Year-end true-ups tell you how far you drifted. They cannot tell you whether you should have. That takes a value chain to measure against, and a memory deep enough to hold one consistent story across every filing.
The year-end true-up is a poor destination
Most multinationals still run transfer pricing on a twelve-month loop. Policies are fixed in the budget cycle. Intercompany transactions run at provisional prices. In the last weeks of the year someone works out how far each entity landed from its target margin and books an adjustment. Documentation follows, and it describes a position that was settled before anyone wrote it down.
Anyone who has run that process knows where it hurts. The adjustment is large because it clears a year of drift in one entry. It arrives after the commercial year is over, so the cause can no longer be fixed, only the number. Tax, customs and VAT treatment of late adjustments differs by jurisdiction, and some of those differences are expensive. By then the person who knew why the Singapore entity ran hot in the second quarter may have moved on.
Operational transfer pricing is the better end state. Prices are set, monitored and corrected inside the financial calendar. At each close, actual results are set against what the policy expects. Variances are looked at while the facts are fresh, small corrections are made as they arise, and what reaches year-end is a short list of explained exceptions. The group still has a true-up, but it is a minor one, and every line of it has a reason attached.
Finding the deviation is only half the job
Most of the discussion about operational TP is about data. Pull the intercompany ledgers and trial balances out of the ERP, calculate entity margins every month, flag whatever falls outside range. That work is necessary, and it is hard enough that many groups have not done it. But it answers a single question: where did results move?
An auditor asks a different one: should they have moved?
Take a limited-risk distributor with a policy range of 3% to 5% operating margin. It closes the third quarter at 2.1%. That is a number, and what it means depends on what the entity is supposed to be doing. If it took on market development work this year, 2.1% may contradict both the policy and the facts. If it bears no inventory risk and lost a major customer, 2.1% points to a pricing error that needs correcting now. If its risk profile has changed, the range itself is wrong. One number, three different responses.
A deviation only has meaning against a reference, and a policy range is a good reference only if it reflects how the business really works. That is why the baseline for operational TP should be the Value Chain Analysis. The VCA records the functions each entity performs, the assets it uses, the risks it bears and who controls them, who develops, enhances, maintains, protects and exploits the intangibles, and from all of that, where profit should sit. In this model the VCA is not a project done for a restructuring or a documentation refresh. It is the live specification that actual results are tested against.
With the VCA as the framework, every deviation can be accounted for in one of four ways:
Commercial. Volume, price, currency or cost movements that fit the entity's role. Explain, record, leave the policy alone.
Operational. Errors or lags in invoicing, allocation or booking. Correct in the current period.
Policy. The range or method no longer fits the facts. Send it to policy review.
Structural. The value chain itself has changed: a function has moved, a risk has been reallocated, a new asset exists. Update the VCA first, then let the policy follow.
Done this way, the year produces a record. Each variance has a classification, an owner, a rationale and a date. That record is worth more in an audit than any amount of year-end reconstruction.
One story across every CbCR
The group is no longer the only party with a view of its own profit distribution. Country-by-Country Reporting hands tax administrations the same table of revenue, profit, tax, employees and tangible assets for every jurisdiction in the group. The OECD's Corporate Tax Statistics 2026 reports that 120 jurisdictions have mandatory CbCR laws for fiscal year 2026. Authorities use the data for high-level risk assessment, and they read it against the local files, the master file, published accounts and each other.
So the group should read the relevant CbCRs together, and in the same way. That means the parent filing and any surrogate or local filings, across the years in sequence, set against local files, the master file, statutory accounts and the agreements underneath them. The questions are the ones an examiner would ask:
Does profit sit where the VCA says value is created?
Where a jurisdiction shows high headcount and tangible assets but thin profit, is there an explanation on file?
Does the profit per employee in one country fit the functions described in its local file?
Do the in-year results, the intercompany agreements and the reported figures agree?
CbCR is a risk assessment tool and not a basis for adjustment, but it decides which groups get looked at first. A group whose documents tell slightly different stories gets examined. That usually happens because the documents came from different teams, at different times, built from different data extracts. Consistency cannot be produced in the final weeks before a filing. It has to be maintained all year.
Why this needs an enterprise memory
In-year deviation accounting against the VCA and a consistent reading of the CbCRs depend on the same capability: finding and joining facts across systems and across time. That is a memory, in the proper sense of the word, and a document store is not one.
A transfer pricing memory has to hold what was true (entities, roles, agreements), what was decided and why (policies, approvals, benchmarks), what happened (ledgers, closes, adjustments), what was said to the authorities (filings, local files, CbCR), and when each of those things was known. It needs to resolve entities consistently, so that one subsidiary has one identity in every source. It needs versioning, so that the policy in force in March can be told apart from the policy in force in November. It needs lineage from every figure back to its source, and live links between an agreement, the policy it supports, the transactions that follow it and the filing that reports them.
Without that, operational TP becomes a monthly spreadsheet exercise, and the multi-CbCR comparison becomes a scramble before each deadline.
A solution architecture
Put together, the design has five layers. Each depends on the one beneath it.
Layer | What it does | What it produces |
|---|---|---|
1. Enterprise memory | Ingests ledgers, agreements, policies, benchmarks, filings and rulings; resolves entities; versions everything; keeps lineage to source. | One indexed, time-aware record of facts and decisions. |
2. VCA baseline | Models functions, assets, risks and DEMPE by entity and value chain, and derives the expected profit allocation. | The reference every deviation is measured against. |
3. Operational monitoring | At each close, ingests actuals, compares them to the VCA-derived expectation and the policy range, and routes variances. | A deviation register, classified as commercial, operational, policy or structural. |
4. Consistency engine | Builds the CbCR view from the same data and reads it across jurisdictions and years, reconciled to local files, master file and accounts. | Contradictions found before filing, not after a query. |
5. Judgement and defence | Puts human sign-off at the points that need it and assembles the evidence for any position on demand. | An audit-ready trail built as decisions are made. |
Two design choices matter more than the rest. First, the VCA sits in the middle, not at the edge. Monitoring, consistency checks and defence all reference it, and when the business changes, it is the first thing to change. Second, professional judgement is built in at defined gateways. Software can detect, classify and compile. Deciding whether a variance is commercial or a sign that the policy has failed is a judgement call, and a person should make it and be on record as having made it.
Where infer360 fits
The infer360 TP Intelligence Stack was designed around this reading of the problem. The pieces map directly:
My Repository and Enterprise Tax Memory form the memory layer: a single, organised and indexed home for global TP data, with continuity between data, agreements, policies and supporting information.
Value Chain Analysis is the baseline, kept current so that it describes the business as it operates now.
Plan and Operate carry the VCA into practice. Plan supports scenario modelling, policy design and risk assessment. Operate compares monthly or quarterly close results with policy so that pricing drift is visible while it can still be addressed.
Benchmark and Document draw on the same underlying data for comparables work, Master File, Local File and Country-by-Country Reporting, which is what lets the documents agree with each other.
Defend retrieves the evidence, the reasoning and the history behind a position when an authority asks for it.
Human-in-the-loop gateways keep qualified people in charge of the judgement calls. The result is what the #BuiltToDefend philosophy has always argued for: governance, evidence and consistency in place before an audit starts.
Book a free TP Diagnostic Workshop
Groups start from different places. Some have sound policies and no in-year monitoring. Others have the data scattered across ERPs, shared drives and inboxes, and a VCA that was last updated for a restructuring. The infer360 TP Diagnostic Workshop looks at your current operating model, shows where the gaps and inconsistencies sit, and sketches what operational TP would look like for your group.
Book a free TP Diagnostic Workshop to see how your CbCR story reads today, and what it would take to make it one story.
Frequently asked questions
What is operational transfer pricing?
Operational transfer pricing means setting, monitoring and correcting intercompany pricing during the financial year, using actual close results measured against policy, so that year-end adjustments are small and explained rather than large and reconstructed.
Why is a Value Chain Analysis the baseline for operational TP?
A margin outside its policy range only becomes meaningful when compared with what the entity actually does. The VCA records functions, assets, risks and intangibles development, so it shows whether a deviation is commercial, an error, a policy problem or a structural change.
Why should CbCRs be evaluated together?
Tax administrations compare CbCR data with local files, the master file, accounts and other filings. Reading them together first lets a group find and resolve inconsistencies before an authority does.
What is an enterprise memory stack for transfer pricing?
It is a connected, versioned and indexed record of the facts, decisions, transactions and filings behind a group's transfer pricing, with lineage to source, so that any position can be traced and any two documents can be checked against each other.
Does technology replace professional judgement?
No. Technology detects variances, classifies them and compiles evidence. Professionals decide what a variance means and what to do about it.