AWRA OpsHub Search

For Fiji

Fiji's VAT is 15%. It was 9% until the first of August 2024. Both are correct.

A rate is not a fact about a country, it is a fact about a country on a date — and almost every finance system, including ours, stores it as though the first were true. Raise a credit note today against an invoice from July 2024 and the right answer is 9%. Ask our own country profile what Fiji's rate was that month and it will tell you 15%, because it holds one rate and no date. Half of our answer here is already safe and half is not; the page below says which is which, with the column names.

The subject
A rate is a fact about a country on a date. 9% to 31 July 2024, 15% from 1 August 2024 — and almost every finance system stores one number with no date attached to it.
What we get right
Your history is safe. The tax rate is stored on each invoice line when the document is raised rather than looked up when it is viewed, so a July 2024 invoice still reads 9% and always will.
What we do not
Credit notes hold no tax split, and our country profile holds no date. A credit against a pre-August-2024 supply has nowhere to record that it reverses a 9% sale, and asked what the rate was that month the profile answers 15. Both are below, with the column names.
Support hours
We are at UTC+3 and you are at UTC+12. There is effectively no shared working day — the largest gap of any market we have written a page for, and you should price it in rather than discover it.

One change, five documents

One rate change, five documents, and we get two of them right

Every row below is the same event — the rate moving from 9% to 15% on 1 August 2024 — seen from a different document. Read the last column as a verdict on us rather than on Fiji.

The change — before

9% to 31 July 2024

The change — after

15% from 1 August 2024

We get this right

An invoice raised 15 July 2024

9% — the rate in force when it was raised

The rate is stored on the invoice line as a decimal, not looked up. Reopen it, reprint it or report on it in three years and it still says 9.

We get this right

A recurring invoice generated on 5 August 2024

15% — the rate in force at supply

Each generated document resolves the rate when it is created and then stores it. A schedule crossing the change produces documents at the right rate on both sides of it.

We do not

A credit note raised today against that July invoice

9% — a credit follows the supply it reverses

Our credit note is a single amount with no lines and no tax columns at all. It cannot carry 9%, and it cannot carry 15% either. The split lives in whatever your accountant journals.

We do not

A quotation issued 20 July, accepted 5 August

A question with a real answer, and the answer depends on the terms and the time of supply

The quotation carries the rate it was issued at and converts carrying that rate. Nothing prompts anybody that the rate moved between issue and acceptance, so the check happens only if a person remembers to make it.

We do not

Asking the country profile what Fiji's rate was in July 2024

9%

It answers 15. The profile holds one rate per country with no effective date anywhere in the file or the resolver, so there is no date to ask it about.

The two rows that pass are the reason to believe the three that do not. Documents already raised are safe — the rate is on the line, it was never a lookup, and no rate change reaches backwards into your history. What is not safe is every NEW document that is about an OLD period, and the reason is the same in all three cases: the system knows what rate it used, and does not know what rate applied.

Two column names, and they are the whole disclosure

The first is `credit_notes.amount`. Our credit note is one figure with a customer, a reason and an optional link to the invoice it relates to — no lines, no tax rate, no tax columns. For most of what a credit note is used for that is adequate. Against a rate change it is not, because there is no place in the record to say that this credit reverses a supply taxed at a different rate than today's. The second is the absence rather than a column: `config/tax_rates.php` holds one rate per country and there is no effective date anywhere in the file, on any row, or in the resolver that reads it. For provisioning a new organization that is correct, because they are registering today. For any question of the form "what was the rate on this date", it is not answerable at all.

We are not going to fix either quietly and then describe it as a feature. Effective-dating a rate is not a column on a config row — it needs a rate HISTORY per country, and every consumer in the codebase resolves a country to exactly one profile today. That is a data-model change with an owner and a migration behind it, and until somebody has done it properly this page will keep saying so. What we are claiming is narrower: your history is safe, we know precisely which three cases are not, and we have written the column names down. A vendor who has never had to answer this question has not had a customer live through a rate change.

The operational half

And "in stock" is a question about where, and when the next boat is

A national stock figure is an average of places that cannot substitute for each other. The distances are short and the schedules are not, which makes availability a calendar question rather than a quantity question.

The main warehouse

Where the number in most systems comes from, and where the goods usually are not when they are needed.

Stock on hand, committed against open orders, and what is already allocated to a transfer that has not left.

A second island branch

Days away by sea, not hours by road. A stockout here is not solved from the main warehouse this week.

Its own on-hand position and its own counts, with transfers confirmed on receipt rather than on despatch.

Goods on the water

Owned, paid for, insured, and in neither location's figure. The single most common place for stock to be invisible.

An in-transit position between two of your own locations, with a despatch date and a confirmation that has not happened yet.

A resort, site or consignment location

Your goods, somebody else's premises, and usually somebody else's counting.

A location you do not own, counted against, with the same transfer discipline as any other.

None of this makes a boat come sooner and we are not going to suggest it does. What it changes is narrower and worth paying for: the difference between "we have forty" and "we have forty, of which eleven are on this island, nine are on the water since Tuesday and twenty are at a resort". The first sentence is a number. The second is a decision.

What this costs today

Four consequences, and every one of them looks like a correct record

None of these produce an error. Each is a document that is internally consistent and describes the wrong thing, which is the hardest class of mistake to find because nothing about it looks unusual.

A credit note that cannot say which rate it reverses

A return against an old supply is credited at the rate that supply carried. Where the record holds one gross figure and no tax split, the correct treatment exists only in the head of whoever posted it — and it is found, if ever, by an accountant reconstructing a year from invoice copies.

A quotation that quietly ages

Issued at one rate, accepted after a change, converted without anybody re-checking. The document is internally consistent and the total is wrong, which is the hardest kind of error to catch because nothing about it looks unusual.

One stock number, four places

A national figure that is the sum of positions which cannot substitute for one another. It is not wrong, it is unusable — and the reorder decision made from it is made blind to the only fact that mattered, which is where the goods are.

Cost that lands after the goods do

Freight, inter-island shipping, handling and clearance invoiced separately and weeks apart. If the receipt closes when the goods arrive, none of it reaches the unit cost, and the margin on a line is calculated from a price that was never the cost.

What we are actually for

Cost, stock across places, and evidence

Everything below is running today. Fiscalisation, filing and payroll are not on this list, and the scope section says so in more detail than a vendor normally would.

Stock by location, with what is on the water

Each branch, island and consignment location holds its own position, transfers are confirmed on receipt rather than on despatch, and goods in transit are a state rather than a gap. In a market where the next resupply is a schedule rather than a drive, in-transit visibility is the feature.

Landed cost that stays open after receipt

Freight, duty, VAT at import, port charges, inter-island shipping and handling attach to the consignment as their invoices arrive, with the unit cost recalculating each time. Which is the only way a cost that arrives three weeks after the goods ends up in the price of them.

The rate stored on the line, not looked up

Every invoice line keeps the tax rate it was raised at as a stored figure. This is the half of the rate-change problem we already get right, and it is worth naming as a capability rather than as an accident: your history does not move when a rate does.

Documents against the transaction

Import paperwork against the consignment, the supplier invoice against the receipt, the credit against the sale — previewable in place and retrievable by the transaction rather than by whoever filed them. Where a tax question is answered by reconstruction, this is what shortens it.

Cost coded to a place as it is entered

Branch, site, cost centre and customer captured at entry rather than reconstructed at reporting time, so per-island and per-branch profitability is a filter instead of an excavation.

Currency held at the rate actually applied

Goods bought in Australian, New Zealand or US dollars and sold in Fijian dollars keep both amounts and the genuine rate on the transaction, so a margin can be explained months later rather than recalculated from a rate nobody wrote down.

Scope, in three parts rather than two

What runs today, what we would build, and where we stop on purpose

Three columns, because "no" means two entirely different things and one list hides which is which. The middle column is work that has not been done and has a price. The right-hand column is work we would decline from a paying customer — and it is the one to demand from every other vendor on your list, because a page without it has not told you where its edges are.

Scope in Fiji, including the two column names

Running in the product today

  • The tax rate stored on the invoice line, so a document raised before a rate change keeps the rate it was raised at, permanently and without a lookup.
  • Stock by location with confirmed transfers, including goods in transit between two of your own places and stock held at a location you do not own.
  • Landed cost on the consignment, open after receipt, so freight, duty and handling invoiced weeks later still reach the goods and move the unit cost.
  • Documents held against transactions — import paperwork, supplier invoices, credits — previewable in place and retrievable by the transaction.
  • Cost coded to branch, site, cost centre and customer at entry, which makes per-island profitability a filter rather than a reconstruction.
  • Procurement approvals that refuse rather than warn above a threshold, with the trail of who authorised what against which budget.
  • Original currency and the rate actually applied retained on every transaction, alongside the Fijian dollar figures.

On the roadmap — and commissionable now

  • Our credit note holds no tax split. One amount, a customer, a reason, an optional link to the invoice — no lines, no rate, no tax columns. Against a rate change that is a real gap, because a credit reversing a supply taxed at 9% has nowhere in the record to say so. First on the roadmap in this market, because it is the smallest of the three gaps here and the one that costs a real business real reconstruction time.
  • Our country profile holds one rate and no date. There is no effective date anywhere in the reference data or the resolver, so "what was the rate on this date" is not a question the system can be asked. This is the bigger of the two and it is honestly a data-model change rather than a feature: it needs a rate history per country, and every consumer in the codebase resolves a country to exactly one profile today. Recorded in the reference file itself and pinned by a test, so a fix has to be deliberate.
  • Nothing prompts you when a rate moves under an open document. A quotation issued before a change and accepted after it converts carrying the rate it was issued at, with no flag. Once effective-dated rates exist this becomes a small rule on top of them; until then it is a human check and we would rather name it than let you assume the system is watching.
  • No output for FRCS fiscalisation. The Revenue and Customs Service operates a VAT Monitoring System that requires approved point-of-sale equipment for businesses in scope. We produce no data in any format it accepts. The data half is buildable against a published specification and against equipment you already have; the approval half is not ours and never will be — see the next column.
  • No Fijian payroll engine. No income tax tables, no provident fund calculation, no filing. We attribute labour cost to branches, sites and cost centres, which is the reporting half rather than the calculation half. Our maintained statutory payroll engine covers Kenya only, which is both the limit and the evidence that a second one is buildable.
  • No support inside your working day, and this is the largest gap in our entire market coverage. We are in Nairobi at UTC+3 and you are at UTC+12 — nine hours, with effectively no shared working day. Your morning is our night. Extending cover is a staffing decision and it is quotable, but as things stand today asynchronous support is not a detail to discover after signing, and we would rather lose the deal here than in month two.

Not ours, by choice — and this column is the reason to believe the other two

  • We will not present ourselves as approved fiscalisation equipment, or as applying to be. That status is the Revenue and Customs Service's to grant and it attaches to certified equipment rather than to a web application. If your business is in scope, the approved device is a local purchase and we would sit alongside it. A vendor who blurs "we can send the data" into "we are approved" is describing a compliance position they do not hold.
  • We will not tell you which of your supplies are zero-rated or exempt. That is a live schedule and a question about your specific goods and services. We hold the rate you configure and put the amount where you tell us to. Anyone answering that in a sales meeting is selling you a liability.
  • We will build the field; we will not be the source of truth for what goes in it. When effective-dated rates exist, the historic rates are still yours and your accountant's to confirm rather than ours to publish and maintain for every country in the profile. We are not a tax content provider and we will not quietly become one, because the day our table is stale is the day it is worse than no table.
  • We do not file, in any country whose rules we do not maintain. Kenya is the single exception and it is one we earned by building and maintaining it. Ask any vendor who offers to file your Fijian returns which of their customers has filed one, by name.
  • We will not claim a local presence. No office in Fiji, no implementation partner, no consultant who has been to your branch. Remote from Nairobi, in English, across the largest time-zone gap in this cluster. We would rather write that sentence than let you infer a bench that is not there.

Nothing in the middle column is a permanent limit, and the first two are things we want fixed for every customer rather than for one. Kenya's eTIMS transmission and its maintained statutory payroll engine exist because Kenyan clients needed them and commissioned them; neither arrived by itself. The credit note tax split is a small, well-defined build. Effective-dated rates are a genuine data-model change and we would rather quote it honestly than describe it as a configuration option. If any of these is what stands between you and a decision, say which and we will scope it as a build: written specification, timeline and price, before you commit to anything. What we will not do is print a date on this page that nobody has paid for.

Read the three columns rather than counting them. The middle one is work, and work has a specification, a timeline and a price. The right-hand one is where we stop on purpose, and it is the column worth demanding from every other vendor on your list — a page without one has not told you where its edges are. One thing deliberately absent from all three: language. English is an official language here and the language of business, so unlike our Francophone and Gulf pages there is nothing to disclose, and we are not going to pad the list to look thorough.

How this starts

Three moves, and the first two need your own system rather than a vendor

01

Find one credit note that crosses the change

A credit raised after 1 August 2024 against a supply invoiced before it. Ask what rate it was treated at and where in the system that is recorded. If the answer is a person rather than a field, you have found the subject of this page in your own data.

02

Ask your system what the rate was in July 2024

Not what an old invoice says — what the system believes the rate was on that date. Most will answer with today's rate and no warning that they were asked a question they cannot answer. Ours does exactly that, which is why it is on this page rather than in a footnote.

03

Then decide which half you are buying

Fiscalisation, filing and payroll are local purchases and we have said so above with the column names to back it. What is left is cost, stock across places, evidence and control — a smaller conversation than most vendors will let you have, and the one we can stand behind.

Questions we are asked here

Straight answers, including the two column names

Our old invoices still show 9%. Is that a bug?

No — that is the system working correctly, and it is the half of this problem we already get right. The tax rate is stored on each invoice line as a figure at the moment the document is raised, not looked up when the document is viewed. So a July 2024 invoice shows 9% today, will show 9% in five years, and would show 9% even if the rate moved three more times. A system that re-resolved the rate on display would rewrite your history every time a budget passed. If you are checking whether your own system does this, raise nothing and change nothing — just open a document from before the change and see whether the rate on it is the rate it was raised at.

What happens when we credit an invoice from before the rate change?

Honestly: less than it should. Our credit note is a single amount with a customer, a reason and an optional link to the invoice it relates to. There are no lines and no tax columns, so there is nowhere in the record to say that this credit reverses a supply taxed at 9% rather than at today's 15%. The correct treatment is not in doubt — a credit follows the supply it reverses — but the system does not hold it, which means it lives in the journal your accountant posts and in their memory of why. **This is on the roadmap rather than a boundary and it is commissionable now.** It is the smallest of the three gaps on this page: lines and a rate on the credit note, defaulting from the invoice being credited. If it is what stands between you and a decision, say so and we will come back with a written specification, a timeline and a price. Kenya's eTIMS transmission exists because a client commissioned it, which is the only reason we make that offer with a straight face.

Can the system tell us what the VAT rate was on a given date?

No, and it will answer confidently anyway, which is worse than refusing. Our country profile holds one rate per country and there is no effective date anywhere in the reference file or in the resolver that reads it — so asked about July 2024 it returns 15, because 15 is the only number it has. We have recorded that in the reference file itself, published it here, and pinned it with a test. **It is on the roadmap rather than a boundary, and it is the larger of the two builds on this page**: effective-dating a rate needs a rate history per country rather than a date on a row, and every consumer in the codebase currently resolves a country to exactly one profile. We would scope it, price it and do it properly rather than bolt a date onto a config file and call it solved. And there is a matching limit in the right-hand column of the scope section that is not a build at all: we will provide the field, and what goes in it for a country we do not file in stays yours and your accountant's to confirm.

We use approved point-of-sale equipment for FRCS. Can you replace it?

No, and this one is a boundary rather than a backlog. The Revenue and Customs Service operates a VAT Monitoring System requiring approved point-of-sale equipment for businesses in scope, that approval attaches to certified equipment rather than to a web application, and it is the Service's to grant rather than ours to claim. We would sit alongside what you already have rather than in front of it. **You should want that answer**, because the alternative is a vendor who blurs "we can produce the data" into "we are approved" — and you discover which they meant during an audit. The data half is different and is genuinely buildable: producing sales data in a published format, against equipment you already own, is a specification-and-price conversation like any other.

How do you handle stock across islands?

As separate places rather than as one number, which sounds obvious and is the thing most systems get wrong by default. Each branch and island holds its own position, transfers between them are confirmed by the receiving location rather than by the sender, and goods on the water are an in-transit state instead of a gap in both figures. What that gives you is the difference between "we have forty" and "we have forty, of which eleven are here, nine have been on the water since Tuesday and twenty are at a resort". What it does not give you is any influence over when the next boat sails — we make the schedule visible and costed and we are not going to imply anything more than that.

What are your support hours, really?

We are in Nairobi at UTC+3 and you are at UTC+12. There is effectively no shared working day: your morning is our night, and your afternoon is our very early morning. This is the largest time gap of any market we have written a page for and we are putting it in the FAQ rather than in a footnote because it is the kind of thing that is fine in the demo and corrosive in month two. In practice support is asynchronous — you write, we answer within our day, you read it the next morning — which works well for configuration questions and badly for anything urgent. **Extending cover is a staffing decision rather than an engineering one and it is quotable like anything else**, but as things stand today it is a real cost of buying from us and you should price it in rather than discover it. If same-day response inside your working hours is a requirement rather than a preference, a regional vendor is the rational choice and we would rather say so.

Do you do Fijian payroll?

No. No income tax tables, no provident fund calculation, no filing, no statutory forms. What we do is attribute labour cost to branches, sites, cost centres and jobs, which is the reporting half of payroll rather than the calculation half — useful alongside a local payroll, useless as a replacement for one. **It is on the roadmap rather than a boundary and it is commissionable now**, on the same basis as everything else here: our maintained statutory payroll engine covers Kenya only, and it exists because a Kenyan client needed it and paid for it to be built and kept current. That is the precedent and also the warning — a payroll engine is not a one-off build, it is a maintenance commitment, and we would quote it as one rather than as a project with an end date.

Why would we buy from a vendor who has never been here?

Possibly you would not, and the page is written so you can decide that quickly rather than in month three. What we can offer instead of proximity is checkability: this page names two things in our own schema that are inadequate for a Fijian rate change, gives the column names, and there is a test in the repository pinning both so that a fix has to be deliberate. That is a kind of evidence a reference story is not. The rest is ordinary and unflattering — no office, no implementation partner, no shared working day, remote support in English. Ask us what happens when the person who implemented your system leaves. Then ask every other vendor on your list what their system does with a credit note against a pre-August-2024 invoice, and see which of them has thought about it.

One credit note, one date

Find a credit raised after 1 August 2024 against a supply invoiced before it, and ask your system which rate it was treated at. Bring the answer and we will tell you honestly which half of your problem we are for.