About the service
I build dashboards that answer the questions a business actually asks: what happened last week, where the margin went, which warehouse is holding goods that do not move. The work starts with a conversation, not with charts. We sit down and list the decisions you make by numbers — who gets the bonus, what to order, which product line to stop — and only after that do I decide which figures deserve a tile on the screen and which are noise that distracts. Data comes from wherever it actually lives: the database of your accounting system, spreadsheets that people keep by hand, exports from a trading account. Every source is described once, in writing, so from that point each number on the screen has a single known origin.
Underneath the interface there is a precalculated showcase instead of a heavy query on every page view. Aggregates are rebuilt on a schedule, so the screen opens in a fraction of a second even when the source tables hold millions of rows, and the time of the last rebuild is printed next to the figures — you always see how fresh they are. On top of that come charts of dynamics and a comparison of periods, filters by warehouse, manager and product that work together rather than one at a time, and export of exactly the selection you are looking at, not of some fixed report written a year ago. Access is divided by roles: a manager sees his own scope, a head of department sees the department, the owner sees everything.
What I say out loud before we start, because it saves arguments later:
— a dashboard shows what is in the source, and it does not cure discrepancies in your bookkeeping; if the stock balance disagrees with the warehouse, we reconcile the data first and draw it second;
— access to the figures is closed by roles, and every export is logged — who took which selection and when;
— I do not put up a metric without a definition: the formula is written next to the number, so it cannot be read two ways;
— the refresh schedule is agreed in advance and shown on the screen, so nobody argues with a figure that is simply four hours old;
— filters, roles and export rules are written down in a short handover note, so your own people can keep the dashboard alive without me.
Underneath the interface there is a precalculated showcase instead of a heavy query on every page view. Aggregates are rebuilt on a schedule, so the screen opens in a fraction of a second even when the source tables hold millions of rows, and the time of the last rebuild is printed next to the figures — you always see how fresh they are. On top of that come charts of dynamics and a comparison of periods, filters by warehouse, manager and product that work together rather than one at a time, and export of exactly the selection you are looking at, not of some fixed report written a year ago. Access is divided by roles: a manager sees his own scope, a head of department sees the department, the owner sees everything.
What I say out loud before we start, because it saves arguments later:
— a dashboard shows what is in the source, and it does not cure discrepancies in your bookkeeping; if the stock balance disagrees with the warehouse, we reconcile the data first and draw it second;
— access to the figures is closed by roles, and every export is logged — who took which selection and when;
— I do not put up a metric without a definition: the formula is written next to the number, so it cannot be read two ways;
— the refresh schedule is agreed in advance and shown on the screen, so nobody argues with a figure that is simply four hours old;
— filters, roles and export rules are written down in a short handover note, so your own people can keep the dashboard alive without me.
