Service · Web development

Build a BI tool on your own data stack.

Your own Business Intelligence tool: embedded in your product, multi-tenant for your customers, or fully AI-driven with natural-language query. We build custom BI when Tableau, Power BI or Looker doesn't quite fit the way your data and users work.

Embedded analyticsMulti-tenantText-to-SQLReal-time streamingRow-level security

A custom BI tool is not a Tableau replacement.

Tableau, Power BI and Looker are market leaders for a good reason: for the average data analyst running ad-hoc queries on a warehouse, there is little they don't do well. The calculation and pivot layer is mature, the visualisation library is deep, and the integrations with Snowflake, BigQuery or Redshift work out of the box. For 80% of "we want to see our data" questions, a standard BI tool is simply the best choice, and building your own would waste both money and attention.

We build custom BI for the other 20%. These are typically organisations where the BI product isn't meant for an in-house analyst, but for an end user outside the organisation: SaaS companies delivering embedded analytics to their customers, multi-tenant platforms that want a separate dashboard environment for each customer, or sector-specific flows (healthcare with EHR data, finance with DORA compliance) that an off-the-shelf product has no room for. We have been working on custom software projects since 2015 and have delivered analytical dashboards for SaaS platforms, financial institutions and healthcare organisations that wanted to make their data product part of their own brand.

The question we start with is almost always the same: who looks at this dashboard, and what will they actually do based on what they see? A Tableau report for your own analyst is not an embedded analytics product for 200 customers. The UX requirements differ by orders of magnitude: end users expect an interface that fits the rest of your product, with logins, branding, performance and mobile support that enterprise BI tools don't provide as standard. Only once that foundation is right do we design the architecture underneath.

Our BI projects start small. Within a few sprints the first tier is in place: a working dashboard on your data sources, with the first few charts and filters your team will actually use. After that we build towards the full feature set in two-week sprints: additional data layers, drill-downs, alerting, exports and, if it fits your business case, an AI layer that turns natural-language questions into SQL and charts. That last one is no longer a demo feature; LLMs are now reliable enough to run text-to-SQL on top of a semantic layer in production, provided you build in the right guardrails.

Who custom BI is relevant for.

SaaS companies with embedded analytics: your customers expect a reporting layer inside your product, not a click-through to Tableau Public or a PDF export. Embedded BI in your own branding is a strong extension of your product: it increases session time, reduces support questions ("where can I see my revenue?") and becomes an argument at renewal. Looker Embedded, Sisense, Cumul.io and ThoughtSpot Embedded are useful alternatives, but they come with a licensing model that quickly becomes expensive as your customer base grows, and they remain recognisable as third-party. Above a certain scale, an analytics layer of your own within your stack is simply cheaper, and recognisable as part of your brand.

Mid-market organisations with a BI need that is too complex for Excel: your data volume or complexity is growing beyond what is still workable in spreadsheets. Tableau or Power BI licences are on the expensive side for the whole organisation, or you want a specific workflow that the generic BI tool doesn't support out of the box. A dashboard layer of your own on top of existing data sources gives your operations and management teams exactly the insights they need, without having to train the whole department on Tableau Desktop.

Industry-specific organisations — healthcare organisations with EHR data and NEN 7510 requirements, financial institutions with DORA and MiFID reporting, energy or utility companies with IoT meter data. The compliance requirements alone are often an argument against a commodity tool. We build a custom BI tool directly around your compliance framework: row-level security, audit logging, end-to-end encryption on patient or customer data, and a data flow that passes the DPIA.

Multi-tenant platforms — a marketplace, a partner platform, a franchise chain or a holding company with multiple business units. Each tenant gets its own data environment, branding and permissions, but everything runs on one shared codebase. Looker and some enterprise tools can do multi-tenant, but come with a cumbersome licensing model. A custom multi-tenant BI layer is almost always easier to maintain once the number of tenants runs into the dozens.

AI-augmented BI — organisations that want their users to "talk" to the data. "Show me Q3 revenue by region with a YoY comparison" is turned by an LLM into a SQL query on your warehouse plus a suitable visualisation. ThoughtSpot has a commercial version of this, but the real power lies in a custom build in which the LLM knows your semantic layer (column names, business definitions, joins) and stays neatly within row-level security. We build this with Claude or GPT-4 as the backend, plus a custom text-to-SQL controller that validates the query result.

Real-time streaming BI — operational dashboards that don't show "last night's data" but live status. Logistics, energy, financial trading, gaming platforms: environments where a dashboard with a 5-minute refresh is too slow. We build real-time dashboards on Kafka, Materialize or ClickHouse streams, often connected to a real-time analytics platform as the top layer.

Three flavours of custom BI.

Most BI projects fall into one of these three profiles. Which one fits depends on who uses the dashboard, which compliance requirements apply and how deep into the stack you want to go. In the first conversation we'll advise which profile is right — and be honest when a standard BI tool is simply the better choice.

Compact project · fixed sprint budget

Embedded dashboard in your product

An analytics layer inside your SaaS product that feels like part of your own UI. Login-aware, in your house style, with the visualisations and filters your customers actually use. Under the hood: a React front end with a chart library such as ECharts or Recharts, a custom query layer on top of your existing data model, and row-level security so that customer A never sees customer B's numbers. We start with the three to five dashboard views most often requested by your customer success or support team, and expand based on what is used in practice. Often a replacement for a slow PDF report or a click-through to an external tool that nobody opens.

React + EChartsRow-level securityCustom brandingExport APISSO integration
Mid-sized project · fixed sprint budget

Multi-tenant BI platform

A dashboard platform on which multiple customers, partners or business units work — each with their own data environment, branding and permission structure. Often connected to a data engineering platform that manages the ETL and the warehouse. The multi-tenant layer is usually where the complexity lies: how do you partition data at tenant level without query plans degrading, how do you scale the RBAC layer so that a single tenant can manage its own roles, and how do you keep operating costs under control as the number of customers grows from ten to a thousand. We build this on a data warehouse (Snowflake, BigQuery, ClickHouse or Postgres with partitioning), with a metadata layer that filters per tenant.

Snowflake / BigQueryMulti-tenant RBACPer-tenant brandingdbt / AirflowWhite-label
Larger project · fixed sprint budget

AI-augmented BI or real-time streaming

A BI layer with a natural-language interface: your user asks a question, an LLM such as Claude or GPT-4 translates it into SQL, runs the query and returns a suitable visualisation. Or: a real-time streaming dashboard that shows live status instead of a nightly refresh. Both use cases are among the most technical BI projects we build, and require thorough LLM integration and a streaming architecture on Kafka or Materialize respectively. For text-to-SQL, we build a controller that validates the generated query against a whitelist of tables and columns. It is not a case of "do whatever the LLM wants", but a controlled semantic layer in which the business definitions are explicitly defined. For real-time, we work with aggregations that update incrementally, so the dashboard stays fast even at high event volumes.

Text-to-SQLSemantic layerKafka / MaterializeClickHouseWebSocket stream

What you get at the end.

A working BI tool that your team manages itself, plus everything around it to keep building without vendor lock-in on a commercial platform.

  • The BI application itselfProduction and staging environments, running in your cloud (GCP/AWS/Azure) or managed by us. A complete React front end and API back end, scaled to the number of users and query volume.
  • Full codebase and architecture documentationSource code, build pipelines, architecture diagrams, ERDs and data-flow documentation from source to dashboard. Your team can continue building on it themselves.
  • Semantic layer and data modelA central data model in dbt or a comparable tool, with business definitions and metrics explicitly defined. One place for "what is revenue", reusable across all dashboards and, if you build that layer, by the LLM text-to-SQL.
  • Admin guide for your IT teamWritten for your administrators: how to create users and roles, how to onboard a tenant, how to add new dashboards, and how to read the audit log. Also includes a runbook for incidents.
  • IntegrationsConnectors to your existing data sources: warehouses (Snowflake, BigQuery, Redshift, ClickHouse), operational databases (Postgres, MySQL), CRM (HubSpot, Salesforce), ERP. Plus webhooks and a dedicated REST/GraphQL API so your team can extract data for other purposes.
  • Compliance packageRow-level security per user and tenant, audit logging on every query, encryption in transit and at rest, and GDPR DPIA documentation. For sector-specific regimes (NEN 7510 in healthcare, DORA and MiFID in finance, the AI Act for text-to-SQL) we deliver the logging and governance documents the regulator expects.
  • End-user training and documentationTwo sessions for your key users plus a short video tutorial for end users. Includes in-app help tooltips so new users don't need support straight away.
  • Management contract (optional)Monitoring, backups, security patches, performance tuning and further development. A fixed monthly price with SLA levels for response time. For multi-tenant platforms this is generally recommended, as RBAC and query performance need continuous attention as usage grows.
Not sure about a large project yet?

Test your idea first: a working prototype in 1 day

With OneDayBuild, we make your idea tangible in a single day for €1,150, so you know whether further development is worth the investment. Decide to go ahead with the full build? Then we deduct the cost in full.

View OneDayBuild →

When custom BI is the right choice.

Four patterns in which we guide organisations towards their own BI tool. If you recognise one, we'd be glad to talk further. If you don't, a standard tool is probably the better option.

Embedded SaaS

Your product needs reporting

Your customers are asking for dashboards inside your SaaS, not a Tableau licence on the side. A built-in analytics layer in your own branding becomes part of the proposition and reduces support questions like "where do I see this number?". A commercial embedded BI tool can work temporarily, but above a certain scale it becomes more expensive than building your own.

Multi-tenant

A dashboard per customer or unit

You serve multiple customers, partners or business units, each wanting their own data environment and branding. A Looker or Power BI licence per tenant becomes unsustainable; your own multi-tenant layer on a shared codebase is both cheaper and easier to adapt to customer-specific requirements. See also our page on a custom KPI dashboard for a simpler variant.

Industry requirements

Compliance is a headache

Healthcare with NEN 7510 requirements on patient data, finance with DORA and MiFID reporting, energy with regulatory logging. The compliance layer sits so close to your data that a standard BI tool becomes more of an obstacle than a solution. A custom tool is built from the start around the right audit trails, row-level security and encryption requirements.

AI & real-time

Standard BI is too limited

You want text-to-SQL ("show me Q3 revenue by region"), or a live-streaming dashboard that lags only a few seconds behind the source event. Both use cases call for an architecture that falls outside the scope of standard BI tools. We build them on a Databricks layer or a custom ClickHouse / Materialize stack.

How a BI project works.

1

Introduction & data discovery

A conversation in which we understand who will use the dashboard, which decisions it supports, and which sources need to be integrated. We also check straight away whether the business case justifies a custom build or whether a standard tool is a better fit. That's an honest question, not a sales funnel.

2

Architecture & UX workshop

A workshop with your team plus interviews with three to five end users. We draw up an architecture blueprint (warehouse layer, semantic layer, frontend, RBAC model) and an initial screen flow for the most important dashboards. By the end, you have a concrete scope and a tech stack choice.

3

Building in sprints

A working build every two weeks. We start with the data layer and one basic dashboard, then expand to the full feature set. Your team tests along each sprint and helps decide where the priorities lie, so the scope often shifts along the way based on what works in practice.

4

Compliance audit and pen test

Before going live: a pen test, a row-level security review, an audit trail check and a data-flow mapping for the DPIA. For industry-specific regimes (NEN 7510, DORA, the AI Act for text-to-SQL) we provide the extra documentation and governance documents your regulator expects.

5

Rollout & further development

A phased rollout to your user group, training sessions for key users, and ongoing management for security and performance. For multi-tenant platforms, this also includes tenant onboarding and setting up your operational runbook.

Frequently asked questions.

What clients usually want to know before we start on a BI tool.

What exactly is a BI tool, and when do you need one?
A Business Intelligence tool is software that brings together, transforms and visualises data from different sources so that an end user can see patterns, KPIs or anomalies without writing SQL themselves. Standard BI tools such as Tableau, Power BI, Looker, Qlik and Metabase do this well for most organisations that want to analyse their own data. You need a BI tool as soon as Excel no longer scales, for example because data comes from multiple systems, because different people need to see the same numbers consistently, or because there are compliance requirements on who may see what. Custom only becomes interesting when the BI tool becomes part of your own product, needs to work multi-tenant, or requires industry-specific compliance that the standard tools don't deliver.
Do you replace Tableau, Power BI or Looker?
Not for the standard use case. For a data analyst running ad-hoc queries on a data warehouse, Tableau or Power BI is almost always the right choice — we're not going to catch up with that functionality, and we don't need to. What we do is build a BI layer where Tableau or Power BI fundamentally doesn't fit. Embedded analytics in a SaaS product, multi-tenant dashboards for your customer base, AI-augmented BI with text-to-SQL, or strict compliance for healthcare or finance. Our software often runs alongside an existing Power BI or Tableau instance: the standard reporting for your own team stays there, and the specific flows move into our platform.
How does embedded analytics work in a SaaS product?
Embedded analytics means the dashboard layer is part of your own product, not a link out to an external tool. Technically, we build this as a React front end that is integrated into your existing app, with a back-end API that handles the queries and applies row-level security so that customer A never sees customer B's data. For visualisation we typically use ECharts, Recharts or Plotly — open-source libraries that we can style fully to match your branding. Commercial alternatives such as Looker Embedded, Cumul.io or Sisense also work, but beyond a few hundred active users they are often more expensive than a custom build, and they remain recognisable as a third-party iframe.
How far has text-to-SQL with LLMs actually come?
Further than most people think, provided you build in the right guardrails. An LLM such as Claude or GPT-4 can reliably convert a natural-language question ("show me Q3 revenue by region with a YoY comparison") into SQL, as long as you provide a good semantic layer: a metadata layer that explicitly defines column names, business definitions and joins. Without that layer, LLMs guess and you end up with wrong queries. With it, this becomes a production feature. We build text-to-SQL with a validator that checks the generated query against a whitelist of tables, columns and aggregations, and with row-level security that prevents the LLM from retrieving data outside the user's scope. For AI Act compliance, we also record which questions were asked and which SQL resulted — this audit trail is mandatory for high-risk AI systems.
Does multi-tenant BI also work with hundreds of customers?
Yes, as long as the architecture is designed for it. The crux lies in two layers: the data partitioning (a single shared table with tenant_id, a schema per tenant, or a database per tenant — all three have trade-offs) and a query optimiser that stays fast enough as the number of tenants grows. We choose the approach based on your data volume and isolation requirements. For hundreds to thousands of tenants, a partitioned table with tenant_id and database-level row-level security (Postgres RLS or the warehouse equivalent) is almost always the right choice. For strict data isolation (in healthcare, for example), we sometimes opt for a schema per tenant with a metadata layer that selects the right schema for each request.
What determines the cost of a BI project?
Three factors weigh the heaviest. One: the number of data sources and the complexity of the ETL — a platform that pulls data from three systems is fundamentally a different project from one that connects to fifteen. Two: the number of dashboard views and the complexity of the visualisations — a KPI overview is quicker to build than an interactive drill-down with cross-filtering and geographic heatmaps. Three: the compliance and multi-tenant layer — strict audit requirements, row-level security and tenant isolation require extra engineering. In the first conversation we discuss a realistic scope and give you a transparent price indication based on it.
Can this be integrated with our existing data stack?
Almost always, and that's usually the point. We build connectors to warehouses (Snowflake, BigQuery, Redshift, ClickHouse), operational databases (Postgres, MySQL, SQL Server), CRM systems (HubSpot, Salesforce), ERP systems and specific APIs. For ETL we use dbt, Airflow or Fivetran, depending on your existing set-up. If you don't have a data warehouse yet, we can build that too, often as part of a linked data engineering platform project, so the BI layer sits on a reliable data foundation.
How long until we can go live?
For a first working dashboard on your existing data stack, the foundation is in place after a few sprints. A full multi-tenant platform with multiple dashboards, RBAC and a compliance layer is a project of several further sprints on top of that, depending on the number of integrations and the complexity of your data model. We work in two-week sprints with a working build each sprint, so you see early on where it's heading and can steer along the way.

Talk to us about your BI tool.

A no-obligation introductory call of half an hour. We listen to who will use the dashboard, which data sources are involved and which compliance requirements apply, and we'll tell you honestly whether a custom solution is the right route or whether a standard BI tool is simply the better choice.

Edit content