From the deployment
Garment manufacturer — SE Asia — Apache Superset embedded in custom ERP (PostgreSQL + ClickHouse) — 9-year client
Reporting locked to a privileged tool. Daily 1-hour morning meeting — 2–3 managers, 6 days/week — to review KPIs and allocate work. Everyone else waited for the meeting to get answers.
Superset dashboards embedded directly in the ERP. Role-based access rolled out across production, sales, and procurement. Daily huddle reduced to a brief exception review — everyone sees their own live data in the system they are already in.
1-hour daily KPI meeting eliminated. Pending orders, pipeline, and showstoppers — visible to every team member, without a report request or a meeting.
The real problem was not the tool. It was the model.
We have worked with this garment manufacturer for over nine years. In that time, we built their custom ERP, their supplier portal, and their data warehouse. We understand how the business runs — and we have watched how their reporting needs evolved as the organisation grew.
For a period, Power BI was the BI layer. It worked, technically. The reports were accurate. But per-seat licensing meant that every new executive or department head who needed data access was a licensing conversation. In practice, this created an invisible ceiling. Rather than expanding data access as the team grew, the cost per user kept the number of users constrained.
Decision-makers were not consuming BI because it required them to log into a separate application, navigate to the right dashboard, and pull what they needed. Each of those steps is small. Collectively, they are enough friction to make most people default to asking someone else for the number — which meant a developer or analyst spending 2–3 days producing a custom report that answered one question, once.
The question we kept returning to was not "which BI tool is better." It was: how do we take reports to where the decision-makers already are, without changing their behaviour?
Marketing has a version of this principle: go where your customers are, don't expect them to come to you. We applied the same logic to internal reporting. The executives were in the ERP. That is where they spent their working day. That is where the reports needed to be.
Why Apache Superset for manufacturing BI deployment
Apache Superset is open source, enterprise-grade, and — critically for this deployment — supports embedded analytics. You can embed a role-aware Superset dashboard into any custom web application. No separate login. No navigation away. The user sees the report as part of the interface they are already using.
For a team building and maintaining a custom ERP, this was the unlock. We were not choosing between Superset and Power BI on feature grounds. We were choosing a tool that could disappear into the existing product — and do it without a per-user cost that would recreate the same ceiling we were trying to remove.
The same budget that had been licensing Power BI seats for a subset of the organisation could now fund unlimited internal access, plus faster report development, plus the infrastructure to serve the data properly.
The company, instead of saving money, could do more within the same budget. Access to data went from a cost centre to a capability.
The technical architecture: Superset, ClickHouse, and embedded ERP dashboards
The manufacturer runs two data sources that serve different purposes. The operational database — PostgreSQL — holds live transactional data from the ERP: orders, production, supplier transactions, inventory. The data warehouse — ClickHouse — holds historical and aggregated data built for analytical queries at speed.
Superset connects to both. Operational dashboards pull from PostgreSQL for live data. Trend analysis, period comparisons, and management reporting pull from ClickHouse. The query routing is handled at the dashboard level — the user sees one interface and never needs to understand the underlying data architecture.
That layer — Superset on top, PostgreSQL and ClickHouse beneath, the ERP feeding both — is what we mean by building a unified data pipeline. The dashboards are the visible part; the structure underneath is what makes them trustworthy.
Role-based access: right data to the right person
One of the constraints of the old model — beyond licensing — was that BI access was effectively binary. You either had access or you didn't. Superset's row-level security and role-based access changed this, and it changed it in a way that directly mapped to how the business was structured.
Three ways the reports now reach people
Embedding in the ERP was the primary delivery mechanism. But we had learned over nine years that BI reports cannot be static — the questions change, the business changes, and a dashboard that was right six months ago may be answering the wrong question today. The delivery model had to match that reality.
What changed
The measurable shift in reporting speed matters, but the more significant change was behavioural. Executives who were not previously engaging with BI started driving new report requirements themselves — because the friction of access had been removed, and because the data was now appearing in the context where they were already making decisions.
Data-driven decision-making moved down from senior leadership to the executive layer. Not because of a training programme or a change management initiative — because the tool was in the right place, in the right form, for the people who needed it.
BI reports are not static deliverables. They are living tools that need to evolve as the business evolves. The right architecture is one where new requirements can be turned around in hours — not days — and where the person with the question can get the answer without waiting for an intermediary. Superset, embedded and role-aware, gets closer to that than any licensed tool we have deployed for this client.
A note on Apache Superset for mid-market operations
Superset is sometimes positioned as a developer tool or a startup cost-cutting measure. Our experience is that it is neither. The role-based access, row-level security, embedded analytics, and scheduled reporting capabilities are enterprise-grade. The absence of per-seat licensing is not a compromise — it is a structural advantage for organisations where broad data access is a strategic goal, not a premium feature.
The constraint is implementation. Superset requires a team that can configure it correctly, connect it to the right data sources, and build dashboards that answer the right questions. Without that, the tool is as underutilised as any other. With it, the economics and the access model are genuinely different from anything in the licensed BI market.