THE RATIO. One judgment, decoded twice: once for anyone, once for the profession.

Headnote

A Canadian data company licensed specialized pricing data to a global custodial bank, on terms barring the bank from sharing it with its own affiliates. The bank shared the data with dozens of affiliates for years anyway, and when the data company demanded records showing how widely it had spread, the bank refused and let those records disappear. The trial judge worked around the missing evidence with a rough damages formula that both sides rejected. The Supreme Court of Canada agreed the formula was wrong, set out a clear legal test for when destroyed evidence counts as spoliation, confirmed the bank had committed it, and sent the case back for damages to be worked out properly.

Factual Background

In 1999, the data company’s predecessor licensed securities pricing data to two related entities: what became Bank of New York Mellon (BNY) after a 2007 merger, and CIBC Mellon, the Canadian joint venture through which BNY served Canadian clients. Both licences barred the licensee from sharing the data with its own subsidiaries or affiliates, and pricing was volume-based per security, even though the data could benefit any number of the licensee’s customer accounts.

CIBC Mellon terminated its own licence in 2011, claiming it no longer needed the data, while secretly continuing to receive it from BNY for free. SS&C discovered this in 2016 through a data-delivery glitch, demanded a full accounting of every affiliate that had received the data and what BNY had earned redistributing it, and told BNY in writing to preserve all related records. BNY refused, offered no explanation, and did not preserve the records. SS&C terminated its own agreement and sued for breach of contract in 2017.

At trial, the judge found BNY breached the licence and faced a damages record gutted by BNY’s non-production. He drew two limited adverse inferences, that unaccounted-for data had reached unauthorized entities beyond CIBC Mellon and that this use was more than trivial, but stopped short of calling it spoliation. He then fixed damages using what he called a rateable approach: since BNY could account for only 44.6 percent of the data, it had to pay for the remaining 55.4 percent at the contract rate, yielding US$5.7 million against SS&C’s claimed US$889.75 million. The Court of Appeal went further, making an express finding of spoliation, but left the trial judge’s damages figure largely intact.

The Question for Determination

When a company destroys or withholds evidence during litigation, what must be proven for a court to call that spoliation, and once it is proven, does the law demand the harshest possible remedy or leave the judge room to choose a proportionate one? And separately, was the damages figure the trial judge actually reached properly connected to the evidence in this case?

Framed precisely: the Court had to settle the elements of the test for spoliation in Canadian civil litigation, decide whether a spoliation finding triggers a mandatory maximum inference or a discretionary, fact-specific remedy, and determine whether the trial judge’s rateable approach to damages, and the adverse inferences underlying it, were adequately tethered to the record.

The Court’s Reasoning

Spoliation is the intentional destruction, alteration, or concealment of evidence meant to subvert a court’s search for the truth. The Court had not addressed the doctrine since 1896’s St. Louis v. The Queen, and it set a four-part test: the evidence was intentionally destroyed or concealed; litigation was ongoing or reasonably contemplated at the time; the evidence was relevant to that litigation; and it is reasonable to infer it was destroyed to affect the litigation. Once all four are proven on a balance of probabilities, a presumption arises that the missing evidence would have hurt the spoliator’s case. The spoliator may try to rebut that presumption; if it cannot, the presumption becomes mandatory and an adverse inference must follow.

What that mandatory presumption actually requires is not fixed by a formula. The Court rejected SS&C’s argument that spoliation should automatically trigger the highest possible damages finding, holding instead that trial judges keep discretion to fashion whatever response, from an adverse inference to striking a pleading, costs, or contempt, fits the scope and impact of what was destroyed. This mirrors the Court’s approach to abuse of process generally and matches how other common law jurisdictions and Canadian provincial procedural rules treat the remedy.

Applying its own test, the Court agreed BNY had committed spoliation: the usage data was demonstrably destroyed or withheld after SS&C’s written preservation demand, litigation was plainly contemplated, the data was obviously relevant, and BNY’s only stated reason was that it disagreed with the merits of SS&C’s claim, which is never a valid basis for destroying evidence. But the trial judge’s two adverse inferences did little more than restate what the untainted evidence already showed and what BNY was already arguing. A proper inference needed concrete findings, such as treating each of the up to 65 potentially unauthorized entities as having accessed and fully used the data, absent an explained reason to find less.

The rateable approach to damages fared no better. It fixed a single rate for all unaccounted-for data based on an unsupported assumption about industry-wide centralized pricing, then applied that rate as if only one entity had used the data once, despite the finding that up to 65 affiliates may have had access for 17 years. When the Court of Appeal later found 65 rather than 44 entities had unlawfully received the data, the damages figure did not move at all, exposing its disconnect from the scale of the breach it was meant to compensate. The rate was also inconsistent with the trial judge’s own stated rationale, and it treated the data BNY could account for as though it had never been shared, when the evidence showed the opposite.

Because the award was untethered from the facts and from the incomplete inferences that were drawn, the Court allowed the appeal and remitted the matter to the Superior Court for a fresh damages hearing on the existing record, open to further evidence if the trial judge permits it. That judge must now draw adverse inferences producing concrete findings on how many entities used the data and how often, and price the result accordingly.

Critical Assessment

The Court’s refusal to impose an automatic maximum-penalty rule is well-grounded. Binding every spoliation finding to the harshest conceivable inference would treat ambiguous, low-stakes destruction the same as calculated, wholesale concealment, and the discretionary approach lines up with how Canadian courts already handle abuse of process generally.

That same discretion leaves real uncertainty for litigants trying to predict what spoliation will cost them. The Court says the inference must close the evidentiary gap rather than restate the obvious, but gives limited guidance on calibrating that fill when the true scope of what was destroyed is itself unknowable. The implicit benchmark, that absent an explained reason to find less, a judge should assume the worst reasonably supportable use, gives the doctrine real teeth, but it also means two judges facing similar spoliation could land on meaningfully different damages ranges, with strong deference owed to whichever choice is made.

The insistence that damages track the scale of the breach, illustrated by the trial judge’s award staying identical even after the pool of unauthorized users grew from 44 to 65 on appeal, is a sound corrective. It closes off the risk that a spoliator benefits twice: once by hiding the scope of its wrongdoing, and again if a rough-justice formula fails to move with whatever scope a court eventually finds.

Implications

If you’re not a lawyer: if you send a written demand asking the other side in a dispute to preserve records, keep that demand. If they destroy or withhold the records anyway, Canadian courts will now presume, once a few clear conditions are met, that whatever was hidden would have hurt their case, and judges have real tools to make that presumption count rather than a token penalty.

For the profession: the four-part test set out here, intentional destruction or concealment, contemplated or ongoing litigation, relevance, and an inference that destruction was meant to affect the litigation, is now the governing Canadian standard, and litigation holds and preservation demands should be built with it in mind. On damages, the ruling is equally useful outside spoliation cases: an award must be traceable to the scale of the breach actually found, and a methodology producing the same number regardless of how much wrongdoing is proven will not survive appellate review.

Case Details

  • Citation: SS&C Technologies Canada Corp. v. Bank of New York Mellon Corp., 2026 SCC 29
  • Court: Supreme Court of Canada
  • Heard: December 10, 2025 Decided: July 31, 2026
  • On appeal from: Court of Appeal for Ontario, 2024 ONCA 675, affirming in part and varying decisions of the Ontario Superior Court of Justice (Koehnen J.)
  • Counsel: Chris G. Paliare, Ren Bucholz, Glynnis Hawe and Catherine Dunne for the appellant; Eli Mogil, Thomas Curry, Brandon Kain and Brian Kolenda for the respondent
  • Key precedents applied: St. Louis v. The Queen (1896), 25 S.C.R. 649; McDougall v. Black & Decker Canada Inc., 2008 ABCA 353; Saskatchewan (Environment) v. Métis Nation, 2025 SCC 4; Naylor Group Inc. v. Ellis-Don Construction Ltd., 2001 SCC 58
  • Key paragraphs: [79]-[83] (spoliation test), [84]-[118] (remedial discretion), [124]-[134] (adverse inferences applied), [135]-[159] (rateable approach errors)