The data gap, ranked by how long it takes to close
| Data | What you probably have | What is needed | Effort |
|---|---|---|---|
| Customer tax registration numbers | Missing for many customers | Held for all registered customers | Months: depends on customers |
| Customer legal name | Trading name as typed | Legal name matching the trade licence | Months: depends on customers |
| Customer address | One free-text block | Structured fields | Weeks: internal reformatting |
| Item master | Grown organically, duplicates | Consistent, deduplicated | Weeks |
| Units of measure | Free text | Codes from a controlled list | Weeks |
| Tax category per line | Applied at invoice level | Per line, coded | System configuration |
| Credit note linkage | Loose or absent | Structured reference to the invoice | System configuration |
| Invoice numbering | Sometimes restarted or duplicated | Sequential and unique | Days |
Read the effort column. The two items at the top are not technically hard. They are slow, because you are waiting on other people. Everything else you control. That is why the top two determine the timeline and should start first.
The full position
This is the part of the programme that businesses consistently underestimate, and the reason is that it does not look like work.
Appointing a provider feels like a project: there are vendors, quotes, a decision, a contract. Cleaning customer records feels like administration.
But the provider decision takes weeks and can be compressed. The data work takes months and cannot, because a large part of it consists of emailing several hundred customers to ask for their tax registration number and legal entity name, then following up with the ones who do not reply, then following up again.
No amount of budget shortens that. It is governed by other people’s response times, and it is the only part of the programme with that property.
Which is why the sequencing advice is always the same: start the data work first, before selection, before integration, before anything. It is the long pole, and every week of delay at the start is a week lost at the end.
Customer master data: start here
For each customer you invoice, you will need:
- Legal name as it appears on their trade licence: not the trading name you have always used
- Tax registration number, where they are registered
- Address in structured fields: street, city, emirate, country, rather than one block of text
- Country code from the standard list
- A reliable contact for invoicing, which is worth capturing while you are asking anyway
The practical approach is a single, clear request to every customer, sent early, with a simple form and a deadline. Then chase. Businesses that fold this into an existing touchpoint (a statement run, a renewal, an account review) get materially better response rates than those sending a standalone compliance email.
Item and transaction data
The second tier is entirely within your control, which makes it schedulable rather than uncertain.
Item master. Consistent descriptions, deduplicated entries, and a unit of measure code from the controlled list for each. Businesses that have accumulated products over years typically find duplicates, near-duplicates and abandoned entries, and the cleanup is overdue for reasons unrelated to e-invoicing.
Tax categorisation at line level. Most SME systems apply a tax treatment to the whole invoice. Under the specification it is applied per line with a defined category code, which means every item needs its correct treatment attached, and that is a classification exercise, not a data-entry one. A business that has been treating exempt supplies as zero-rated will surface that here.
Invoice numbering. Sequential and unique. Businesses running separate number ranges per branch or restarting annually should confirm the result is still unique across the entity.
The transaction types that need mapping
Beyond master data, certain arrangements have no structured equivalent in most current systems and need deliberate design:
Credit notes must reference the invoice they correct. Where credit notes are currently raised as standalone documents, that linkage has to be created.
Milestone and progress billing, heaviest for construction and contracting, where the schedule usually lives in a spreadsheet beside the accounting system rather than in it.
Retention held against a contract, which has to be represented rather than tracked separately.
Advance payments and deposits, properly represented rather than netted off.
Self-billing, where the customer raises the document.
Intra-group invoicing, which has a transition running through 1 January 2029.
The test for whether you have work here is simple: if any part of how you bill lives outside the accounting system, that part needs to be brought into it.
The common misunderstanding
- Starting with provider selection rather than the data, which is the long pole.
- Expecting the provider to fix it. They do not know your customers.
- Sending a standalone compliance email instead of folding the request into an existing touchpoint.
- Assuming trading names will do. Legal names matching the trade licence are required.
- Treating line-level tax categorisation as data entry when it is a classification exercise.
- Leaving billing schedules in spreadsheets outside the accounting system.
- Underestimating it because it looks like administration rather than a project.
What to do next
- Export your customer master today and count how many records are missing a tax registration number.
- That count is your timeline. Start the collection immediately.
- Fold the request into an existing touchpoint for better response rates.
- Clean the item master and add unit codes: internal work you control.
- Check whether tax is applied per line or per invoice in your system.
- List anything billed outside the accounting system and plan to bring it in.
Related questions
Frequently Asked Questions
What invoice data does e-invoicing require?
Customer legal names matching trade licences, tax registration numbers, addresses in structured fields, a consistent item master with coded units of measure, tax categories applied at line level, and structured references from credit notes to the invoices they correct.
Why is customer data the hard part?
Because it depends on your customers responding to you. Collecting tax registration numbers and legal entity names from several hundred customers is not technically difficult and it takes months, and no amount of budget shortens it.
Can our provider clean the data for us?
No. A provider transmits and validates; it has no way to know your customers’ legal names or tax registration numbers. That work is unavoidably yours, which is why it should start before provider selection rather than after.
How do we get customers to respond?
Fold the request into an existing touchpoint (a statement run, a renewal, an account review) rather than sending a standalone compliance email. Response rates are materially better, and then you chase the remainder.
Is the trading name we always use good enough?
No. The legal name as it appears on the trade licence is required, and for many customers those differ. It is one of the two items that has to come from the customer rather than from you.
What is different about tax categories?
They move from invoice level to line level, with a defined code per line. That makes it a classification exercise rather than data entry, and a business that has been treating exempt supplies as zero-rated will discover it at this point.
What if our billing lives in spreadsheets?
That is the clearest signal you have work to do. Milestone schedules, retention tracking and advance payments held outside the accounting system have no structured equivalent to transmit, so they need to be brought in before go-live.
How do credit notes change?
They must carry a structured reference to the invoice they correct. Where credit notes are currently raised as standalone documents, that linkage has to be created, usually a system configuration matter rather than a data one.
What is the very first thing to do?
Export your customer master and count the records missing a tax registration number. That count is your timeline, and it tells you more about your readiness than any vendor conversation will.
Do we need this data for customers who are individuals?
The requirements differ for supplies to unregistered customers and individuals, where a simplified form of invoice is generally permitted with fewer mandatory fields. That is worth establishing early if you sell to consumers at volume, because it changes how much of the customer master work actually applies to you, retail and food and beverage businesses in particular should not assume the full B2B data requirement applies to every transaction.
How many records are missing a tax registration number? That number is your timeline, and it is the one measurement worth taking this week.
Check my compliance status 058 101 9570
Last reviewed 27 July 2026. Rates, thresholds and deadlines change, the e-invoicing provider deadline has already moved once. Confirm current requirements with the Federal Tax Authority before acting, or ask us to check your position.