Dana Tousova, our VP of Marketing, and Tereza Mlynarova, our Head of Product Development, interviewed Adela Brandejsova, Data Analyst Team Lead at Notino, about how the implementation and adoption are going and how Accurity is helping with consolidated reporting and harmonizing the terms across such complex e-commerce reporting.
I am leading a team of five data analysts in Notino. The team's primary responsibility is managing the reporting portal, which contains a significant number of reports. Some of the reports date back to the early days of Notino! The reports are done within various tools, mainly in Power BI.
The main driver for selecting the tool was maintaining the same naming and data in the background for all our reports, increasing the trust in data for the business users, and aligning the business definitions across the reports.
It was a request from both – stakeholders and analysts. Analysts wanted to increase efficiency so that they eliminate repetitive reports, increase transparency about what metrics and terms each report consists of, harmonize the terms, enable business terms definitions, and facilitate searchability across all the reports within the portal.
For users or consumers of the reports, it facilitated the usage of the reports. It lowered the need to ask the data team about the definitions, so they are more self-sufficient in consuming the reports, and it increases the trust in the data and correctness of reports in Notino. As a result, we can see that the need for users to do their own reports to be able to trust the data has decreased as they trust the reports that are shared with them. So, it enables efficiencies within the organization because of reducing the creation of duplicate reports.
It is worth mentioning that there was also a push from the management based on the internal transformation and creation of the data office department to own the whole data domain in the company.
There are power users within certain domains or teams that are supported and empowered by the data team. However, the data team does not maintain the data quality of such reports as the usage is only limited to that domain or user. Such reports also reside outside of the reporting portal in dedicated storage.
There are also analysts outside of the data team. There is, for instance, a data office department with, e.g., a data science team, which also creates some of the reports. Here the cooperation and support of our data team are closer.
Not really – it was a logical next step. It was a continuous process of cleanup, a continuous increase in the number of reports, and the start of work on the new reporting structure. This means we are to revisit the reports and edit our data sources.
We have a documentation wiki, and we train users to use it; however, it has limits as the terms are not precisely described there. This documentation also serves the data team as it includes metrics, calculations, and links to data sources. But as I said, its adoption varies across individuals. Apart from this, no dedicated tools were in place yet, like a business glossary or data catalog.
Not yet. The work is currently based around when we are looking at a particular report, we have it connected to our new data server, there is already a modified data part, and we are observing that not only the metrics and their definitions are matching, but also the data sources are matching. This means that it goes from the same source and with the same filters so that there is an exact match. While doing this, we are mapping the differences and directly correcting and editing them. As a result, what is then written in the business part can be trusted. We encourage business users to utilize these descriptions when using their own reports, like those in Google Analytics.
Accurity is not that much part of the ecosystem yet. We were focusing on cleaning the reports and defining the metrics and calculations. All the business terms are being defined in the online environment outside of the existing tool stack. We are adding tags for respective departments to the reports so that the business owners of respective terms can follow up on this and can define how the metric should be specified. This naturally brings to attention individual discrepancies in terms or metrics, enabling their further final definition and harmonization. For us, this is an extension to the Power BI report server. So, we have databases where we calculate the data and where our data source we use for reports. Based on this, we are trying to document all as best as possible. However, we are not yet at the stage where we would map reports to business terms or reports to data sources.
One of my colleagues is using it, and we liked it, but it gets overwhelming at times with having even ten children, so we will leave this for a later stage. But in general, we decided that we would define the basics first – the basic calculations so that we would not get overwhelmed by too many details and have the fundamentals described well and be able to build on them. So now, it is a lot of cleanup work that is being done ex-post in older reports, but in the future, every new report will be built faster.
Tereza: Yes, we can recommend that in the future: define the report, define metrics, and then build the report. This will then go much faster, but naturally, such cleanup is needed first.
To illustrate the scope, there are approximately twenty reports for each team, so in total 300 reports that are really used, so it is a lot of work to do with the definitions and cleanup. Hence, the team prioritized the most used reports, removing not used reports on the go. For the most crucial metrics, there are data quality tests defined that we want to show in the reports indicating that the data behind the reports are correct and the data were loaded correctly.
There are currently a lot of requirements for support from business users and other ad hoc necessities, so the capacity to work on the project is only sometimes available. Thus, it is more of a long-term initiative where we prioritize the key reports, then prioritize based on domains or departments, and then also clusters of data that are repetitive. An example is e-commerce reports with data from Google Analytics, mainly sessions repeated in several reports from different views.
We have around 150 terms defined, which is roughly 25% of the scope. We significantly reduced the scope of the terms that needed to be defined by focusing on the most important definitions. For example, we will not define “sales last year” but only what “last year” means. At the same time, we decided to keep the documentation on the report level for some of the very specific metrics of individual reports, as we would like to keep the business terms definition on a maintainable level and decide later if to move more terms from these very specific reports to general definitions. Such an iterative approach helps us to move forward faster.
It is the data team and business owners of the reports – business managers/stakeholders/domain data owners whom we discuss with. We also presented the tool and the initiative at the internal demo of our web and innovations department. Concerning the Browser Dictionary plugin that enables the highlighting of the business terms in the browser, this was presented only to a narrow group of users as it will need more of an internal adoption, so they would like to phase this one in.
The general feedback is positive; everybody thinks such an initiative is needed and valuable. There were even questions from our colleagues who could not wait to start using the Browser Dictionary plugin.
Yes, we plan to do internal demos monthly. They relate to data quality tests that we are doing, and we give ten-minute updates about our progress and status during the internal demos. In addition, the management of the web and innovation department is presenting the status to top management to give them an update on the initiative.
We estimate to move to the next phase and map the terms to data sources in Q2 of 2023. We have sort of a middle step in place, as all reports have their own table in the database, so we already have some progress in this. So, it simplifies the work for developers, and we can focus on the business terms' definitions.
Tereza: Once your core terms are defined and you connect them to sources, this will further facilitate the consolidation.
We are still considering the pros and cons if to use this plugin or if to go via the documentation of respective reports. I can see the biggest advantage of the Browser Dictionary plugin in the fact that it is not needed to keep the mapping of the report and term, and the plugin can be run on any web page. However, since we are focusing now also on data quality tests, we need to have data catalog in place.
With the latter option, going through the database and showing the description directly in the report documentation might be easier for users to adopt. With the latter option, it also plays into the fact that users are already used to seeing the descriptions in the report documentation and a need to refresh the reports often. The refresh is done for some crucial reports every five minutes or so or when resetting the report's filters.
Tereza: We have recently focused on integration with reporting tools such as Power BI, and we plan some improvements in these matters so that Accurity can do such mappings.
Not really, the usability is fine and straightforward, which simplifies the adoption. The benefits gained from harmonizing the terms are notable and appreciated already, even in such an early adoption stage.
We had a lot of use cases and ideas on how to use it at the very beginning, which we needed to narrow down to enable good progress and adoption. For instance, we originally wanted to do search in terms, having one place where the terms will be located and using it as a wiki. But then we realized we needed to start with the most important things that are being explained often and based on this, defined the crucial parts that needed to be done first.
We appreciate your consultation services during the PoC stage. Now we are confident we can manage the implementation ourselves; since we started with the crucial terms, the description of the terms is straightforward. And it is clear there are terms we need to be very specific with as we are getting different values from different systems for metrics like “items in stock”. So, we need to be very precise in such definitions so that the differences in various applications are minimized, and we also need to consider latency when data is loaded.
The important part is to maintain the business terms and have this as an ongoing, long-term initiative that has clear productivity benefits.
Implementation in Notino is, despite its early stages, on a good track to bring the benefits stemming from the harmonizing of the terms and how this is crucial in an e-commerce company with a robust number of reports to have an intercompany alignment on metrics and calculations. Besides this, it is bringing efficiencies while removing duplicates and minimizing the need for support of business users related to reports, empowering business users and increasing trust in data for business decision-making.
We described the use case for consolidated reporting and what steps are ahead for companies like Notino in our recent 4 Steps to Enrich Your Reports with Trustable Data That Everyone Can Rely On article.
If you want to get started with our Accurity Business Glossary and Data Catalog solution you can sign up for our SaaS version for free – start evaluating the features and the value, they can bring to you and your business.