One vendor outage in Timmins exposes a contract nobody had stress-tested

The situation

The email arrived at 6:40 in the morning, an automated incident notice from their core software vendor stating that its platform was experiencing an outage affecting all customers, with no estimate for when service would be restored. Halima, who had spent years as an actuary before co-founding the company with Piotr, an architect by training, read it twice before waking Piotr to tell him their entire operation had just gone dark, unable at first to tell whether this was a brief blip or something far more serious.

Their Timmins-based firm built risk and structural modeling reports for engineering and construction clients across the north, work that combined Halima's statistical background with Piotr's grasp of how buildings and infrastructure actually behaved under stress over time. The company had grown steadily over several years to somewhere between five and twenty million dollars in annual revenue, and nearly every deliverable it produced ran through one specialized modeling platform they had licensed from a single vendor years earlier, back when it was the founders' first year in business and, at the time, it was the only tool on the market that did what they needed well enough to trust with client-facing work.

The plan, for most of the company's life, had simply been to keep using that platform and let the business grow around it rather than around any particular alternative. Switching to a different tool would have meant retraining staff, rebuilding years of model templates from scratch, and risking errors in the transition that could show up in a client's structural sign-off months later, so the dependency had never felt like a deliberate decision so much as a fact of how the business worked, the water it swam in rather than a choice anyone actively revisited.

The vendor relationship was old, familiar, and had never caused a real problem in years of steady use, which was part of why nobody had looked closely at the underlying contract in a long time. Their largest client by far was a mid-sized firm run by an architect named Ewa, who relied on Halima and Piotr's modeling reports to sign off on structural elements of her own projects on tight deadlines set by her own clients. When the platform went down, Ewa's firm had two active deadlines that depended directly on reports Halima and Piotr could not produce, and Ewa wanted to know, immediately and in writing, what was being done about it and who was ultimately responsible for the delay now hitting her own projects.

The gap nobody had noticed

The outage lasted just over three days before service was fully restored, which was long enough to blow through both of Ewa's deadlines and force her firm to seek extensions from its own clients, something she was not pleased to have to do because of a vendor problem entirely outside her control and, from where she sat, entirely invisible to her until it hit her desk. Her frustration was directed, understandably if not entirely fairly, at Halima and Piotr, since she had no direct relationship with the software vendor at all and no way to apply pressure of her own.

Once the immediate fire was out and service was restored, the contract with the vendor came under close review for the first time in years, and what it did not contain turned out to matter far more than what it did. There was no meaningful service-level commitment tied to any real consequence if it was missed, no obligation on the vendor to notify customers promptly or transparently during an outage rather than after the fact, no data escrow arrangement that would let the company access its own modeling files independently if the platform became unavailable for an extended stretch, and no exit assistance clause describing how data and templates would even be transferred out if the relationship ever ended. The company had, in effect, built its entire operation on a platform it had no real leverage over and no real fallback from, a dependency that had simply never been tested until now.

That gap became the actual subject of the dispute, once the initial anger over the outage itself settled into something more like negotiation. Ewa wanted assurance, in writing, that this would not happen again during one of her own deadlines, and she wanted some form of compensation for the disruption to her firm's own client relationships. Halima and Piotr wanted the same operational assurance for their own sake, but they also did not want to damage a vendor relationship they were still, practically speaking, stuck with for the foreseeable future, since no other platform on the market could be adopted quickly without real risk to every other client relationship the business depended on.

The three parties' interests only partly overlapped, which made the whole situation harder to resolve than a simple two-sided negotiation would have been. Ewa's priority was compensation and a guarantee that her deadlines would never again be exposed to this particular risk. The founders' priority was fixing the contract's real gaps without provoking the vendor into walking away from a mid-market client it might not consider worth the trouble of accommodating. The vendor, for its part, was willing to improve the contract somewhat to retain a customer of the company's size, but had no interest whatsoever in taking on open-ended liability for a client three steps removed from its own contract and unknown to it entirely.

What we did

1. Reviewed the full vendor agreement clause by clause against what had actually happened during the outage, identifying the specific missing protections, service-level commitments, data escrow, notification obligations and exit assistance, so the renegotiation could target real, demonstrable gaps rather than a general complaint about a bad few days, with each proposed change tied directly back to a moment in the outage where the existing contract had offered no protection at all.
2. Assessed the company's practical leverage honestly and without wishful thinking, confirming that switching vendors quickly was not realistic given the retraining and rebuilding of years of model templates it would require, which shaped the negotiation strategy toward improving the existing relationship rather than threatening to leave it, a threat the vendor would likely have seen through immediately given how few real alternatives existed.
3. Communicated separately and transparently with Ewa about what had gone wrong technically and what the company intended to fix contractually, distinguishing clearly between assurances the founders could actually deliver themselves and demands that depended entirely on a third party's cooperation, so her expectations were grounded from the start in what the company genuinely controlled, rather than in assurances that sounded reassuring in the moment but depended on a vendor she had no relationship with.
4. Drafted proposed contract amendments for the vendor renewal, adding a real service-level commitment with service credits for extended outages, a data escrow arrangement giving the company independent access to its own files during any future disruption, and a defined exit assistance period, prioritizing the protections that mattered most operationally over ones that would be hardest for the vendor to accept.
5. Negotiated the renewal directly with the vendor's account and legal teams, framing the requested changes as reasonable and proportionate given the size of the account and the demonstrated operational risk, while accepting early that some requests, including a meaningfully higher liability cap, were not going to move given the vendor's standard terms applied uniformly across its full customer base.
6. Offered Ewa a limited, documented commitment covering priority support status and an agreed scheduling buffer built into future deadlines tied to her projects specifically, rather than a broad cash compensation payment the founders were not in a realistic position to offer, given how little the vendor's own limited liability to them under the very contract in dispute actually left on the table to pass along.
7. Built an internal contingency plan alongside the contract changes, including periodic independent exports of active project files to a separate storage system, so the company would not be entirely dependent on the vendor's own infrastructure even with the new escrow clause formally in place, since a contractual right to data means little in the middle of an active outage if no working copy exists anywhere else.
8. Documented the whole arrangement in a short client-facing memo for Ewa and, eventually, other major clients, explaining in plain terms what protections were now in place and what the company's own contingency plan covered, which turned a defensive negotiation into something the founders could actually point to as evidence of a more resilient business going forward, rather than an episode they would rather clients simply forgot about.

The outcome

The renewed vendor contract included a genuine service-level commitment with credits for extended outages, a data escrow arrangement, and a defined exit assistance period, real improvements over what had existed before and protections the business had simply never had to rely on until this dispute forced the question. The vendor did not agree to raise its overall liability cap, and it declined to offer any compensation tied specifically to the earlier outage, treating the new terms as forward-looking rather than as a remedy for what had already happened, a position it held firmly throughout the negotiation.

Ewa accepted the scheduling buffer and priority support commitment rather than the direct cash compensation she had initially wanted, once it became clear that the founders had no meaningful leverage to extract cash from a vendor that had no contract with her at all and no exposure to her firm's own client relationships. The working relationship with her firm continued afterward, though it took several months of steady, reliable delivery before the earlier frustration fully faded from her side of the table.

Neither side of the overall negotiation walked away with everything it had wanted going in, which is roughly what a genuine compromise among three only partly aligned parties tends to look like. The founders absorbed the reality that they would remain dependent on a single vendor for the foreseeable future, now with meaningfully better protections but still without full operational control, and Ewa absorbed the reality that a vendor problem three steps removed from her own contract was never going to produce a cash settlement in her favour no matter how legitimate her frustration had been.

The compromise nonetheless left both relationships, with the vendor and with the client, intact and workable, which was the outcome the founders had prioritized from the very start of the process over any attempt to extract maximum concessions from either side. The company has since applied the same contract-review discipline to its two other significant vendor relationships, treating this outage as the reason it finally looked closely at agreements it had signed and never revisited.

What you can learn from this