A Cleaner's Border-Crossing Schedule Draws a Residency Fight

The situation

The letter from the Canada Revenue Agency did not ask a question. It stated one, flatly, as a conclusion: Farid had understated the number of days he worked in Canada during the prior tax year, and the amount of tax owing would be adjusted accordingly. Attached was a spreadsheet built from a calendar the agency's own reviewer had reconstructed, filled with dates that did not match anything Farid remembered writing down, along with a short paragraph noting that a response was required within a set number of weeks or the reassessment would simply stand.

Farid and his wife Karima owned a small rental property in Bracebridge, and Farid also worked for a commercial cleaning company whose contracts ran on both sides of the Canada-United States border. Some weeks he cleaned office floors in Ontario. Other weeks the same company sent him across for contracts near the border, and his pay stubs did not separate the two, since payroll simply totalled his hours regardless of which side of the line he had worked on. The arrangement was informal enough that nobody had thought to track it precisely at the time, and that gap was exactly what the reassessment was built on.

Karima worked as an early childhood educator in town, and their household income sat on the lower end, which made the amount in dispute, under fifteen thousand dollars, feel much larger to them than the number itself suggested. It represented weeks of Karima's own wages, and the couple had already been stretching the rental property's income to cover a roof repair that year. A coworker of Farid's, Lan, did similar cross-border shifts for the same employer, and Lan's own file had already gone through a CRA review the year before. That earlier case had shaped how the agency approached Farid's file, and not in a way that favoured him, since the reviewer assigned to Farid's file had also handled Lan's.

The file had started with a different lawyer, who had filed an initial response before withdrawing from the matter for reasons unrelated to its merits. By the time it reached our office, a partial record already existed, some of it useful, some of it built on assumptions we would need to unwind. The clock on the response deadline was already running, and Farid and Karima were unsure which parts of what had already been submitted they could still change, or whether raising new arguments this late would look inconsistent to the reviewer on the other end.

What the documents showed

The first task was to find out what evidence actually existed, separate from what the previous lawyer's letter had claimed. Farid did not keep a diary of his work locations, but he did have fragments: text messages arranging shift swaps, a company scheduling app with a login history, fuel receipts, and bank statements showing purchases made near the border on certain dates. None of it was a residency log. Together, it was closer to one than CRA's reconstructed spreadsheet had been, though it took several evenings of comparing dates against one another to see just how close.

Cross-referencing the scheduling app against the fuel and purchase records let us build a day-by-day account that covered roughly eight months of the disputed year with reasonable confidence. Where the app showed Farid clocked into a specific site and a fuel receipt from the same day placed him near that site's location, the two together made a reasonably strong case for where he had actually been. For the remaining weeks, the records were thinner, sometimes just a single data point with nothing to corroborate it, and we told Farid plainly that some of those days would likely stay contested no matter how the file was argued, because there was simply nothing left to point to.

The previous lawyer's initial response had conceded more days to CRA's version of events than the underlying evidence supported, apparently to move the file along quickly rather than to fight for every week. That concession was not binding, but reversing course on points already conceded in writing meant explaining, credibly, why the new position was better supported than the old one. We built that explanation around the scheduling app data, which had timestamps CRA's own reviewer could independently verify against the employer's records, giving the argument a foundation the reviewer did not simply have to take on Farid's word.

What became clear reading the file end to end was that CRA's reconstruction had leaned heavily on the pattern from Lan's earlier, already-settled case, applying a similar week-by-week split without separately confirming it against Farid's actual schedule. That was useful to know, because it meant the reassessment was partly a template, not a finding specific to Farid, and templates are easier to push back against than individualized findings once the pattern is named openly in a response letter. It also meant the negotiation had a natural anchor point: if CRA could not point to anything specific about Farid's own movements beyond the template, the burden shifted back toward showing why the template fit at all.

What we did

1. Requested the full audit file from CRA, including the reviewer's notes on how the disputed days had been calculated, because the original letter listed conclusions without showing the underlying reasoning, and we needed to see exactly which assumptions were doing the work before we could argue against them effectively. What came back confirmed a hunch: the reviewer's notes referenced Lan's earlier file directly, telling us the reassessment leaned on a template rather than a week-by-week review of Farid's own movements, and that distinction became the anchor for everything that followed.
2. Audited the previous lawyer's submission line by line against the evidence Farid actually had, flagging which concessions were supported and which had been made without a documentary basis, so we knew precisely where we had room to argue and where reopening a point would simply waste the negotiation's limited goodwill. That review turned up several weeks conceded in the first response with no supporting record behind them at all, concessions we would need to justify reopening carefully rather than simply reversing outright.
3. Rebuilt the work calendar from the scheduling app, fuel purchases, and bank records, cross-checking each entry against at least one other source before including it, which gave the reconstructed calendar more credibility than a memory-based account would have carried and let us stand behind every date we put forward. That cross-referencing eventually covered roughly eight of the twelve disputed months with two independent sources apiece, turning a folder of scattered receipts into a calendar the reviewer could check against the employer's own records rather than take on faith.
4. Identified the weeks with no reliable evidence and discussed them directly with Farid and Karima, so they understood from the outset which parts of the reassessment were realistically contestable and which were not, rather than promising a reversal we could not deliver and leaving them to discover the gap later. That conversation happened before we took any negotiating position with CRA, so neither of them was blindsided by a concession made on their behalf, and it let them plan around a likely shortfall ahead of the final number.
5. Drew the connection to Lan's earlier file, showing CRA in writing that the disputed weeks followed a pattern consistent with a template applied across similar cases rather than a finding tailored to Farid's actual movements, and asking the reviewer directly to confirm whether Farid's calendar had been independently checked. The reviewer never gave a direct answer to that question, which was itself telling, and the silence became part of our argument that the assessed days deserved fresh scrutiny rather than automatic deference to a pattern borrowed from someone else's file.
6. Negotiated week by week with the CRA reviewer, trading the well-documented months for concessions on the days where our evidence was strongest, rather than trying to win every week at once, which would have risked losing the negotiation's momentum entirely and hardening the reviewer's position on the weaker weeks too. Over several rounds of correspondence, that approach moved the reviewer off the template figure for the eight strong months entirely, leaving only the thinly evidenced weeks for a separate, more modest conversation instead of one all-or-nothing exchange.
7. Confirmed the final split in writing, including how it would apply to the rental property income already reported, to make sure the settlement did not leave a gap that could trigger a second reassessment down the line once the current file closed. Farid and Karima's rental income had never been part of the dispute, but leaving that boundary vague in the settlement letter could have invited a future reviewer to reopen the question, so we made the separation explicit rather than assumed.
8. Reviewed the settlement with Farid and Karima line by line before it was finalized, so they understood exactly which weeks remained assessed against them and why, rather than receiving a final number with no explanation attached to it. That walkthrough also gave them a chance to flag anything that still looked wrong before signing off, and it left them able to explain the outcome to their own accountant afterward instead of just handing over a number neither of them fully understood.

The outcome

CRA agreed to remove close to half of the disputed days from the reassessment, based on the reconstructed calendar for the eight well-documented months. The remaining weeks, where the evidence was thin from the start, stayed on CRA's books largely as originally assessed. Farid and Karima ended up owing additional tax on those weeks, though the total was meaningfully lower than the original letter had claimed, and the payment amount was small enough to absorb without touching the rental property's reserve fund.

The settlement did not erase the underlying problem, which was that the cross-border arrangement had never been tracked properly while it was happening. We were honest with Farid and Karima that a fully documented case, built as the work happened rather than reconstructed a year later, would likely have held up better across the whole period, not just the eight strong months, and that the outcome here reflected the limits of reconstructed evidence as much as the strength of the argument itself.

The switch from the previous lawyer had cost the file some ground too. Points conceded in the first response could be revisited, but not without spending time and credibility that could otherwise have gone toward strengthening the weeks with genuine evidence, and it is likely a small number of the contested weeks would have gone differently had the file never changed hands.

Since the file closed, Farid has started keeping a simple daily log of which country he worked in, a habit that costs him almost nothing but would have changed the shape of this dispute considerably had it existed from the start. Karima's rental income reporting was confirmed clean throughout, which meant the property itself was never at risk, only the characterization of Farid's employment income, and the two of them now review the log together at the end of each month as a matter of routine.

What you can learn from this