T661 Writing

Preparing a T661 Narrative Your CPA Can Actually Work With

A T661 project narrative covers technological advancement, systematic investigation, and the technical context that made new knowledge necessary. Here's how each part should read — and what CPAs typically flag.

Daniel Croft · Founder Finance Contributor 2026-04-01 6 min read

What your CPA sees when they open your draft

When a CPA sits down to review a T661 narrative, they're looking for three things in roughly this order: does the advancement section describe a technical outcome at the right level of specificity, does the uncertainty section explain why investigation was the necessary path, and does the investigation section show what was actually done — including dead ends.

Most founders get the concepts right. They understand their own work. What produces revisions is structure: leading with business outcomes instead of technical ones, writing uncertainty as difficulty rather than knowledge gap, and treating the investigation section as a project summary rather than an experiment log.

Three questions, three answers

The structure

A T661 narrative answers three questions in order: (A) What advance in technology or knowledge was attempted or achieved? (B) What technical gap made systematic investigation necessary? (C) What did you specifically do — what approaches, what experiments, what findings, including failures?

CRA allows the narrative to be written as flowing prose rather than three labeled sections. But the underlying structure must answer all three questions — and in roughly this order. Starting with C (what you did) before A (what advance you achieved) is the most common structural mistake.

Section A: the advance — lead with technology, not product

The advancement section should answer: what specific advance in technology or scientific knowledge was attempted or achieved by this project? Focus on the technical level, not the business outcome.

Before (too product-focused)

"We built a recommendation engine that improved conversion rates by surfacing relevant products to users."

After (technical advance)

"We developed a cold-start recommendation approach for sparse user-interaction datasets that extended beyond published collaborative filtering methods, enabling meaningful recommendations with fewer than N user interactions by incorporating [specific technical mechanism]."

Same project. The first version describes a business outcome. The second describes a technical advance. CRA reviews the second kind. Business outcomes belong in section C as context for why the advance matters — not in section A as the advance itself.

Section B: uncertainty as context, not difficulty

The uncertainty section explains why the work required investigation rather than simply applying existing knowledge. It's context for sections A and C, not a standalone test.

Better framing

"No documented method resolved [specific technical constraint] for [specific data characteristics or requirements]. The absence of applicable prior art in the public domain made systematic investigation the necessary path forward, rather than implementation of existing techniques."

Ordinary development challenge language — 'this was complex to implement,' 'our team hadn't done this before,' 'the codebase was large and hard to modify' — describes effort, not technological uncertainty. A CPA will flag it and ask you to reframe. Better to identify the exact knowledge gap before you start drafting.

Section C: investigation — show the path, including the dead ends

This section should answer: what did you specifically do to investigate? What approaches did you test or analyze? What intermediate findings did you uncover? What was the final outcome, including work that turned out to be a dead end?

Include failed approaches explicitly. A narrative that describes only successful paths raises questions. Systematic investigation involves trying things that don't work — those failures are legitimate, claimable SR&ED outcomes. An investigation that only succeeded is either unusually lucky or underreported.

CRA reviewers look for named approaches, specific experiments or analyses, intermediate findings, and a clear link between the investigation and the uncertainty identified in section B. 'We tried several approaches' is not specific enough. 'We investigated [approach A] using [method], which produced [finding], leading us to reject it in favour of [approach B] because of [reason]' is.

Getting to a draft your CPA can work with efficiently

Draft section by section, not from memory as a single essay. Start with the advance (A), then the uncertainty context (B), then the investigation (C). Ask the engineer who did the work to read and flag anything technically imprecise — if they'd add technical detail, a CRA reviewer would likely want it too.

Attach your evidence list — PRs, design docs, test logs — so your CPA can match the narrative to the evidence and identify gaps before review. Leave 3–4 weeks for CPA review and revisions before the SR&ED filing deadline. Complex claims can require multiple revision rounds.

This guide is for general educational purposes only. T661 narrative requirements depend on the specific facts of each SR&ED project. All SR&ED claims must be reviewed by a qualified Canadian CPA before filing.

T661 narrative CPA filing project description