Your data before your system

This is the most important station on the path, the least exciting, and the most deferred.

When an e-invoicing project slips, the stated reason is always “integration”. The real reason, in most cases, is that the data that has to appear in the document does not exist.

Integration is weeks of work. Data is months. Start with the first and defer the second and you have sequenced your project backwards.

Why this part specifically fails

Because it is not IT work, and everybody thinks it is.

A customer’s tax registration number is owned by sales. Item coding is owned by procurement or the warehouse. Units of measure are owned by the warehouse. The entity’s correct legal address is owned by finance.

A project handed to a team that owns none of that data, and has no authority to change it, ends as a monthly gap list that never moves.

So the first decision here is administrative rather than technical: every list gets a named owner, a date, and a completeness percentage reported to management monthly.

The seven lists

#	The list	What must be complete	Who owns it
1	Legal entities	Legal name, tax number, address in the required form	Finance
2	Customers	Tax number, address, type: business or consumer	Sales
3	Items	Approved coding, description, tax classification	Procurement
4	Units of measure	Standard coding for every unit in use	The warehouse
5	Taxes	Status and rate per scenario, and the basis of any exemption	Finance
6	Payment methods	A standard code for every method defined	Finance
7	Numbering	A consecutive series per document type, with no gaps	Finance

Row two is what lengthens projects, because it is the only one you cannot complete alone: a customer’s tax number comes from the customer. Writing to a thousand customers and waiting for replies is months rather than weeks.

Rows three and four look easy until you open the list and find three names for one item, and a unit of measure called “carton” with no definition of what is in it.

Row six surprises people most: many systems require every payment method you have defined to be mapped to a standard international code list. Skip it and you submit an invoice that is accounting-correct and rejected for a reason with nothing to do with accounting.

The number that measures your readiness

One measure replaces ten status reports.

The percentage of invoices issued in the last three months that would have been accepted if submitted to the authority today.

Computing it needs no integration and no provider: take a sample of two hundred real invoices, check each against the required field list, and count how many are complete.

The first figure is always a shock. Between 30% and 60% is normal on a first measurement. That is healthy: knowing it today is a thousand times better than learning it from the first batch of rejections.

Then remeasure monthly and report it to management. One moving number does what a ten-page report does not.

The customer data campaign — how it is run

This is the longest task on the project and the least under your control, so it is run as a campaign rather than as a task.

Four waves, a fortnight each.

Wave	Who	Channel	Expected
1	The top 20% of customers by value	A call from the account manager	High response
2	The remaining active customers	Email with a ready form	Moderate
3	Non-responders	A reminder, carrying the mandate date	Some move
4	What is left	Commercial escalation: no data, no invoice	Settled

Wave one is not the largest by count but by effect. Twenty percent of customers usually represent most of the invoice value, so completing their data alone moves your readiness figure visibly.

Wave four needs management cover before wave one starts. A sales rep cannot tell a customer “I will not invoice you” unless that is written, approved policy. Agree it in week one, not in the week you need it.

And do not ask a customer for what you can find yourself. Tax registration numbers are published in the public registers of most countries in the region, and an internal team filling a thousand of them is faster than a thousand emails awaiting a reply.

The order of work

Start with what you do not control the timing of, rather than with what is easiest.

1. Customers first — because it depends on somebody else replying. Start it today.
2. Items and units of measure — long internal work, needing a decision rather than effort.
3. Taxes and payment methods — bounded work, fast once decided.
4. Entities and numbering — the fastest, and usually already right.

The order is deliberate: the element whose speed you do not control starts first.

What you actually do in this station

• Build the seven-list table with a named owner and a completion date for each.
• Compute the readiness percentage on two hundred real invoices. Put the number somewhere the finance director sees it.
• Send the data update letter to your customers this week, rather than after choosing a provider.
• Inventory duplicate items and build the merge list.
• Remeasure after a month, and put the figure in the monthly management report.

The three commonest mistakes

One: handing the file to IT. It becomes a project with no authority over its own data.

Two: deferring the customer mailing. It is the longest task and the least controllable, and it always ends up last in the schedule.

Three: measuring progress in tasks rather than in readiness. Ten tasks close and the percentage of acceptable invoices has not moved.

The acceptance test for this station

1. Name the seven lists and each one’s owner in your company, by name.
2. State your current readiness percentage and how you measured it.
3. Explain why you start with customers rather than items.
4. State what you do about payment methods and why.
5. Show the duplicate item list and the merge plan.

What comes next

You know the size of the work. The next station decides who does it: choosing your route — a module inside your system, a provider, or building it yourself, and what each really costs.