This article is based on written responses from Keyni founder and CEO Jared Navarre to questions from Greg Kihlström for The Agile Brand Guide’s Expert Mode series.
The AI layoff announcements read like weather. A front moves through, the numbers spike for a quarter or two, and then conditions normalize — which is why most of us have filed them under labor market, temporary, someone else’s department. Marketing leaders in particular have watched the coverage with a kind of detached sympathy. Our headcount didn’t move much. When budgets loosen, we’ll rehire.
Jared Navarre thinks that reading gets the story exactly backwards. Navarre is founder and CEO of Keyni, a consultancy that has advised more than 250 organizations, and he was named CEO of the medical technology company Onnix in February 2026. His argument is that the layoff is the part that gets reported and the least important thing happening. What matters is the requisition nobody reopens, and the fact that a lot of companies are quietly testing a new operating model in public without saying so.
The Layoff Is the Headline. The Vacancy Is the Story.
Navarre’s read is that the announcement is a lagging indicator of a decision already made somewhere less visible. He points to changes that don’t generate a press release — an org chart that thins out one box at a time, a team that absorbs another team’s scope during a reorg nobody called a reorg.
You see the redesign in quieter places…vacancies disappear, teams combine, job descriptions widen, managers become player-coaches, and senior people use AI to reach directly into work that once required analysts, coordinators, and several handoffs.
For a marketing organization, that’s a specific and recognizable set of events. The junior analyst who left in March never got backfilled, and the dashboards still get built. The coordinator role got folded into the demand gen manager’s job during the last planning cycle. None of that shows up as a layoff. All of it shows up in what a marketing leader can actually staff two years from now, because the entry-level rungs are where marketers historically learned attribution, media buying, and how a brief becomes a campaign. Navarre is blunt about what the numbers really measure: not what AI can do, but “what leaders believe the future cost of producing an outcome should be, and how much transition risk they’re willing to take now.” That’s a budget assumption dressed up as a technology story. If your CFO has internalized it, your 2027 headcount plan is already written whether you’ve seen it or not.
Role Compression Redraws the Job, Not the Task List
Most of the preparation marketers did was for task automation. The pitch was relief: the machine takes the tedious work, the human moves up a level. Navarre says that’s not the shape of what’s arriving. The unit of change isn’t the task, it’s the role, and the thing being eliminated is the handoff between roles.
Inside a marketing organization, one person can now research, draft, segment, analyze, report, and iterate with far fewer handoffs. The work hasn’t vanished…the handoffs have.
Anyone running a lean marketing operations function has already felt this, and it cuts both directions. Navarre allows that compression “can be incredibly powerful when the person gains leverage and owns the outcome.” Then he names the failure case, which is the one most companies are actually running: it “can be brutal when the same person inherits the responsibilities of several specialties without the time, authority, pay, or recognition that should come with them.” The practical question for a CMO isn’t whether to compress roles — that’s happening. It’s whether compensation, title, and decision authority moved along with the scope. If a manager now owns work that used to require three people and a weekly sync, and her comp band and approval limits are unchanged, the org didn’t get more efficient. It just moved cost from the P&L onto one person, and it will find out how much it moved when she leaves.
AI Is a More Comfortable Story Than Operational Failure
Navarre made this case publicly before we asked him about it. In a June 2025 piece for Built In, he argued that AI is taking the blame for layoffs that broken systems actually caused. Asked what the AI explanation conceals when a company would have made the cuts anyway, he was specific.
It papers over the decisions that made the organization expensive in the first place…poor hiring discipline, unclear roles, duplicated work, bad prioritization, too many approvals, and projects that survived long after their purpose did.
His explanation for why the substitution is so appealing is almost unkind: AI “sounds futuristic, inevitable, and impersonal,” where “operational failure sounds like somebody was responsible.” Worth saying plainly that Navarre sells the alternative diagnosis — Keyni’s business is fixing operating models, so an argument that layoffs are really an operational debt problem is an argument that generates consulting work. Readers can test it independently, though. Gartner surveyed 350 executives at companies with at least a billion in revenue and found that roughly 80% of those piloting AI or autonomous technology had cut headcount — and that the firms posting strong returns weren’t the same firms reporting the AI-related reductions, as Fortune reported in May. Cuts and returns came apart. That’s third-party evidence for the mechanism Navarre describes, and it’s the kind of thing a marketing leader can look for internally: if last year’s reduction was attributed to AI, ask what specific process the AI now performs and what the cycle time was before. If nobody can answer, the root cause sits somewhere else. His line for it is the sharpest thing in his answers: “You can remove people from a broken system and leave the system fully employed.”
Judgment Is Observable, If Anyone Is Watching
The premium skills everyone lists — judgment, adaptability, systems thinking — are famously unmeasurable, which is a convenient reason to keep rewarding output volume instead. Navarre rejects the premise. His position is that judgment shows up as behavior, and behavior can be watched.
The people you want ask better questions before moving faster. They identify assumptions, distinguish reversible decisions from irreversible ones, change their mind when the evidence changes, and think about who inherits the consequence downstream.
He adds a marker that translates directly to an AI-saturated marketing team: these are the people who know when a model’s output is useful, when it’s “merely fluent,” and when to throw it out. That’s the human-in-the-loop capability every AI content workflow depends on and almost no performance review measures. Navarre’s suggested instruments are unglamorous — scenario-based interviews, decision logs, honest postmortems, tracking avoided rework and risks surfaced early. The harder part is the reward. He wants “authority, autonomy, meaningful compensation, and an individual-contributor path that doesn’t require becoming a people manager,” which for most marketing departments means creating a senior IC track that doesn’t currently exist and moving decision rights down a level. On measurement he gives no ground at all: “Judgment often looks like the expensive mistake that never happened…if leadership still calls that hard to measure, the problem is the measurement.”
For the marketing leader doing headcount planning this quarter without a Fortune 50 budget, Navarre’s advice starts with two things to stop. Stop shopping for AI before you understand the work, and stop opening the exercise with the question of who could be removed. Then pick one value stream — campaign launch, lead routing, quarterly reporting, whichever one people complain about — and write down the outcome, every step, every handoff, the errors that keep recurring, and who owns the result when it breaks. Establish what time, cost, and quality look like today. Test AI against that baseline rather than against a vendor’s demo.
Which brings the whole thing back to the vacancy nobody reopens. The companies making that call well are the ones that mapped the work first and then decided what didn’t need a human in the middle of it. The ones making it badly are removing the humans and keeping the confusion, and they’ll pay for it in eighteen months when they try to rebuild a capability they never understood. Navarre’s filter is a good one to steal: ask where AI can remove latency without removing judgment. A smaller company doesn’t need a Fortune 50 budget for that, he says. It needs “the discipline to avoid automating confusion.”
