When to follow the pack — and when to break away

There is a well-worn theory in economics that explains why otherwise intelligent organisations keep making identical decisions — even when the evidence is thin. It is called an informational cascade, and according to Rafael Gomez, a professor of employment relations and director of the Centre for Industrial Relations and Human Resources at the University of Toronto, it is playing out in Canadian workplaces right now.

"When it's hard to discern what the right answer is, it's actually not a bad idea to follow someone you think is credible," Gomez, pictured above right, told HRD. "You're piggybacking on someone else's presumed investment in that decision."

The logic is sound in theory. If a major bank or the federal government mandates that employees return to the office five days a week, smaller employers have reason to follow. They are outsourcing the research cost of a hard decision to a source they trust.

But Gomez is quick to flag where the model collapses: "Where informational cascades go wrong is when the first mover who may be credible — like government, like a big bank — didn't actually do the research and just sort of moved on a whim, or moved because of other pressures."

Everyone copies the leader. The leader was wrong. And a great many organisations end up in the same wrong place together.

The RTO wave: a cascade in action

Return-to-office mandates are perhaps the clearest example of this dynamic. As a handful of high-profile institutions began publicly announcing return-to-office requirements in 2023 and 2024, the pressure on other employers intensified almost overnight. RBC, Scotiabank, BMO, and TD all announced four-day-per-week in-office requirements within months of each other in mid-to-late 2025 — mirroring moves by Amazon and JPMorgan in the US. Ontario's Treasury Board president acknowledged that the province "studied other governments and businesses" before issuing its own full-time mandate. That is the cascade, stated plainly.

Christine Vigna, Chief People Officer at Dejero, pictured above left, watched it happen from the inside. "Some organisations have moved really quickly because other large employers were doing it," she said. "They're just following that trend rather than really thinking about what it means for their organisation."

READ MORE:  1 in 5 employees admit to using AI in job interviews

Gomez adds a less obvious layer: before the pandemic, many large firms in the knowledge sector were actively encouraging staff to work flexibly, redesigning offices to cut the cost of downtown real estate. Workers resisted. Then the pandemic made remote work the norm — on workers' terms. "When it's workers demanding it on their terms, that's when employers push back," Gomez said. "They want to dictate the terms under which that ability to work from home happens."

The push back to the office, in other words, is as much about power as productivity. For Vigna, the more urgent concern is the cultural damage from mandates issued without sufficient internal reflection.

"There's a severe underestimation of the cultural cost of how those are being implemented," she said. "There's a real loss of trust across employees, inconsistency in terms of how those mandates are being implemented, and a lack of clarity around the why employees are being asked to return."

The data bears this out. A University of Pittsburgh analysis tracking more than three million tech and finance workers across S&P 500 firms found abnormally high turnover of around 14% following RTO mandate implementation — with the exodus disproportionately hitting senior, skilled, and female employees. Statistics Canada's own time-use research reinforces the point: teleworkers are 12–14 percentage points more likely to report being satisfied with their work-life balance than their in-office peers. The organisations at greatest risk, Vigna argues, are those that followed the pack without aligning RTO policy to their own workforce model, talent strategy, and business goals.

AI adoption: the next cascade

If return-to-office is the cascade that already crested, artificial intelligence may be the one still building. And the pattern, according to both Vigna and Shawn Gibson, Chief Human Resources Officer at Info-Tech Research Group, is disturbingly familiar.

"Many AI adoptions are failing," Vigna said. "That's because many organisations hadn't actually defined the business problem they were trying to solve with AI. They just wanted to say, 'Hey, we use AI.'"

Five major industry reports published in early 2026 reached the same conclusion: the failure isn't the technology — it's the absence of governance and implementation frameworks. Only 25% of companies have fully operational AI governance programmes, even as 82% report moderate to extensive AI use.

Gibson saw this pressure first-hand. "Some organisations are a little fearful. And some just get caught up in the novelty or competitive pressure of being on the cutting edge."

Info-Tech experienced its own version of the challenge when exploring an AI-powered HR assistant — a compelling idea for a global workforce spanning six countries and multiple time zones. Phase one, loading the assistant with internal policies, moved ahead smoothly. Phase two stopped them cold.

"The second phase was whether an AI chatbot could answer confidential information," Gibson explained. "What gave us pause was recognising that privacy guardrails, governance, and overall risk needed closer attention."

Rather than racing ahead because the market was moving, Info-Tech slowed down. "It's often because the pace of adoption gets ahead of the governance around it," Gibson said.

Dejero took the opposite approach from the outset. In November 2024, the company rolled out a 24-month strategic plan for AI implementation built around three lenses: business outcomes, board and shareholder accountability, and — most critically for Vigna — employees.

READ MORE: Return-to-office mandates: is there a retention battle brewing?

"There's a real responsibility for all employers right now to help upskill their employees," she said. "I hope that the work we are doing — increasing everyone's AI literacy and overall skills — is that when they decide it's time to seek out a new opportunity in the market, their AI skills make them an ideal candidate."

Crucially, Dejero approached AI as workforce transformation rather than an IT initiative, investing early in education, encouraging experimentation, and normalising failure. "The organisations that are seeing the most success with AI right now are usually the ones that have normalised learning early instead of waiting for perfect certainty," Vigna said.

Gibson's advice to peers is similarly direct: "Fail fast, figure out what works. Just start. Even if it's a failure, you can document your experience and train back.

Knowing when to break away

So if cascades can carry both good and bad information, how does an HR leader know when to follow and when to break rank?

Vigna offers a clear decision rule: "It's when the external trends stop matching the internal evidence."

The signals she watches for are concrete: declining employee engagement despite following a popular practice; productivity that fails to improve after technology adoption; retention risk among key talent; and leadership struggling to operationalise a policy consistently. "When you see your leadership team struggling to operationalise a policy consistently," she said, "that's a great signal that you haven't thought through the realities of where your business is at. Are you structured for this? Is your organisation mature enough to manage the change?"

The cascade: % of Canadians working mostly from home, 2016–2025

Four consecutive annual declines as major employers announced return-to-office mandates

Sources: Statistics Canada Labour Force Survey (May 2025); Statistics Canada time-use study (2024); 2727 Coworking RTO Canada 2026 analysis; Angus Reid Institute (July 2025)

Gibson frames it as a discipline question. "Is this truly solving the problem, or are we getting caught up in the hype?" he said. "The right move is not always the most visible one. It's the one that actually works for your people, your managers, and your business."

Gomez offers the longest view. The real competitive advantage, he argues, belongs to the organisation willing to do its own work. "I think you could do better if you were the employer that said, 'I'm not going to follow the leader. I'm going to do my own research. I'm going to figure out what's best for my company,'" he said. "That would give you the ability to hire the best talent, keep and retain the best talent, and perform better than your competitors — not follow your competitors."

The AI interview paradox

In a final irony, the cascade logic has now reached the hiring process itself — and it is creating a genuine ethical knot.

Organisations are increasingly using AI to screen résumés, evaluate candidates, and score interview responses. At the same time, some candidates are using AI tools to generate real-time scripted answers, creating what Gibson calls "AI talking to AI" — a dynamic that bypasses human judgement on both sides.

"What we're finding is that sometimes we get a sense that candidates are using AI right in the actual interview with the recruiter," Gibson said. "That's not acceptable — it's not evaluating the candidate properly."

Vigna pushes back on the employer reaction. "It's a little hypocritical, because so many employers are now using AI in the hiring process themselves," she said. "AI is used to screen résumés, to evaluate candidate responses and score them. It's a tad hypocritical that employers can use AI for all of their processes, and yet we're saying that employees should not."

At Dejero, they have reconsidered their assessment approach entirely. Rather than penalising AI use, they probe it. "We don't shy away from candidates using AI, but we dig into those post-assessment conversations: 'What was your prompt? How did you use it? How could you have improved your output?'" Vigna said. It is, she argues, a more honest way to assess both AI competency and underlying knowledge at once.

Whether employers follow that model — or follow the pack on AI detection tools — may itself be the next cascade to watch.