In this guide 6 sections
A data dictionary is a written list of every field a report uses, with its source, its exact meaning, its calculation and an owner. It stops arguments about what a number includes, because anyone can read the definition instead of guessing or asking whoever built the dashboard.
What goes into one entry
Each entry describes one field or measure. It names the source system and table, states in plain words what the number counts, and records the formula when the figure is calculated.
It also names a person, because when a definition is disputed someone must decide, and that someone should be written down. Without an owner a dictionary is only a document nobody trusts, so the next dispute about a number goes straight back to guessing.
Keep entries short. Two or three lines per field is enough, so a sales manager, a finance lead and a new analyst can read the same page and agree on it.
- Field name as shown on the dashboard
- Source system, table and column
- Plain-language meaning and what it excludes
- Formula, filters and time basis
- Owner and date last reviewed
A sample entry for sales
Net sales: invoice line value after trade discount, excluding tax and freight, for invoices with status Posted, taken from the invoice line table in the ERP. Credit notes are subtracted in the month they are posted, not in the month of the original invoice, which keeps closed months from changing.
That entry settles a familiar argument. The sales team counts orders when they are booked, finance counts invoices when they are posted, and both call the result sales.
Add a second entry for booked orders and put the two side by side, so the gap between them gets a name and an owner.
Sample entries for stock and receivables
Stock on hand: quantity in the warehouse bin table at midnight, excluding items in quarantine and goods in transit, valued at moving average cost from the item master. Owner is the stores manager.
Overdue receivable: invoice balance where the due date is earlier than today, excluding disputed invoices flagged in the collections log, with ageing counted from due date, not invoice date. Owner is the credit controller.
Notice what each entry pins down. The midnight cut-off, the quarantine exclusion, the moving average cost and the due date that starts the ageing clock are exactly the details that make two reports disagree.
- Stock: cut-off time, quarantine rule, costing method
- Receivables: due date versus invoice date, disputed items
- Sales: gross or net, tax, returns, posting date
- Headcount or hours: contractors in or out
How it ends the argument about numbers
Meetings stall when two people show different totals, yet almost always both are right under their own definition of the word. A dictionary turns that debate into a lookup, so no one needs to be blamed for a number that was only defined differently.
It also protects you when people leave. The analyst who knew that region codes changed in an old migration will not be there forever, so if it is written down, the next person simply reads it.
Link each dashboard tile to its entry, because a small info icon that opens the definition costs little and saves many messages.
Start small and keep it alive
Skip the warehouse columns. Begin with the ten measures on your management dashboard and write them with the people who own them, using a shared spreadsheet or a wiki page.
Review it whenever the ERP changes, a new report is built or a number is questioned, and put a review date on each entry. Stale entries are worse than none.
Next step: pick the three numbers your team argues about most, write their entries this week, and add the info icons to the matching dashboard tiles.
Data dictionary template
A single-sheet spreadsheet template with eight columns: field name, business meaning, source system, table and column, formula, exclusions, owner and last reviewed date. Three example rows come filled in for net sales, stock on hand and overdue receivable, and a blank block below links each dashboard tile to its entry.
Open a blank worksheet to printQuestions owners ask
Is a data dictionary the same as a data catalogue?
Not quite, because a dictionary explains what each field means, while a catalogue is a wider inventory of datasets, locations and access rules. Start with the dictionary.
Who should write the definitions?
The business owner of each measure should write the definition, with an analyst recording it, so finance defines receivables, stores defines stock and sales defines bookings. An analyst working alone will describe what the query does, which is rarely what the number should mean to the people reading it.
Where should the dictionary live?
Keep it somewhere people already open, such as a shared workbook, a wiki or a page linked from the dashboard tile itself. Files in forgotten folders go stale.
What if two departments will not agree on a definition?
Keep both, give them different names, such as booked sales and invoiced sales, and show each on its own tile with its own entry in the dictionary. Agreement can wait, and confusion stops now.