Anonymised, illustrative composite. A share purchase agreement for a specialty foods distributor carried a broad arbitration clause. When the buyer discovered a fabricated customer contract eighteen months after closing and sued for fraud, the arbitration clause reached the claim anyway.
At a glance
A buyer had acquired a specialty foods distributor by share purchase for $4,200,000. The agreement's indemnification article capped the seller's liability for breaches of representations and warranties at 15% of purchase price — standard enough — but carved out fraud, stating that fraud claims remained available “at law, without limitation.” Buyer's counsel had negotiated that carve-out specifically and was proud of it.
The same agreement, in its general provisions near the back, contained a standard dispute-resolution clause: “any dispute arising out of or in connection with this Agreement shall be finally resolved by arbitration.” Nobody on the buyer's side had connected the two clauses at signing.
Eighteen months after closing, the buyer's finance team discovered that one of the target's largest customer contracts — the one most heavily cited in diligence to support the revenue multiple — had been fabricated. The counterparty on the contract had never actually placed the orders it recorded; the paperwork had been created to inflate trailing revenue ahead of the sale.
The buyer sued in the Superior Court of Justice for fraudulent misrepresentation, seeking rescission and damages, confident that the fraud carve-out from the indemnity cap meant this claim was not subject to the deal's other limitations. The seller moved to stay the court action in favour of arbitration.
Purchase price $4,200,000. The fabricated contract represented $380,000 of annual revenue, roughly 9% of the top line the deal multiple had been built on. Damages claimed: $1,700,000, calculated as the price premium attributable to the fabricated revenue at the multiple both sides had used to value the business.
Arbitration Act, 1991 s.7(1) leaves nothing to the court's discretion once a valid arbitration agreement and a proceeding covered by it are established: “the court in which the proceeding is commenced shall, on the motion of another party to the arbitration agreement, stay the proceeding.” Shall — not may. The court does not weigh whether arbitration is a good idea on these particular facts; it stays the action if the statutory conditions are met.
s.7(2) then lists the only grounds on which a court may refuse the stay, and it is a closed list: incapacity at the time the arbitration agreement was entered into; the arbitration agreement being invalid; the subject-matter not being capable of arbitration under Ontario law; undue delay in bringing the motion; and the matter being proper for default or summary judgment. Fraud in the underlying transaction is not on that list, and the buyer was not arguing that the arbitration agreement itself was invalid or fraudulently induced — only that the seller had committed fraud in the deal the agreement was part of. Ontario law does not treat that distinction as taking a claim outside an otherwise-valid arbitration clause; a claim being about fraud is not the same as a claim that the arbitration clause itself is invalid.
The buyer's stronger instinct — that the SPA had already carved fraud out of its limitations — ran into a drafting reality: the indemnity cap and the arbitration clause are two different provisions doing two different jobs. One limits how much can be recovered and for what; the other decides where any dispute gets resolved. Carving fraud out of the first says nothing about the second unless the agreement says so, and this one did not.
The court granted the stay. The fraud claim proceeded to arbitration under the clause's own procedure — a single arbitrator, private hearings, and (because the clause itself stated the arbitrator's decision was “final and binding, with no right of appeal”) no route back to the courts even on a question of law. s.45(1) of the Act would ordinarily allow a leave-based appeal on a question of law where the arbitration agreement is silent on appeals — but only where the agreement is silent, and this one was not.
The buyer still won on the merits — the fabrication was not seriously disputable once the arbitrator saw the customer's own records — and recovered damages. What it lost was the forum: a private arbitration rather than a public court judgment, with no appeal available regardless of how the arbitrator reasoned the decision.
Had the buyer litigated instead of being stayed, it would have had the benefit of the court's ordinary discovery and production process against the seller to establish whether the fabrication pattern extended beyond the one contract, the possibility of a public, precedent-setting judgment, and (had the arbitration clause not excluded appeals) a broader review than the arbitrator's private, largely unappealable decision. None of that was available once the stay was granted — the clause both parties had treated as boilerplate turned out to control where an eight-figure fraud claim would actually be heard.
The tell was in the negotiation history, not the final document: buyer's counsel had fought hard, and successfully, to carve fraud out of the indemnity cap, and treated that as job done on the fraud issue. Nobody asked the parallel question about the arbitration clause a few pages later. A carve-out negotiated into one clause does not travel to another unless someone deliberately puts it there.
A 30-minute call is enough to tell you whether AI pays for itself here.
Canadian small-business transaction data is not published anywhere, so most valuations in this country quote an American benchmark. The Deavo–Treadstone Acquisition Index is a daily record of Canadian listings built to replace that: asking-price distributions by province and city are published now, and days on market, departure rates and asking-to-sale spreads follow as the series lengthens. Leave an email and we will tell you as each measure lands.
No pitch, no listings. One email when the first report lands.
