In brief
ZATCA illustrates possible methods without promising approval. Correct restrictions and direct attribution first, then demonstrate comparative accuracy and follow the notification's actual scope, period and conditions.
A proposed alternative apportionment method needs to show how eligible shared inputs are used more accurately than the correctly constructed default. A management cost-allocation model can provide evidence, but it is not a VAT approval and may measure a different commercial objective.
Article 51(8)-(10) supplies the application and direction framework. This guide explains method design; the approval guide covers the notification and application conditions. None of the original calculations below is a ZATCA-approved method for a particular institution.
Establish the eligible pool before designing a driver
Apply expenditure restrictions and non-economic-use analysis first. Remove input VAT exclusively and directly attributable to qualifying taxable supplies and to exempt supplies under Article 51(1)-(2). Only the eligible shared economic pool requires proportional deduction.
In ZATCA’s Financial Services Guideline, second edition, May 2026, section 8.6, Example 24, page 61, direct taxable VAT of SAR 22,500 and exempt-only VAT of SAR 9,000 are treated separately. The 40% fraction applies to SAR 30,000 shared VAT, giving SAR 12,000. Applying a proposed sector percentage to every purchase would therefore start from the wrong pool.
Reconcile the default’s supply values, geography and exclusions before alleging distortion. A classification error is not evidence that an alternative method is needed. See the financial-institution recovery ratio.
What ZATCA’s official method illustrations establish
The Input Tax Deduction Guideline, third version, May 2026, section 13.1, unnumbered method table, page 57 gives possible methods and suitable circumstances:
| Official design illustration | Condition that makes the measure relevant |
|---|---|
| Transaction numbers | Each transaction uses a similar quantity of inputs regardless of its value. |
| Input-based measure | Particular supply values distort the comparison with relative input use. |
| Employee/FTE measure | A high proportion of inputs relates to employees and time can be accurately tracked. |
| Floor space | Rental and associated costs form a high proportion of inputs. |
| Sectoral method | Distinct sectors have separately measurable non-attributable input VAT, with a calculation for each sector. |
These are official design illustrations, not numerical taxpayer approvals. The table expressly requires a verifiable basis, does not promise approval for a particular method and does not exhaust the available alternatives. A label such as payments sector is not proof that all its transactions consume equal resources.
Define sectors from actual activities and records
A useful sector boundary distinguishes activities whose shared input use can be measured reliably. Explain the products, supplies and costs in each proposed sector, including functions that serve more than one sector. A department, legal entity and VAT sector are not necessarily the same thing.
Head-office support might be allocated to measured sectors or retained in a further shared pool under the proposed method. There is no general rule in the reviewed table that head office can never be a sector, or that an unsupported sector ratio is automatically a fallback whenever detailed data is missing.
For an insurer, actual general/life policy administration and distribution arrangements matter. For a fintech, a distinct lending activity may first require direct exempt attribution rather than a complex sector method. The design should explain those facts before selecting percentages.
Match each driver to the costs it purports to measure
We recommend documenting the causal link between each pool and its proposed driver. Platform processing data may support some technology costs; occupied area may support some premises costs; reliable time/FTE data may support some people-related services. These are candidate measures requiring evidence, not universal strong/weak rankings prescribed by ZATCA.
Transaction counts can fail when events differ materially in complexity. Headcount can fail when a small team uses unusually expensive systems. Supply values can remain appropriate for a particular pool or sector even when they distort another. Compare the alternatives using the relevant data rather than reject revenue or select floor area automatically.
The VAT pool, driver denominator and qualifying-use numerator must also have consistent scope. Do not divide taxable transaction counts from one platform by all events in the group or weight a VAT deduction using an unrelated total-payroll statistic.
Dariba illustration: a method can reduce recovery
Assume restrictions, non-economic use and direct attribution have already been dealt with. Eligible shared input VAT is SAR 1 million, and a correctly constructed default gives 40%, or SAR 400,000 recovery. Original Dariba facts are used below; reliable supporting data is assumed solely to illustrate the calculation.
| Shared VAT pool | Input VAT | Proposed supported qualifying-use share | Proposed deduction |
|---|---|---|---|
| Technology | SAR 600,000 | 25% | SAR 150,000 |
| Premises | SAR 300,000 | 40% | SAR 120,000 |
| Other shared services | SAR 100,000 | 30% | SAR 30,000 |
| Total | SAR 1,000,000 | 30% weighted result | SAR 300,000 |
The proposed result is SAR 100,000 below the default, not an entitlement to switch methods immediately. The 30% aggregate reflects the VAT amounts in these pools; an unweighted average of 25%, 40% and 30% would be a different calculation and cannot replace the pool-by-pool working.
Dariba’s analysis. A lower proposed outcome can reveal a need to examine whether the default reflects use, but neither a higher nor a lower result proves comparative accuracy. The supporting measures must actually represent input use and the applicable ZATCA approval or direction must be obtained.
Build a reviewable comparison and application
Financial Services section 8.7, page 62 calls for the default’s impact, the proposed allocation factors and an explanation of why the alternative accurately reflects input use. It requires formal approval specifying the period before an alternative may be used.
We recommend reconciling each pool to invoices/returns, each driver to a reproducible system report, shared support to its allocation and every excluded cost to its reason. Compare periods that are representative of actual activity and show sensitivity to missing or unstable data. A three-year history can be useful on appropriate facts, but the reviewed provisions do not prescribe a universal three-year back-test or a standard acceptance threshold.
Explain differences from management, transfer-pricing or regulatory models. Those models can measure profitability, risk or cost ownership rather than VAT input use. Consistency is useful evidence; identical results are not a legal requirement established by the source table.
Operate within the notification rather than invent operating rules
Article 51(9) allows an approval period of no more than five years; the actual period and terms come from the notification. Article 51(10) permits a ZATCA direction. The default’s annual actual-values reconciliation does not itself prescribe identical previous-year-driver and final-quarter rules for every bespoke method.
Include proposed recalculation, changes in activities, missing-data handling and renewal arrangements in the application, then follow the terms actually approved. An internal materiality threshold is a recommended control proposal, not a statutory exemption allowing costs or changes to be ignored. Do not infer unilateral switching, automatic retrospective operation or guaranteed approval from the guideline’s possible-method list.
Use the design review to correct both directions
The useful outcome may be a supported application, a correction to direct attribution, better measurement under the default or recognition that further information is needed. A well-documented model makes those decisions visible. It does not guarantee an improved percentage, a fixed term or recurring savings.
Continue with recovery review decisions and the financial-institution VAT framework.
Key takeaways
- Only eligible shared economic input VAT requires a proportional method after restrictions and direct attribution.
- ZATCA's unnumbered method table gives possible transaction, input, FTE, floor-space and sector measures with suitability conditions.
- A management allocation model is evidence to test, rather than a VAT approval or a required identical result.
- The original three-pool model gives SAR 300,000 recovery versus SAR 400,000 under the assumed default; neither outcome proves accuracy by itself.
- No universal three-year back-test, materiality safe harbour or automatic sector fallback is established by the reviewed sources.
- Formal approval or direction and the actual notification terms control operation; the maximum five-year period is not a guaranteed term.
Sources
- ZATCA - VAT Implementing Regulations, Arabic tenth edition, April 2025Article 51, PDF pp51-53; controlling Arabic text.
- ZATCA - Input Tax Deduction Guideline, third version, May 2026Sections 13.1-13.3, pp56-58; unnumbered possible-method table p57.
- ZATCA - Financial Services Guideline, second edition, May 2026Section 8.6 Example 24 p61; section 8.7 approval and comparative report p62.
General information only. Actual supplies, facts, applicable periods and controlling Arabic legislation determine the result. Official examples are distinguished from Dariba analysis and original illustrations. Seek advice or clarification where the arrangement or source material leaves a specific issue unresolved.