Module 6: Disputes and audits

After this module you will be able to manage a review of a client's claim from the first letter to the closing agreement or appeal: answer an information request in a form the reviewer can use, prepare the technical lead without scripting them, concede early the items that should be conceded, defend the items the evidence supports, and keep the client informed throughout. Review is used here for every form of post-filing scrutiny.

What this module covers

How a review starts, what it tests and how to set the client's expectations.
Answering the information request: order, indexing, references and deadlines.
Using the technical lead effectively in written responses and meetings.
Conceding weak points early and defending strong ones.
Preserving the client relationship through the process.

How a review starts and what it tests

Every one of the seven authorities selects claims for review on a risk basis, and some also run random programmes. A review is therefore not an accusation, and the client should hear that first. The second thing to explain is what the reviewer is trying to establish, which is the same everywhere: was there a real technical unknown; did the people claimed actually work on it; do the costs reconcile; and is the story the same in every document. The third is how the process runs: questions in writing, a deadline for each response, possibly a meeting with the technical staff, a written finding and a right to challenge it.

Answering the information request

Reviews commonly begin with a written request for project descriptions, cost schedules and supporting evidence. Read every question and map it to the file before drafting anything; a question you cannot map to a document is a weak point, and you want to know now. Answer the questions asked, in order, each with a short statement of the position followed by references to specific documents by name and date. Index the documents and attach only those cited. Never present a document created during the review as a record made at the time; if something has been reconstructed, say so and say from what.

What is requested	What the reviewer is testing	What to send	Pitfall
Project descriptions	Whether there was a technical unknown and the narrative matches what was filed	The filed narrative and the full file version, with the evidence index	A rewritten narrative that differs from the filed one
Evidence of uncertainty	Whether the unknown was stated before the work, not after	The earliest dated document stating the unknown, and the baseline records	Undated documents, or metadata showing creation during claim preparation
Evidence of the work	Whether the work was systematic	Test logs, experiment records, tickets and records of abandoned approaches	Success stories only; the failures are the evidence
Basis of time allocations	How the staff proportion was arrived at	Time records or the documented method, reconciled to payroll	Round number allocations applied to whole departments
Cost reconciliation	Whether the schedule ties to the ledger, payroll and invoices	The reconciliation, with ledger extracts and invoices for sampled items	A schedule that cannot be traced to the accounts
Contracts and funding	Who bore the risk, who owns the results, where the work was done	Contracts, grant and intercompany agreements with the relevant clauses marked	Discovering during the review that the work was funded or performed for another person
A meeting with technical staff	Whether the people who did the work recognise the narrative and can explain the unknown	The technical lead, prepared as set out below	Sending the sales lead or a manager who was not there

Using the technical lead

If the review proceeds to a meeting, the reviewer wants to speak to the people who did the work. Prepare the lead by walking through the narrative and the evidence together, not by scripting answers. Explain what the reviewer is trying to establish and that conceding the routine parts of the project strengthens the claim. Agree the division of labour: the lead answers technical questions in their own words and refers to real documents; the practitioner answers questions about the scheme, the calculation and the file. A lead who explains the uncertainty plainly and concedes the parts that were routine is the best asset a claim has; a lead coached to use the words of the definition is the worst.

Conceding weak points early

Before the first response goes out, review the claim as the reviewer will and sort the items into three groups: the evidence supports the position; the evidence is thin or reconstructed; the activity sits within an exclusion or the cost fails a category rule. The second and third groups are weak points. Remove them, recompute the claim, and state the revised position and the reason in the response. In several countries a disclosure made before or during a review reduces the penalty that would otherwise apply, and the country pages state the rules.

Conceding is not capitulation. Where the file holds a dated statement of the unknown, records of the systematic work and a reconciled cost, defend the item in writing, with references, through each stage the country's process allows; every one of the seven countries provides a route to challenge a finding. The decision to escalate belongs to the client, taken on your written advice.

Preserving the client relationship

Clients lose confidence in a review for three reasons: surprises, silence and blame. Prevent surprises by telling the client at the start what the process is, what the weak points are and what the range of outcomes looks like. Prevent silence with a short written update at each step. Prevent blame by never putting technical staff in the position of defending a narrative they did not write. When the review closes, write a short lessons note for the file and the client: what was challenged, what was conceded and why, what was defended and why, and what changes in next year's process.

Illustrative example

Illustrative example. Harbour Robotics Ltd, a hypothetical company, filed a claim on eligible costs of 100,000: 70,000 for the gripping algorithm project and 30,000 for a project to integrate a new conveyor controller. The tax authority opens a review and asks for the project descriptions, the evidence of uncertainty for each project, the basis of the time allocations and the cost reconciliation. Mapping the questions to the file, the practitioner finds that the gripper project has a dated kickoff note, test logs for four approaches and a reconciled schedule, while the conveyor project has a narrative written at claim time and no record that predates the work. The response concedes the conveyor project, recomputes the claim on 70,000 and says why, and answers the remaining questions in order with references to eleven named documents. The reviewer accepts the revised claim, and the benefit is recomputed on 70,000 at that country's rate.

Across the seven countries

Country	How the review runs
United States	Research credit claims can be selected for examination through the IRS's normal return selection processes. Taxpayers may request review of an examination determination by the IRS Independent Office of Appeals before paying the proposed deficiency.
United Kingdom	HMRC selects R&D claims for compliance checks on a risk basis using the information on the additional information form and other data. Information requests in a compliance check usually set a 30 day deadline, which can be extended on request.
Ireland	The Code of Practice for Revenue Compliance Interventions provides for Level 1 compliance support interventions, Level 2 risk reviews and audits, and Level 3 investigations. A qualifying disclosure made before a Level 2 intervention begins reduces the penalty.
Canada	The CRA risk-assesses each SR&ED claim on receipt and either accepts it as filed or selects it for a financial review, a technical review or a coordinated review. The RTA communicates preliminary findings and provides an opportunity to present additional information.
Australia	The department selects registrations for examination on a risk basis, and before making a negative finding issues a notice of its proposed finding and gives the applicant an opportunity to respond. A finding binds the Commissioner of Taxation.
New Zealand	Inland Revenue selects claims for review on a risk basis and may engage external technical experts. The disputes process requires a notice of proposed adjustment, a notice of response within 2 months, a conference phase, statements of position and adjudication.
Singapore	IRAS may request further information on R&D claims before finalising the assessment, including project descriptions, the basis for concluding that the three requirements are met, and the expenditure breakdown.