Three choices that cost revenue first

Paying service credits our contract let us dispute

On a Saturday morning in November 2023, a software update distributed by a convenience-store client's payment processor left 11,400 card terminals at 2,180 of its stores unable to take cards. Our contract committed us to a four-hour on-site response but excluded faults caused by the client's other suppliers from service credits.

Our dispatch desk did not recognise a single cause for almost seven hours. It worked the calls one by one, sent technicians to 340 stores and told store managers that help was on its way. Our remote team then identified the common fault, arranged a rollback with the processor and restored most terminals remotely; the last sites recovered after 31 hours.

The client sought US$1.4 million in credits. Our general counsel and chief financial officer believed the exclusion applied in full, and on the wording they were probably right. I argued that the seven hours were ours: the misdirected visits had delayed the remote fix and given store managers false expectations. I proposed crediting US$620,000, our estimate of the cost attributable to that delay, and writing a mass-incident procedure into the contract. The chief executive agreed; the finance team recorded its dissent in the board paper.

We paid the credit, gave the client a written timeline of our errors and introduced a clustering rule that obliges the desk to consider a single incident when tickets concentrate on one device model. We trained 64 shift leads to act as incident leads.

The client renewed for three years in 2024 at about US$36 million a year, though its procurement team had several reasons and I would not attribute the renewal to the credit. In a 2025 outage at a different client, the desk declared a single incident after 22 minutes.

Next time I would telephone the client's head of store operations within the first hour, whatever the dispatch screen showed. Much of the client's anger came from hearing about the pattern from its own stores before it heard from us.

Committing to 940 kiosk installations when the client wanted 1,600

In early 2024 a quick-service restaurant group asked us to install self-order kiosks at 1,600 restaurants within 26 weeks, finishing before a September menu launch. Our trained crews could complete roughly 900 to 950 sites in that window without returning to the long overnight sequences that had damaged quality before.

Our chief revenue officer proposed hiring 220 installers through staffing agencies and extending crew hours. We had tried that on a terminal rollout in 2022: 14 per cent of sites needed a second visit within thirty days, and two agency installers were injured lifting equipment in the same month. I was not willing to repeat it at greater scale. The commercial team was right that declining part of the work would hand a competitor a relationship we had spent three years building.

I offered the client 940 restaurants on a committed schedule, with weekly reporting of completions and revisits, and walked its operations team through the capacity model behind the figure. The client accepted and gave the other 660 restaurants to another installer. We gave up about US$7.9 million of installation revenue and whatever maintenance income might have followed it.

Our crews finished the 940 sites four days before the launch, with 3.2 per cent needing a second visit. Scheduled hours stayed within our fifty-hour cap in every week but two, and both exceptions were approved by regional managers. In 2025 the client added 410 restaurants to our maintenance contract, worth about US$2.6 million a year. I cannot say how its experience with the other installer influenced that decision, and I have not asked for comparative figures we have no right to see.

What I would change is the timing of the capacity conversation. I would show the model at the bid stage, before the client had fixed its date. Its operations lead told me later that an earlier discussion might have moved the menu launch by a fortnight, which would have let us take on more of the work safely.

Keeping an acquired dispatch desk open after the deal model closed it

Brisendale acquired Sallowmont Technical Services of Spokane, Washington, in April 2022 for US$74 million. Sallowmont brought 310 technicians across Washington, Oregon, Idaho and Montana, around US$61 million of annual revenue and a 46-person dispatch and remote-help desk. The investment case assumed we would close that desk within six months, move its work to Indianapolis and Querétaro, and save US$2.6 million a year.

I had not taken part in the diligence. When I spent time in Spokane during the first month, I found that the desk held knowledge our systems did not: which mountain passes closed in winter, which rural stores could be reached only through a network of 120 subcontracted technicians, and which client managers expected a call from a named dispatcher. Querétaro's day shift also ended before the Pacific evening peak.

Amberlode's operating partner wanted the six-month plan kept, arguing that each month of delay would make the integration harder. I recommended keeping the desk for eighteen months, moving only scripted remote troubleshooting first, and documenting the regional routing knowledge before any role changed. The board accepted my plan with a quarterly review of the savings shortfall.

We realised US$0.9 million of savings in the first year rather than US$2.6 million. Sallowmont technician attrition that year was 12 per cent, against the 24 per cent our integration advisers had warned might follow a desk closure, and the region's two largest clients renewed. By 2024 the Spokane desk had become our western dispatch hub with 38 people, and Querétaro handled routine western remote work.

I would do two things differently. I would ask to join operational diligence before signing, because the six-month assumption could have been tested then at no cost. I would also have put the eighteen-month plan in writing to the Spokane team in the first week. I waited six weeks while the board considered it, and we lost four experienced dispatchers to rumour in that interval.