Data engineering and analytics platforms that actually get used. BigQuery, pipelines, ETL and BI – built so the right people get the right numbers, on time and trustworthy.
Most organizations I meet have plenty of data. The problem is that it is scattered across ten different systems – CRM, ERP, spreadsheets, export files, SaaS reports – and nobody has the full picture. This leads to a familiar pattern: two departments calculate the same metric and get different answers, the discussion is about the numbers instead of the decisions, and trust in the data erodes over time.
A data platform doesn't solve everything overnight. But it establishes a single source of truth – a data warehouse where all sources land in a shared model – and a system for ingesting, transforming and presenting data in a trustworthy way. The key is to start with the decisions that need to be made, not with the technology. What questions does the business need answers to every week, every month, every quarter? The platform is built backwards from those questions.
I often find that the biggest challenge is not technical but organizational. Getting different departments to agree on shared definitions – what is an 'active customer', really? – is at least as important as making pipelines work. My approach is to build in iterations: one pipeline at a time, one dashboard at a time, with continuous feedback from the business.
There is a growing trend where everyone talks about data lakehouse as a universal solution. In practice, the choice between warehouse, lakehouse or lake depends on what type of data you handle and what you want to do with it. A data warehouse (like BigQuery) is excellent for structured data and fast aggregations – perfect for dashboards and reports. A data lake is right for raw data, machine learning, and situations where you don't know in advance what questions you want to ask.
A lakehouse tries to combine the best of both worlds – and it works, but with a complexity that many underestimate. My recommendation is to start with a clear warehouse model for your key data (sales, customers, finance) and complement with a lake for raw data from logs, sensors or external sources. This gives you a simple starting point that can be extended as needs grow.
The weakest point of a data platform is often the pipelines. They are built quickly, rarely documented, and once they work they are forgotten – until one day they silently break and a report shows incorrect numbers for two weeks before anyone notices. I build pipelines with three principles: they must be monitored (alert when data doesn't arrive), they must be reusable (same transformation logic should not be duplicated), and they must be tested (quality checks along the way).
For most modern platforms I recommend ELT architecture (Extract, Load, Transform) rather than traditional ETL. The difference is that raw data is loaded first and transformed afterward – giving flexibility to redefine transformations without re-importing data. The tool choice (dbt, Dataform or custom code) depends on team competence and platform ecosystem.
The organizations we work with typically face at least one of these.
The numbers exist – but they are in CRM, ERP, spreadsheets and export files. Nobody has the full picture.
Two departments calculate the same metric and get different answers. The discussion is about the numbers, not the decisions.
Someone spends days stitching together Excel files. Insights always arrive too late.
A project was kickstarted but got stuck. Now there's half-finished infrastructure that nobody uses or maintains.
A selective portfolio where senior expertise makes the biggest difference.
Map your data sources, needs and bottlenecks. A realistic plan for a platform that actually gets used.
Robust, monitored pipelines from source systems to data warehouse – scheduled and alert-enabled.
BigQuery or equivalent as your single source of truth, modeled for extensibility.
Dashboards and reports that answer your business's actual questions – not just pretty charts.
Data ingestion from CRM, ERP, APIs and external sources – including scraping where needed.
Tests, documentation and access controls so data is trustworthy and shareable.
A clear process from first conversation to delivered result.
Free initial conversation about what decisions the data should support and where it lives today.
Audit of sources and needs. You get an architecture and prioritized plan.
Pipelines, warehouse and dashboards built iteratively with weekly demos.
Documentation, code and knowledge transfer. Your team owns the platform.
Transparent pricing with no hidden costs. All prices exclude VAT.
Map sources and needs with architecture recommendation and prioritized plan.
Defined-scope project – pipelines, warehouse and dashboards with clear milestones.
Ongoing platform work, pipelines and reporting with priority access.
Läs mer om ämnet i våra artiklar.
Answers to the questions I hear most often.
No. BigQuery is often a strong choice due to low operational overhead and good price/performance, but the platform is chosen based on your requirements, existing cloud environment and team competence.
Yes. I start with an audit of what exists and what is worth building on.
Often yes. I have built large-scale data collection including scraping and integration with limited-interface systems.
Yes. A platform without usable dashboards provides no value – BI and reporting are a natural part of the delivery.
With pipeline tests, clear modeling, documentation and monitoring that alerts when something deviates – so errors are caught before they reach a report.
Next step
Do you have an ambitious idea or a technical decision where it pays to think right from the start? Get in touch – no obligation.
Get in touch