Data & Analytics · Cross-border services
How Landthor built DataQuick — a single source of data that gives leadership reliable figures the same day.
The opportunity
The client helps thousands of people settle in a new country. As the business grew, its data did not keep up: customer, sales, finance and payment records lived in separate systems — a CRM, an accounting ERP, an operations database, a payment gateway and numerous spreadsheets.
Every leadership question — Which channels actually bring paying clients? How many real leads did we get last month? Who closed what? — required days of manual reconciliation, and the answers rarely agreed.
Decisions were being made on instinct and outdated exports. Several of the reference figures also turned out to be incorrect: duplicate records and misattributed spend that no single system could detect.
Our approach
Landthor designed and built DataQuick on a layered architecture: data is captured once and progressively refined until it is ready to support a decision, with every metric traceable back to the event it came from.
Webhooks and serverless functions stream CRM and payment events in real time; scheduled jobs complete the history and act as a safety net.
Bronze stores the raw data, Silver cleans it and applies the rules, Gold serves the metrics. Any number can be traced back to its source.
Role-based access, personal data kept out of the reporting layer, and alerts when a feed stops updating.
Self-service dashboards across four areas, fast and consistent because they read from the same model.
Why it worked
The goal was not a better-looking dashboard, but figures leadership could decide on with confidence, knowing where each one came from.— Landthor, on the DataQuick engagement
The impact
The platform did more than consolidate reporting. By making the data traceable, it corrected decisions the business had been making on the wrong figures and removed the outages and blind spots that had eroded trust.
Channel attribution was rebuilt from session and form signals, reducing the "unknown" channel from a major blind spot to a residual. The ranking changed: the channel that appeared to lead on revenue dropped several positions, and an organic channel that had been almost invisible turned out to be first.
A historical spike in leads turned out to be a CRM migration counted as new demand. Excluding it corrected the lead total and the qualification rate: the kind of distortion only a single model can detect.
Revenue, commissions, revenue per rep and attendance rate all read from the same Gold layer, so every team works from the same figures.
The platform's heaviest query had previously taken production offline. Materialising and scheduling it removed the outages and reduced the compute cost substantially, under real load with real users.
Sales-call outcomes that had stopped being recorded were recovered across the full history, and the CRM was reconfigured to send every update as it happens.
New visibility
Before: marketing spend never tied to results
Before: revenue with no reliable origin
Before: per-rep performance was estimated
Before: no data on sales-call attendance
Before: every report gave a different figure
Before: failures surfaced weeks later
What made the difference
Where a channel could not be determined, we said so. What remains as "unknown" is genuine, never an estimate presented as data.
An incorrect number almost always comes from a process, not a formula. We reported those gaps back to the business so they could be fixed at source.
Role-based access, data health alerts and materialised views keep the platform responsive with the whole team connected at once.
The best model is worthless if no one opens it. DataQuick lives in the dashboards each team already uses.
Looking ahead
With the platform live, the next steps are real-time attendance rate, more detailed commission logic and full marketing ROAS once ad spend is connected. Each one is now a query rather than a project.
Request a diagnostic