SCNET · Enterprise IT · Ankara, Türkiye

Sanal Çekirdek

If two reports show different numbers, you can trust neither.

The job of business intelligence is to end the argument, not to produce charts. Sanal Çekirdek builds the layer on one metric dictionary and a governed data model.

Trust does not grow with dashboard count; it usually shrinks. Where one metric has three definitions, meetings are spent verifying the number.

The metric dictionary

Every metric gets one definition, one owner and one place where it is calculated. Until what 'active customer' means is written down, each team invents its own version and the reports diverge.

  • Definition, calculation location and owner are recorded together
  • Definition changes are versioned and announced
  • One name is never given to two calculations
  • The dictionary is reachable from the reporting interface

A governed data model

When queries connect straight to source tables, every report carries its own business rules. A shared model keeps join and filter logic in one place so report authors do not rediscover the rules.

  • Business rules live in the model, not in the report
  • Authorization is applied at the model layer
  • A model change shows which reports it affects
  • A source change is applied in one place

Designing for the user

An executive, an analyst and a field user arrive at the same dashboard with different questions. Dashboards are designed from the user's question, and each metric carries the threshold at which the number is bad.

  • Every metric carries a target or a threshold
  • Drill-down is one click away
  • Mobile use is designed rather than shrunk
  • Unused dashboards are retired on a periodic review

Freshness, performance and trust

When the data was last refreshed belongs on the dashboard. A user who does not know which moment a number belongs to will distrust even a correct one — and a slow dashboard is abandoned quickly.

  • Last refresh time is visible on the dashboard
  • Load failures are surfaced to the user
  • A load-time target is set and monitored
  • Historic figures do not change retroactively

How we work

  1. Write the metric dictionary with its owners
  2. Build the governed data model
  3. Design dashboards from the user's question
  4. Set freshness and performance targets
  5. Measure usage and simplify the portfolio

How success is measured

  • The same metric returns the same result in every report
  • Dashboard load time sits inside target
  • The proportion of actively used dashboards rises
  • Definition changes are announced with versions

Frequently asked questions

Which BI tool?

The tool follows your existing environment and user profile. What decides is not the product but whether a metric dictionary and governed model exist; without them no tool ends the argument.

Must we abandon spreadsheets entirely?

Not at all. The aim is that corporate metrics come from one source; an export path for analyst work can remain open. The problem is a hand-built table standing in for the official report.

Where does business intelligence end and analytics begin?

Business intelligence shows what happened; advanced analytics predicts what will. Both use the same data foundation but answer different questions and need different validation.

If one metric is coming out differently in different reports, the dictionary is where we should start.

Start with the metric dictionary