Sign in

Blog · How-to guides

The exports by industry: the four files each desk drops, and what is in them

Every measure on this site is computed from a handful of exports a company already has. This hub gives, for twelve industries, the three or four files, which system each comes from, the columns that matter in each, the identifier that joins them, and the control total to note before exporting, so that a team knows exactly what to drop before it drops anything.

The short answerThree or four files per desk, from systems the company already runs: a ledger of record, an activity or event log, a master that describes the accounts, and, where the desk has one, a register of what was promised. Each file is a table of identifiers, dates and amounts with a control total from the source. The identifier that joins them is the account, in whatever form the desk uses: customer code, facility, intermediary, operator, policy, client. Names never leave.

The measures on this site sound like they need a data project. They need three or four files a company already has, exported on the same period cut with a control total noted. This hub gives the files per desk, the system each comes from, the columns that matter, and the join.

The pattern, once

File What it is Always needed for
Ledger of record What happened: invoices, deliveries, trades, fees, premium Every measure
Activity or event log What the company did: calls, visits, inquiries, tickets Coverage, recency, surge
Master What the accounts are: segment, size, owner, hierarchy Norms, roll-up, tiers
Register What was promised: contracts, agreements, controls, commitments Compliance, renewals, production

Twelve desks

Industry Ledger Activity Master Register Join key Control total
General B2B sales Invoice lines from the ERP CRM activities CRM accounts with owner; customer master with segment Targets Account ID Month revenue
Foodservice distribution Delivery ledger Route handheld log Account master with menu type, covers Order guides Account ID Month delivered value
Pharma Prescription data by account Call log Account master with tier Access file, dated Account ID Scripts per period
Commercial banking Balances and product holdings CRM activities Customer master with segment, group Facility register Customer ID Month-end balances
Insurance broking Placement ledger Activity log by policy Client master with sector, size Renewal outcomes; claims file Policy ID Premium in force
Healthcare supplies Order lines by ship-to CRM activities Ship-to and facility masters, dated GPO and system contracts; price files Facility ID Month revenue
Freight brokerage Load ledger with rates and costs Tender log Shipper and carrier masters; lane master Contracted lanes Load ID, lane Loads and gross margin
Asset management Transfer agent transactions and positions Wholesaler contact log Intermediary master with channel Platform agreements Intermediary ID, strategy Month-end assets
Professional services Time entries and bills Pursuit-coded time; CRM Client master with sector, size; roster Proposal log; engagement letters Client ID Billed fees
Compliance Testing log; issue register Role matrix; roster Obligation and control registers Obligation ID, control ID Register counts
Sports Sales ledger by asset Delivery log; access scans Inventory master with rate card Contracts with package discount Partner ID, fixture Contract value
Supply chain PO lines and receipts Item master with planned lead time; supplier master Contracts PO line, supplier, site Receipts per period

The columns that matter

In every file: the identifier, the date, the amount or count, and the one or two fields that place the row in the hierarchy. Everything else is dropped at export or in the data map. Names, contacts, notes and addresses never leave.

The control total

Before exporting, note the figure the source shows for the same period: revenue on the finance screen, premium in force, month-end assets, register counts. The file's total should match it. It is the one check that tests the file against the world.

Where it goes wrong, everywhere

Exported from a report, not a data view. Subtotals, merged headers, page breaks.

Different period cuts. The ledger on the month; the activity log on the last 30 days.

Names in the file. Exported because they were there.

No control total. A filtered export of half the period, analysed as whole.

Four files, every desk

Covirage maps each desk's files once and computes from them on every upload, with the validation report first. The export preparation guide covers the eight rules that make an export usable, and the files not connectors guide covers why the export is the integration.

Questions people ask

Do the exports need to be clean?

They need to be raw: one row per fact, identifiers not names, one date format, no subtotals. Raw exports from a data view are usually fine; exports from a formatted report are not. The validation report says which on the first upload.

What if a desk lacks one of the files?

Most measures need two. Concentration and dormancy need only the ledger. Coverage needs the activity log. Share of wallet needs the master. The hub marks which file each desk's core measures depend on, so a missing file removes a measure rather than the report.

How big are the files?

A year of a mid-sized company's ledger is tens of thousands of rows and a few megabytes. Activity logs are similar. Masters are small. None of it needs a connector; a monthly export on the same period cut is the integration.