Case study · Nezasa AG · February 2023

A data-infrastructure roadmap for Nezasa

Gap analysisTechnical due diligenceData infrastructureData modelling and storage

Client

#

Nezasa, founded in Zürich in 2012, makes TripBuilder, software that tour operators and travel agencies use to build and sell trips. At the time of this engagement, its teams worked across Switzerland, Portugal and Chile.

Challenge

#

Nezasa wanted to put its data to work twice: in analytics for its own decisions, and in data-driven features for its customers. Before investing in either, it needed to know whether its data and infrastructure could carry them.

The data sat in many places: the product’s own databases, website analytics, event tracking and logs, external sources and the business tools teams used every day. Three questions needed answers: what data the company actually had and how it moved between systems, whether the infrastructure would cope as data volumes and data-science work grew, and what to build first.

Our approach

#

Business first. Workshops with Nezasa’s team covered how the company ran and where it wanted to go, so the technical review started from the company’s goals rather than its systems.

Then the data. We examined what data Nezasa held, how structured and complex it was, and how it flowed through the company’s systems.

Then the infrastructure. We assessed whether it could handle growing volumes of data, analytics and data-science work.

"The collaboration is pragmatic, lean, agile. Communication is direct and to the point. Feedback loops to improve anything about the collaboration can take place anytime and always in a constructive way with direct impact. The insights and know-how Denis and his team brought to the table was diverse and perfect to get our data infrastructure to the next level."

– Patrick Hammer, CPO & Co-Founder at Nezasa

Diagram: Nezasa’s applications, web analytics, event collectors, logs, external sources and business tools feed data replication into analytical storage. Data modelling cleans, merges, aggregates and anonymises the data, and the results feed machine-learning development, exploratory analysis and business intelligence tools. Reverse ETL sends results back to the source systems.
The architecture from the review: data from Nezasa’s applications and business tools is copied into one analytical store, cleaned and modelled there, and then feeds business intelligence, exploratory analysis and machine learning.

Results

#

A target architecture and a roadmap. We identified where Nezasa’s operational processes and data infrastructure could improve, recommended a target architecture, and set out a roadmap for collecting, storing, accessing, processing and analysing data. In workshops, we worked through the findings, priorities and open questions with Nezasa’s team.

"The biggest value is the clarification and thus acceleration of important feasibility and opportunity questions. Thanks to the input that we got from SamuylovAI, we quickly had a good understanding what opportunities there are and what is needed to get started."

– Patrick Hammer, CPO & Co-Founder at Nezasa

Built in-house. Nezasa’s own team built the data platform, with data replication, modelling and business intelligence.

"The evaluation written by Denis are not only used for immediate decision taking, but will also serve as work of reference down the road thanks to the conceptual work done. The most exciting might actually be what is still to come, but we know we have a partner at our side who shares the same result-oriented, forward looking work approach as we do."

– Patrick Hammer, CPO & Co-Founder at Nezasa

What transfers

#

Map the data before you plan the analytics. Knowing what data exists, how it moves and where the infrastructure will strain shows the gaps before a build runs into them.

Agree the plan with the people who will build it. Working through priorities with Nezasa’s team gave them a plan they could carry out themselves and keep using as a reference.

Data foundations pay off again with AI. The same platform later let us build Nezasa an internal knowledge assistant in one week. Groundwork laid for analytics shortens the path to AI.

Planning analytics, data products or AI on data spread across many systems? We review what you have and set out what to build first. See Advisory & Implementation, or Start a conversation.

Download as PDF

Want AI working in your company?

Tell us which process you want to improve and where the data sits. Nothing to commit to.