# What Building Quantsole Taught Us About Production Analytics
Lessons from building Quantsole, Rodan’s BI and AutoML platform, and how they now shape our production analytics engineering.
Published: 2026-08-20
Author: Rodan
Quantsole began as a straightforward proposition: business intelligence should make decisions easier, not create another layer of work.

We wanted to bring data ingestion, reporting, visualisation and machine-assisted analysis into one environment. The goal was not to produce more dashboards. It was to shorten the distance between a business question and a decision someone could act on.

Building the platform taught us that the difficult part of analytics is rarely drawing the chart. It is establishing what the numbers mean, whether they can be trusted and how they should change what happens next.

## What we built

Quantsole combined several capabilities that are often spread across separate tools:

- connections to operational, commercial and customer data
- automated transformation and reporting workflows
- interactive dashboards for recurring performance questions
- natural-language exploration for users who did not work in SQL
- machine-assisted pattern analysis and AutoML experimentation where the data supported it
- analyst review for validation, interpretation and deeper investigation
- deployment patterns designed around different data and access requirements

The interface was deliberately simple. Behind it sat the harder work: reconciling inconsistent definitions, handling changing source systems and making automated outputs explainable enough for people to trust.

## Lesson one: a dashboard is not an outcome

It is easy to mistake visibility for progress. A dashboard can show that margin declined, a customer segment changed or an operational backlog increased. It cannot decide who owns the response or what action should follow.

The most useful analytics systems are designed around a decision. They make clear what changed, why it matters, who needs to respond and how the effect will be measured. Without that operating context, even an excellent dashboard becomes another screen to check before returning to the spreadsheet that actually runs the process.

## Lesson two: automation needs judgement around it

AutoML can accelerate repeatable modelling and classification work. It does not remove the need to understand the commercial question, test whether the training data represents reality or decide whether an output is safe to use.

Quantsole reinforced the value of a human-in-the-loop model. Automation handles repeatable analysis and surfaces signals. Analysts and domain experts validate the result, investigate exceptions and connect the evidence to the wider business context.

That combination is usually more valuable than either extreme: a fully manual reporting process that cannot scale, or an automated system that produces confident answers without sufficient oversight.

## Lesson three: deployment is part of the product

Analytics cannot be separated from the environment in which it operates. Security requirements, access controls, data residency, source-system reliability and internal ownership all shape what can be built.

A technically impressive model is of limited use if it requires sensitive data to move somewhere the organisation cannot approve. A reporting layer will not remain trusted if ownership of definitions and transformations is unclear. Deployment architecture is therefore not an implementation detail added after the analysis. It is part of the analytical product itself.

## Lesson four: reusable engineering matters more than a universal dashboard

Different organisations ask different questions, but the engineering patterns repeat: ingest evidence, establish definitions, validate quality, expose the right views, automate recurring analysis and retain a clear route for review.

Those patterns have informed work ranging from multi-market audience intelligence to customer and occasion analysis and stakeholder survey reporting. The front-end experience changes, but the need for reliable data, repeatable logic and decision-ready outputs remains consistent.

The lasting value of Quantsole was therefore not a promise that one dashboard could solve every analytical problem. It was the reusable engineering approach underneath it.

## Where Quantsole sits today

Quantsole is not currently marketed as a standalone product. Its components, architecture and lessons now inform how Rodan designs analytics and intelligence systems around specific operational and commercial decisions.

That is a more useful application of the work. Instead of asking organisations to adapt their problems to a fixed BI product, we can apply the relevant patterns within the systems, governance requirements and working practices they already have.

The principle remains the same as when we started: analytics should reduce the effort required to reach a sound decision. The difference is that we now treat the dashboard as one part of a wider production system, rather than the final destination.

Explore our work in [Analytics and Intelligence](/what-we-build/analytics-intelligence) or learn more about [Data and Analytics Engineering](/data-analytics-engineering).
HTML: https://rodan.io/insights/what-building-quantsole-taught-us-about-production-analytics