Chief Strategy Officer Canada

Four decisions, what each one cost and what I would change

Capital deployed in 86 acquisitions: C$1.21bn; As chief strategy officer, 2021–2025
Share of revenue from the United States: 36%; Sorrelgate, 2025; 19% in 2020
After-tax return of the 2021–2023 acquisition cohorts: 21%; Against an 18% hurdle; cohort review, 2026

Decisions & consequences

Selling the hospitality group despite a promise never to sell

By 2023 Sorrelgate's hospitality group owned nine businesses selling point-of-sale, reservation and kitchen software to independent restaurants in Canada and the northern United States. They earned C$118 million of revenue and employed 640 people. Their customers were moving to systems from providers that made most of their income on card processing and priced the software cheaply to win it. Organic revenue fell 3 per cent in 2022 and 5 per cent in 2023. Competing would have taken at least C$60 million for payment processing, hardware financing and round-the-clock support, none of which Sorrelgate did elsewhere.

Since 1999 every letter to a prospective seller had said that Sorrelgate never sold. Our founder and executive chair, Lucien Archambeau, believed one sale would cost us founders for years, and two group presidents agreed with him. I recommended a sale because the alternative was to keep 640 people in a business we would not fund while telling its customers we were committed to it.

I asked the board first to adopt written conditions under which we would ever sell, so the decision would rest on a rule rather than on the case in front of us. Lucien argued against the sale at two board meetings, and the minutes record his dissent. The board adopted the conditions in March 2024 and approved a sale process in May.

Of three buyers approached, a financial buyer offered C$231 million and planned to merge the businesses and close most offices. Pellara Commerce, a payments and restaurant technology company, offered C$212 million and committed to keep them operating. I recommended Pellara; one director argued for the higher bid. The board chose Pellara, giving up C$19 million, and the sale closed in October 2024.

Two years on, Pellara employs about 560 of the 640 transferred staff and keeps seven of nine offices. The proceeds funded eleven acquisitions in our Civic and Education groups during 2025. Four founders then in talks with us asked about the sale, and one withdrew.

We now send the written conditions with our first letter to any founder. What I would do differently is the order of telling. The nine general managers learnt of the decision the day after the vote, and two heard rumours first. I should have told each of them in person beforehand, conditions in hand, and accepted the risk of a leak.

Choosing small US acquisitions over a US platform

In May 2021 an adviser brought Sorrelgate a private-equity-owned US company selling student-information and school-transport software to about 1,400 school districts, at about C$940 million. US customers then supplied 19 per cent of our revenue; the purchase would have doubled that overnight and was almost five times larger than anything we had bought.

Our balance sheet and our model were at stake. Financing it would have taken net debt above three times earnings for at least two years, and the business ran on centralised management and sales quotas that our decentralised groups had never operated. Our chief financial officer, Nadia Pemberton-Roy, and the investment committee chair, Wendell Prychodko, favoured bidding for the scale, management bench and US investor story a platform offered.

I argued that our advantage was knowing a vertical's customers, not running a large, centrally managed company in an unfamiliar market. I proposed instead a programme of small US acquisitions in verticals our groups already served in Canada, a US deal team and an unchanged 18 per cent hurdle. That meant giving up speed, a scale story investors liked and a chance at a good business. The committee asked for a five-year comparison of both paths, which I prepared with the finance team, and our chief executive backed my proposal.

A Minneapolis deal team, opened late in 2021, grew to seven. Between 2021 and 2025 we bought 27 US businesses for C$612 million, mostly in municipal permitting, court records, school administration and dealer software, each folded into an existing group.

US revenue rose from C$167 million in 2020 to C$497 million, 36 per cent of the total, in 2025. The 2021 to 2023 US cohorts are tracking a 19 per cent after-tax return against the hurdle, and net debt has stayed below 1.5 times earnings. One purchase failed: a Wisconsin county-records business whose founder left early and whose product proved further behind than our diligence had found. We wrote it down by C$14 million in 2024.

Each group now reports its US exposure to the board separately. What I would change is leadership: for two years our US businesses reported to Canadian managers who did not know state procurement or US payroll practice. I should have insisted on US-based leaders in each group from the first purchase, as we did from 2023.

Making Canadian hosting mandatory in a decentralised company

In 2022, 41 of the 59 businesses then in our Civic and Education groups hosted public-sector customers' records themselves, in a patchwork of customer premises, small colocation contracts and, for eight of them, cloud regions outside Canada. Public bodies were changing what they asked for. Tenders increasingly required records to be held in Canada, an independent security assurance report and a list of subcontractors with access to the data, and a few provinces' public-sector privacy laws have at times restricted storing personal information abroad. In 2021 we lost two municipal tenders on those requirements alone, and insurers pressed us after a contained ransomware incident at one small business.

Our largest customer base was at stake, and so was our model: general managers run their own technology, and a central mandate would be our first. Several general managers opposed it as an intrusion, and our chief financial officer questioned a programme cost of C$38 million over three years.

I proposed shared hosting on the Canadian regions of two commercial cloud providers and one Canadian colocation operator. It would be mandatory for any business holding public-sector personal information, with an opt-out for any business that gave a written reason its group president and I accepted. We gave up part of the autonomy we had promised general managers and accepted a central cost that the smallest businesses would feel most.

A central team of 24 built the service, one shared security assurance report and standard contract terms for data location. Businesses migrated in waves from late 2022, each with a named central engineer.

By the end of 2025, 36 of the 41 had moved. Five opted out, three because customers ran the software on their own premises and two because their products were being replaced. Our win rate on Canadian tenders requiring in-country records rose from 31 to 52 per cent, hosting cost per customer fell 12 per cent after the second year, and questionnaires that used to take a business weeks are now answered from the shared report. Two general managers resigned over lost control.

The programme gave us our three tests for any central requirement. What I would do differently is the charging model. We allocated costs by customer count, so small businesses subsidised large ones in the first year. Charging by usage from the start would have cost far less goodwill.

Repricing data conversion after AI cut the hours

In 2024 implementation services brought in about C$118 million, or 21 per cent, of our Civic group's revenue, and converting customers' historical records took about 40 per cent of implementation hours. Six Civic businesses tried AI-assisted tools that propose how old fields map to new ones, flag inconsistent records and draft conversion rules for a specialist to check. Conversion effort dropped by between 38 and 52 per cent, and errors found at customer acceptance testing dropped too.

Because our contracts priced conversion on estimated hours, using the tools group-wide would remove about C$20 million a year of services revenue. The question was who kept the saving, and public-sector procurement officers would eventually see our answer.

Bram Oosterhuis, who succeeded me as president of the Civic group, proposed quoting conversions on historical hour estimates for two years, on the grounds that the value to the customer had not changed. Our chief financial officer was worried about a visible fall in services revenue. I proposed fixed, published prices by scope band, set by the number of source systems, record volumes and years of history, about a quarter below the previous average. Our chief executive read both memos side by side and piloted mine in fourteen businesses from January 2025, group-wide from July. We gave up about C$13 million of first-year services revenue and the comfort of telling investors AI had not touched us.

Customers saw the bands before signing. At Bram's insistence we kept an hourly rate card as an appendix, used only for changes in scope, because several public bodies' purchasing rules require one. We retrained 46 conversion specialists for testing and configuration work; nine left.

In 2025 Civic services revenue fell 11 per cent, but services gross margin rose from 24 to 37 per cent. The median implementation shortened from about ten and a half months to under seven, so maintenance and subscription revenue began sooner, worth about C$6 million in the year. We also beat more incumbents quoting hourly.

The other three groups adopted scope pricing during 2026. Looking back, I consulted people in the wrong sequence. I designed the bands with our sales leaders and only then showed them to customers' procurement officers. Starting with procurement officers would have put the rate-card appendix in the design from the outset, not after two contracts stalled.

Discuss a leadership challenge