Business Intelligence

Organizing the data before building the dashboards

Sales reporting moved from manual Excel cross-checks across two systems to a single governed source in BigQuery.

Luda is a Healthtech company that connects pharmacies with the digital customer. Through its in-house integrations, it provides a value-added service that helps pharmacies in Spain’s National Health System sell online, closing a gap they often lack the services and resources to bridge on their own.

Industry

Software & Technology

Location

Madrid, Spain

Stack

Google Cloud, GKE, Terraform, Airbyte, Airflow, BigQuery, Looker

Overview

Data from MongoDB and MariaDB unified in a single data lake in BigQuery.

Infrastructure deployed on Google Cloud and defined with Terraform.

Looker dashboards covering sell-in and sell-out, margin, stock and a map of the pharmacy network.


Background

About Luda

Luda is a service company that works with pharmacies. Pharmacies in Spain’s National Health System face a gap when it comes to online sales: reaching the digital customer takes services and resources they often don’t have. Luda closes that gap, acting as the link between the digital customer and the pharmacy through the integrations it builds in-house.

To run that network, Luda’s partners, sales and key account teams need a clear view of what is sold, where, and through which pharmacies.


The challenge

Sales data spread across systems

Two environments and a lot of Excel

Luda’s operational data sat in MongoDB and MariaDB databases. Aggregated sales information lived in Luda’s native CRM, while pharmacy data was managed in Salesforce. Any sales report by product or by pharmacy meant exporting from both environments and crossing the data in Excel. For every ad hoc request, recalls Daniel Conde, Business Development Manager at Luda, that meant “at least a couple of hours of data handling.”

Every new request went to engineering

When a request went beyond what internal tools already covered, it landed on the development team. “It wasn’t agile at all,” says David Freiria, Software Engineer at Luda. Engineers spent time on reports for key account managers and sales, such as pharmacy performance or customer classification, instead of on larger projects. The historical database was large, but it was not structured or modeled for analysis.

Dashboards without governance

With no shared foundation, each person built their own dashboard with the data they needed. Many of those dashboards were eventually abandoned because nobody kept them up to date.


“Everyone built their own dashboard independently. There was no data governance."

Daniel Conde, Business Development Manager, Luda Partners

“Everyone built their own dashboard independently. There was no data governance."

Daniel Conde, Business Development Manager, Luda Partners


The decision

Why Astrafy and Google Cloud

Luda chose Astrafy, with Google Cloud as the platform, based on recommendations and on the relationships Astrafy has with its existing clients. Astrafy’s track record with large clients gave the team “a lot of confidence,” Daniel says. The project was supported by Google funding.


The approach

Data first, dashboards second

Luda deliberately reversed the usual order. Instead of starting with dashboards, the team first organized and normalized the data, and only then built visualizations on top of it.

On the technical side, the main concern was not risk but design. Luda needed a structure that could adapt and scale as new data sources were added.

A single data lake in BigQuery

Astrafy deployed the infrastructure on Google Cloud, with GKE, Airbyte and Airflow handling ingestion and orchestration. Data from MongoDB and MariaDB is ingested, cleaned and transformed, unifying the company’s sources in BigQuery.

Clean datasets by design

Infrastructure is defined with Terraform. Data that doesn’t match a dataset’s definition is cleaned or discarded, which keeps the structure “very clean,” as David puts it, as new data keeps arriving.

Sales dashboards in Looker

Once the data was in BigQuery, Astrafy built the Looker dashboards. They cover products sold, stock, margin, profit, sell-in and sell-out, sales per pharmacy, and a map of the pharmacies Luda sells to.


“We now have a very scalable data lake that lets us keep adding data constantly."

David Freiria, Software Engineer, Luda Partners

“We now have a very scalable data lake that lets us keep adding data constantly."

David Freiria, Software Engineer, Luda Partners


Data in daily decisions

A weekly view of the pharmacy network

Reading trends by product and region

The partners team now opens the dashboards every week to follow trends and compare performance week over week. With sales aggregated and modeled in one place, the team can see how products and regions behave across the network, down to the province level. Because Madrid has a large presence in the network, it also works as a reference for how the rest of Spain is likely to behave.


“We have a tracking tool we didn’t have at hand before."

Daniel Conde, Business Development Manager, Luda Partners

“We have a tracking tool we didn’t have at hand before."

Daniel Conde, Business Development Manager, Luda Partners

Prioritizing zones and pharmacies

This changed how Luda prioritizes. The team can see which products are trending in one area, anticipate which will trend in other areas in the following weeks, and decide which pharmacies to target.

From consumers to builders

The partners team also learned to build dashboards during the project and now continues that work independently.


The collaboration

How we worked: close, responsive and flexible

Weekly follow-ups and instant answers

The two teams held weekly follow-up meetings and used a shared chat for day-to-day questions. Whenever Luda had a doubt, David says, they “got an answer instantly.”


“I felt the relationship was very close and at the same time professional."

David Freiria, Software Engineer, Luda Partners

“I felt the relationship was very close and at the same time professional."

David Freiria, Software Engineer, Luda Partners

Scope that adapted to the project

Scope was adjusted during the project as needs evolved, something that “with other teams can be more cumbersome,” according to David.


Results

One source of truth and time back for the team


  • One source of truth: All data is registered and normalized in one place, so every team works from the same numbers.

  • Faster access to data: Business requests no longer depend on manual exports and Excel. Data reaches the business faster and with more flexibility.

  • Time back for engineering: Engineers who used to build ad hoc reports now spend that time on larger projects.

  • Faster decisions: Teams spend less time building and maintaining dashboards, and more time acting on trends.

With the data foundations in place, Luda plans to bring AI into every department, using tools such as Gemini to help teams work faster and generate more business.


“Having the information already built makes decision-making easier."

Daniel Conde, Business Development Manager, Luda Partners

“Having the information already built makes decision-making easier."

Daniel Conde, Business Development Manager, Luda Partners


Book a call

Building dashboards on data you don't trust yet?

We help teams get the data foundations right on Google Cloud first, so the dashboards that come after get used. Book a call to talk through where your data stands today. Book a call.

Latest case studies

Turn your scattered data & AI initiatives into a coherent strategy

Turn your scattered data & AI initiatives into a coherent strategy

Turn your scattered data & AI initiatives into a coherent strategy