Somebody in Your Company Cannot Legally Work Tomorrow

In most UAE companies, visa and document expiry tracking is a spreadsheet. It has a column of conditional formatting that turns cells amber and then red, and one person watches it. That arrangement works for years and then fails on a specific Tuesday, because the person was on leave, or because the renewal was started and stalled somewhere, or because a document that had never expired before quietly did.

The failure is not that nobody knew the date. It is that a date is not a process. A document expiring is the end of a chain that has to start weeks earlier, involves a typing centre or a portal or a medical or a broker, has a cost, and can be blocked by something entirely unrelated — an unpaid fine, a missing attestation, a company licence that itself needs renewing first. A register that holds dates but not the state of each renewal tells you what is coming and nothing about whether anybody is doing it. What is needed is that every renewable document is a record with an owner, a lead time, a state and a cost, and that the alert fires against the state rather than the date.

That is the answer. The rest of this article is the documents people forget, why lead times are the real variable, and what an escalation has to look like before it works.

The register is always incomplete, and predictably so

Employee visas and labour cards are always in the spreadsheet. The ones that go missing are consistent enough to list:

• Passports. Several jurisdictions require a minimum validity remaining, so a passport is effectively expired well before its printed date.
• Emirates ID, which follows its own cycle and is what most other processes depend on.
• Medical insurance, mandatory, and frequently renewed as a group policy whose individual members drift out of step with it.
• Professional and trade licences — the engineer's registration, the driver's permit, the technician's certification, the safety card a site requires. These are held by individuals, needed by the business, and tracked by neither.
• The company's own documents: trade licence, establishment card, lease. These are upstream of everything else, so one of them lapsing blocks every employee renewal behind it, which is the failure that turns a small problem into a large one overnight.
• Vehicles, equipment and their permits, which are the same problem in a different department and usually a different spreadsheet.
• Dependants, where the company sponsors them, because a lapse there is still your employee's crisis and will still be your problem on Sunday morning.

The pattern is that the register covers what HR administers and misses what other departments rely on. That is a data-ownership problem in its ordinary form — one domain, several consumers, no agreed owner — and the general treatment is naming the function that carries the consequence.

Lead time is the variable, not the date

Everybody tracks the expiry. Almost nobody tracks how long a renewal actually takes, which is the only figure that determines when to start.

The lead time is not one number. It varies by document type, by nationality, by emirate, by free zone versus mainland, by whether a medical or an attestation is in the chain, and by the season. A company that uses a single reminder interval for everything will be too early on some and too late on others, and the ones it is late on will be the complicated cases — which are also the expensive ones.

The practical fix is unglamorous: record the actual elapsed time on every renewal you complete. Start date, completion date, document type, category. After a year you have your own lead times, specific to your workforce and your processes, and the reminders can be set from evidence rather than from a guess. This is a measurement exercise that costs a field and pays permanently, and it is the same discipline that makes any automation defensible, as set out in measuring what a process actually takes before and after.

Two refinements worth adding once the basic data exists. Blockers as their own state: a renewal waiting on a fine, an attestation, a signature or an upstream company document is not a renewal in progress, and merging the two hides every case that is actually stuck. And a queue view by owner, because the constraint is almost always one person's capacity in a particular week rather than the total volume.

An alert that nobody can absorb is not a control

The reason expiry alerts fail is rarely that they were not sent. It is that they were sent to somebody who could not act, or sent so often that they became background noise, or sent to a person who left.

Four properties separate an alert that works from one that is merely delivered:

It goes to a role, not a person. A named individual leaves, changes department or takes leave. A role always has somebody in it. This one change prevents the most common total failure.

It escalates on a schedule, upwards. If nothing has happened by a defined point, the next alert goes to the manager, and after that to the level that can actually clear a blocker — because most stuck renewals are stuck on a decision or a payment rather than on effort.

It carries the action, not just the fact. "Expires in thirty days" is information. "Expires in thirty days, renewal not started, owner is X, last renewal of this type took twenty-two days" is a prompt.

It is answerable in seconds from a phone. Otherwise it will be handled on WhatsApp, which is where the record then lives — the pathology described in what happens to a control that moves into a chat thread.

There is a fifth property that is really a design choice: at some point an alert should become a block. A system that will let you assign somebody to a site on Monday when their site pass expired on Friday is a system that has chosen to warn rather than to prevent. Which of your documents should block what is a decision worth making deliberately, in a room, once.

The cost nobody adds up

Renewals have a direct cost — fees, medicals, typing, insurance premiums, broker charges — and it is usually spread across several accounts with no view of the total. Three consequences follow.

The annual cost per employee is unknown, which means the loaded cost of an employee is understated, which means every job cost and every charge-out rate built on it is wrong in the same direction. That is the subject of what an employee actually costs once employment costs are included, and visa and document costs are the component most often left out entirely.

Renewal spend is not forecastable, so it arrives as a series of unbudgeted small amounts and is treated as noise.

And nobody can compare doing it in-house against using a service provider, because the in-house cost — mostly somebody's time — has never been measured.

Attaching a cost and an elapsed time to each renewal record fixes all three at once, and it is a field rather than a project.

What to do in the first fortnight

1. Complete the register. Not the dates — the list of document types, including the ones other departments depend on and the company's own. Ask each department what it would be unable to do if a document lapsed; the answers will not all be in HR's sheet.
2. Make each document a record with an owner and a state, not a row with a date. Even in a spreadsheet, this is the change that matters.
3. Start recording lead times and costs today, so that in a year the reminders are evidence-based.
4. Decide which documents block what, and implement two or three of those blocks rather than all of them.
5. Route alerts to roles with an escalation path.

The register is one of the four pieces of employee data that everything else depends on, alongside service dates, the wage structure and time capture — the four that the payroll article argues should be brought into the first phase of a project rather than the second. None of it is difficult. It is simply nobody's project, which is why it is still a spreadsheet, and why the first sign of failure is a person standing at a gate.

If the honest answer is that you do not know how complete your register is, that is a short diagnostic rather than a programme — and the work it leads to is the ordinary kind: delete the alerts nobody can act on, then automate what is left, in that order rather than the other one.