Data Analysis Tools

[[email protected]]

In order to be able to collect and analyze anything at all, it is important to set up data collection and analysis at all stages, which requires the use of various tools and platforms that help process and visualize data. Below is a diagram describing the basic structure and necessary components for creating reports:



Data Sources

Data sources are the platforms and systems from which pharmacy email database we collect information.

Examples of such platforms: CRM systems (for example, AMO CRM, Salesforce, HubSpot), advertising platforms (Yandex Direct, Google Ads, Hybrid, VK Ads, etc.), web analytics tools (Yandex.Metrica, Google Analytics, Adobe Analytics), social networks (VK, Meta*, Instagram*, LinkedIn).

These sources provide us with the raw data we need to analyze the effectiveness of marketing campaigns and other business processes.

ETL/ELT (Extract, Transform, Load / Extract, Load, Transform

ETL is the process of extracting data from various sources, transforming it into the required format and loading it into a data warehouse. Examples of platforms: Airbyte, OWOX, Improvado.

ETL tools help combine data from different sources, clean it, and prepare it for analysis. Without an ETL process, data from different systems may be incompatible or incomplete.

Data Warehouse

A data warehouse is a centralized repository where collected data is stored and organized for subsequent analysis. Examples of platforms: Clickhouse, Google BigQuery, Snowflake.

A data warehouse serves as a central location for storing structured data. It enables fast and efficient querying and reporting, making it easy to analyze large amounts of data.

Data Visualization

Data visualization is the process of presenting data in a graphical or visual format (like familiar reports for marketers, for example), which simplifies their interpretation and analysis. Examples of platforms: Qlik, Tableau, Power BI, Looker Studio.

Data visualization tools help you create interactive dashboards and reports that help you quickly assess results and make informed decisions. Visualization makes complex data more accessible and understandable to all team members.

And, having selected all the platforms at all the necessary stages, literally with one wave of a magic wand you get your desired report!

Let's say we want to evaluate the effectiveness of an advertising campaign in Yandex Direct.


Data Sources: We collect data from Yandex Direct, CRM systems (for example, AMO for lead and sales data) and Yandex Metrica to analyze user behavior on the site.
ETL: We use the OWOX platform to extract data from all these sources, transform it and clean it.
Data Warehouse: We upload cleaned data into Google BigQuery, where it is centrally stored and organized for analysis.
Data Visualization: Using PowerBI, we create dashboards and reports that show key performance metrics.

Attribution in Performance Marketing
Here you are, as a true analyst (and most likely you are not the only one there) spent an eternity to collect your report and are ready to breathe a sigh of relief, but no such luck! How to interpret data is a new quest. Building a report is only the first step towards an objective assessment of marketing efforts. It is important not only to collect data, but also to take into account the possible subjectivity of the data and try to derive it in the most fair way.

This is where attribution comes in, helping to identify which channels and touchpoints contributed most to achieving the end goal. Here’s an example of a subjective reading of the data:

For example, a user first sees a banner on Yandex Display, then clicks on an ad on social networks, studies prices on a website, receives an email with a promotion, interacts with content on a blog, and finally makes a purchase after receiving a promo code by email. When using the Last Touch attribution model, the entire contribution will be attributed to the last channel - the email newsletter, which does not take into account the role of previous interactions. On the other hand, with First Touch attribution, all credit will be attributed to the first touch - the banner, which ignores subsequent contacts.

Is such a one-sided assessment of the contribution of channels fair? Certainly not in all cases. Often, when looking at the results in this way, there is a risk of turning off a channel that, in fact, has a great deal of weight in customer communications.

There are different types of attributions for different types of interpretation:

Last-Click attribution – all credit is given to the last channel before conversion.

First-Click attribution – all credit is given to the first channel through which the customer interacted with the brand.

Linear Attribution – Credit is distributed evenly across all touchpoints.

U-Shaped Attribution – Most of the credit goes to the first and last touches, with the remaining credit distributed among the touches in between.

W-Shaped Attribution – Credit is distributed between the first touch, the lead generation touch, and the last touch before purchase.

Data-Driven Attribution (Algorithmic) – Uses machine learning to dynamically evaluate the contribution of all channels.

I would really like to now go on for a couple dozen pages about how important it is to look at data from different angles, using different models, which is better to choose for different cases, but, unfortunately, this is a separate topic for an article.