An Electrician's Markup Assumption Meets the Real Invoices

The situation

The letter that reached Tamar's mailbox did not open with a question. It opened with a number: a projected shortfall in the six figures, built from a formula a CRA auditor had applied to two years of business activity. Tamar had never been through an audit before, and that number made the rest of the letter hard to read.

Tamar worked as an electrician, but not in the traditional employed sense. Most of the work came through gig-style dispatch platforms that matched licensed electricians with homeowners and small commercial clients around Waterdown, jobs that ranged from a two-hour panel swap to a week rewiring an older house. Materials were bought as needed, sometimes marked up modestly and passed through to the client, sometimes billed close to cost when a platform set the price itself. Records existed, but they lived scattered across several apps, a shoebox of paper receipts, and a spreadsheet Tamar updated inconsistently between busy stretches, a pattern common enough among gig-platform tradespeople that Tamar had never thought of it as a risk until the letter arrived.

When the CRA selected the file for review, the auditor did not ask for the full two years of records up front. The auditor asked for a sample, and Tamar provided what was easiest to find quickly - a folder covering roughly thirty jobs pulled from whichever platform's app happened to export data cleanly. From that sample the auditor calculated a single average markup percentage on materials, then applied that one rate to every job across the whole audit period, turning a small and possibly unrepresentative slice of the business into a projected income figure far above what Tamar had actually reported. Neither the platforms nor the supplier accounts that fed the real picture were consulted before the number was set.

Dimitri, Tamar's bookkeeper, flagged the number as wrong within a day of seeing it, but neither of them had handled a projection assessment before, and the CRA's cover letter set a response deadline that left little room to build a full rebuttal from a standing start. Dimitri could see the shape of the error - a markup rate from a handful of higher-margin jobs stretched across two years that included plenty of jobs priced at or near cost - but proving that shape with evidence, rather than simply asserting it, was a different undertaking entirely. That was the point at which Tamar called our office.

What made this urgent

Projection assessments carry a particular danger: once the CRA extrapolates a rate across a full period, that projected number becomes the default position unless the taxpayer can show, with evidence, that actual results differed. The burden effectively shifts. Tamar was not being asked to prove innocence in the ordinary sense - the audit had already assumed the extra income existed, and any response had to demonstrate, invoice by invoice where possible, that the sampled markup rate did not hold across the whole period.

The urgency had three layers. First, a finalized reassessment would carry the disputed amount plus arrears interest running from the original filing deadlines, not from the date of the letter, so every month of delay compounded what Tamar would eventually owe if the projection stood as written. Second, Tamar's income as a gig-platform electrician was already uneven year to year, and a reassessment landing anywhere in the roughly $50,000 to $150,000 range the file was heading toward was not a sum an ordinary contractor household could absorb without serious disruption - a payment arrangement at that scale would follow Tamar for years.

Third, and least expected, the file did not stay still. Partway through the response window, the audit was reassigned, and the officer who took it over, later working alongside Kostas on the appeals side once the matter moved past the audit stage, took a different view of what counted as adequate substantiation. The original auditor had been willing to accept invoice totals summarized by supplier. The new reviewer wanted supplier-level detail matched to specific jobs, which meant work already prepared had to be reformatted, not built from nothing, but reformatted under a shortened timeline that arrived without warning.

That kind of mid-file shift is not unusual once a case moves between officers, but it caught Tamar off guard at a point when the file had looked close to settled on the original terms. The urgency, in the end, was not only the size of the projected number. It was that the target kept moving, and each move reset the clock on how quickly a defensible record had to be produced.

What we did

1. Pulled the complete transaction history, not just the sample the auditor had relied on - full platform statements, supplier account records, and bank activity for the entire audit period, going back to Tamar's earliest dispatch account - because a projection built on thirty jobs could only be effectively challenged with evidence covering all of them, and partial records would have left the auditor's rate unchallenged for everything outside the original sample.
2. Reconstructed the real markup job by job, matching each material purchase against the invoice it was billed on across roughly two hundred jobs, which showed the markup varied widely by job type and by which platform had priced the work, rather than sitting at one flat rate as the auditor's formula had assumed, undercutting the core premise the projection was built on.
3. Separated platform-priced work from independently priced work, since jobs where a dispatch platform set the client price left Tamar with little to no markup at all, a category the original thirty-job sample had underrepresented. That gap skewed the auditor's blended rate upward, because the small sample leaned toward the higher-margin independently priced jobs, and correcting the mix materially lowered the true average once every job across the full two years was counted properly rather than extrapolated from a slice that was never representative to begin with.
4. Rebuilt the submission in supplier-matched format once the file moved to a new reviewer who wanted detail tied to specific suppliers rather than summarized totals. Reformatting work already completed, rather than starting the reconstruction over from nothing, kept the response on schedule despite the change in expectations, and it mattered because a reviewer inheriting a file mid-audit is less inclined to extend goodwill on a late or incomplete package than the original auditor had been.
5. Corresponded directly with the appeals contact, Kostas, to walk through the revised submission line by line rather than relying on the written package alone. A submission of that size often gets read against the auditor's original assumptions unless someone is available to explain the gaps and answer questions as they come up, and the conversation surfaced two supplier records the reviewer had initially missed, which strengthened the file further before any decision was made.
6. Addressed the penalty question separately from the income question, arguing that inconsistent but genuine recordkeeping, kept by someone running a one-person trade business between jobs, did not amount to the kind of conduct that supports the heaviest penalties, since the two issues are assessed on different standards and conflating them in the response would have cost more than it saved.
7. Negotiated a reduced reassessment once the supplier-matched evidence narrowed the disputed range substantially, accepting a smaller adjustment for the stretch of the audit period where Tamar's own records genuinely had gaps that no amount of reconstruction could close, rather than pushing every dollar into a longer and costlier dispute that the remaining gap in the record was unlikely to win outright.
8. Set up a recordkeeping structure going forward, consolidating invoices, supplier statements, and platform payout summaries into one running file updated weekly rather than in occasional catch-up sessions. The point was to make sure that if another review ever came, the fuller record would already exist rather than needing to be rebuilt from scratch under deadline pressure the way this one had to be the first time.

The outcome

The final reassessment landed well below the CRA's original projection, with the supplier-matched invoices carrying most of the weight in bringing the number down from the six-figure starting point to a fraction of that within the disputed range. But it did not land at zero. For a portion of the audit period, particularly the early months where Tamar's records were thinnest and several supplier receipts had simply not survived, the CRA maintained a smaller adjustment that reconstruction could not fully displace, and Tamar accepted it rather than pursue a formal appeal over a comparatively small remaining balance that further litigation was unlikely to erase entirely.

That is the honest shape of the result: a loss contained, not avoided. The original projection would have been a serious financial setback, the kind that reshapes a small operator's finances for years. The negotiated figure, by comparison, was manageable, spread over a payment arrangement Tamar could sustain without disrupting the business or taking on new debt. The penalty component, argued separately from the income question, was reduced as well, which mattered almost as much as the principal amount given how penalties and their associated interest compound over an extended payment period if left unchallenged.

Tamar has kept the new recordkeeping system in place since, updating it weekly rather than in the occasional catch-up session that had been the old habit. It was, in Tamar's own words relayed to us afterward, a lesson learned the expensive way - that a folder of the easiest thirty invoices is not the same thing as a defensible record, and that a sampling-based audit rewards whoever can produce the fuller picture fastest, especially once a file starts moving between reviewers with different expectations about what counts as proof.

What you can learn from this