with red. For example, you can see that 15% of registered users (New Members) make the first purchase and become New Buyers, which is a good tendency. 86% of people, who made a purchase in the previous time period, didn’t buy anything in the analyzed time period and eventually became Casual Buyers, which is a negative tendency.

The OWOX BI analysts created user segments with the help of SQL queries. As a result, they received a table containing UserIDs, personal user data, and the segment name.

Next, the analysts used an SQL query to form another table with the main efficiency rates for each of the segments:

  • Number of users in a segment and the segment percentage in the customer base.
  • Number of orders: total and completed in a segment.
  • Average revenue per user.
  • Number of orders per user.
  • Total number of orders and the segment share in the general turnover.
  • Changes in the number of users in a segment (growth rate).

Step 3. Create reports

Using the OWOX BI BigQuery Reports add-on , the analysts imported data from Google BigQuery to Google Sheets and created 3 tables.

1. A table that demonstrates users who transited to another segment or remained in the same one.

A table to demostrate users who transited to another segment or remained in the same one

The “Clients” metric shows the number of users, the “StartSegment” column demonstrates the user segment in the previous period, and the “EndSegment” column demonstrates the user segment for the current time period. For example, in line 7 we can see how many customers switched to Good Buyers from Casual Buyers, and, again, it’s a good tendency. But we can see a totally opposite situation in line 10, which is a bad tendency. Line 5 represents customers who remained Inactive. This means that BUTIK. needs to communicate with these customers more often or more effectively and persuade them to start buying again after 6 time periods of being inactive.

2. A table that demonstrates the current data on each user within a set time period.

A table to demostrate users who transited to another segment or remained in the same one

The table displays the current list of customers who were members of each of the 9 segments. This aforementioned report also shows all personal user data for direct communication: email address, phone number, birthday, name, gender, loyalty program status, average revenue per user and the total number of user bonuses. With this data at hand, BUTIK.’s marketing specialists can set up personalized ads for each user segment. For example, you can group Casual Buyers with the 0101000 activity (2 purchases within 7 months), and send them an invitation to a secret sale.

Moreover, the information from the report helps save ad budget, allowing to exclude huge segments of users from the target audience that the company already communicates with, using direct marketing channels. Soon the data in the table will be enriched with more detailed information about each customer, allowing BUTIK.’s marketing specialists to take account of the brand, category and price of customer choices while forming ad strategy.

3. A table that displays metrics of buying activity across customer segments within the analyzed time period, compared to the previous period.

A table to demostrate users who transited to another segment or remained in the same one

This report helps BUTIK.’s marketing specialists track KPI changes for each customer segment:

  • Revenue generated by a customer segment, and its share in the total company’s turnover.
  • Comparative figures for buying activity: buying frequency and average revenue per user.
  • Order completion rate: the percentage of orders that were paid out.
  • Changes in the number of users from the active segments. A positive tendency shows user growth in active segments (Good Buyers, Very Good Buyers, New Buyers) and user reduction in the passive segments (Sleep, Inactive, Casual with the buying activity of ***000, **0000 and *00000). To get more detailed information on the segment changes, BUTIK.’s marketing specialists resort to the schema of possible user transitions from one segment to another. This schema also allows them to see how well the communication with customers was organized within the reporting period.

This wasn’t the easiest task to process Google BigQuery data and segment customers based on buying activity. We’re glad that BUTIK. is already using the results obtained from the case we worked on together. A special thanks to Yana Parshutina for her attention to detail and involvement in the project.

Alyona Samovar,
Web Analyst at OWOX BI
Alyona Samovar


  • Using Google and OWOX BI tools, BUTIK. managed to collect the complete and detailed data for creating segments, in a single cloud warehouse.
  • The OWOX BI analysts helped the company automate report creation. Now BUTIK.’s specialists can analyze important KPIs across segments and tables, with the account of data on each customer from any segment.
  • The company uses the obtained data from the reports to personalize communication with users.

We hope that you found this success story useful for your business. If you have any questions or interesting cases on using segments, kindly share them in the comment section.