What to Cut
The stack audit found your voids. It also found what you're spending to defend things that aren't assets.
Every Canadian media company carries a mental list of its real assets: a broadcast license, a content library built to satisfy a quota, an audience number reported to a board every quarter, a format that’s always made money so nobody questions it. Almost none of that list survives one honest question- if an AI agent, a platform, or a subscriber had to choose you specifically, would any of it give them a reason to?
Apply that question and most of the list fails. What’s left isn’t complicated. It’s just short.
Everything that fails the question isn’t an asset. It’s a liability, and it almost always has a line item attached to it - money, headcount, executive time, spent every year defending something that stopped producing leverage a while ago. Some of that spend is legitimate: real infrastructure, real contracts, real renegotiation work. Some of it is money spent defending something that was never an asset in the first place, dressed up as strategy because it’s been on the P&L long enough to look load-bearing.
That second category is the subject this week. Not what to build. What to stop paying for.
The test.
An asset produces leverage without you defending it. A liability requires ongoing defense just to keep functioning at its current, already-diminished value. Apply that test to four things Canadian broadcasters spend money protecting every year, and three of them fail it immediately.
The license.
A CRTC license is a permission slip, not a moat. It costs money to keep - compliance headcount, regulatory counsel, the filings, the interventions, the hearings. That spend is defending your right to operate. It is not building anything an AI agent, a platform, or a subscriber has a reason to route through.
The tell is what the license actually does when a real threat shows up. It didn’t stop CBC from losing Hockey Night in Canada. It didn’t give a regional broadcaster leverage when a US streamer entered its market. It has never once been the thing that made an audience choose you. Every dollar spent on license defense above the legal minimum to keep operating is a dollar spent maintaining a document, not a business.
Cut it to the floor. Do the compliance you’re required to do. Stop treating the regulatory relationship as a growth strategy - it was never generating growth, it was generating paperwork that felt like progress.
The catalogue, defended for volume.
CanCon quota compliance produces content. It does not produce a relationship with the person watching it. Worse: as of this year, that catalogue is training data. Every hour of content produced to hit a percentage, rather than to build an audience that would notice if it disappeared, is now feeding a system that can approximate its style without paying for it again.
This is not an argument against making Canadian content. It’s an argument against making it to satisfy a quota instead of to build a relationship. The distinction shows up immediately in the numbers: content built to convert an audience into subscribers, list members, or repeat visitors has a job and a measurable return. Content built to hit a percentage has neither. It has a compliance officer who can confirm the box is checked.
If your production budget has a line for “quota content” that nobody can connect to a retention number, that line is a liability wearing a content library’s clothes. Redirect it toward the formats that build a direct relationship - even if that means producing less, at higher cost per hour, aimed at fewer people who actually come back.
Reach, chased without a list.
Programmatic spend, platform ad boosts, download-chasing on podcast platforms you don’t control - all of it optimizes for a number that looks good in a monthly report and produces nothing that survives a platform decision. Reach without relationship is not a smaller version of an owned audience. It’s a different thing entirely, and it disappears completely the moment the platform that generated it changes an algorithm, a policy, or a fee structure.
Canadian private radio spent a decade doing this at scale: buying reach, reporting reach, defending budgets by pointing at reach - while the actual list, the email addresses and phone numbers and SMS opt-ins that would have survived any of those platform decisions, went unbuilt. Ask most operators today whether that list exists and the honest answer is no. Reach was reported. A list was never built.
The fix isn’t cutting all platform spend to zero - some of it is legitimate top-of-funnel work. The fix is refusing to count reach as an asset on any internal document ever again, and redirecting whatever fraction of that budget was being defended as “audience growth” toward the mechanics that convert a viewer into someone on a list you own: a newsletter signup embedded in the stream, an SMS club tied to a live event, a membership tier attached to the content that’s actually irreplaceable.
Voice-tracked “local,” protected because it’s cheap.
This is the one that costs the most to say out loud, because it’s usually the profitable line. Voice-tracked and syndicated programming running under a local call sign is cheap to produce and margin-positive, and it has been protected for exactly that reason through a decade of cuts that took the local morning shows, the call-in formats, and the drive-time programming instead.
Here’s what that protection actually bought: the format AI replicates first, kept fully staffed and funded, while the formats AI cannot replicate - live, local, participatory, a human who knows the community and the community knows back - were the ones eliminated to protect that margin. The cost discipline was real. The asset selection was backwards.
Voice-tracked content generating real local revenue today can stay, priced honestly as what it is: syndicated content wearing a local callsign, not a local relationship. But it should stop being defended as “local programming” in any conversation with a board, because it is the single fastest thing on this list to become fully commoditized by the AI layer, and every dollar spent protecting its current margin is a dollar not spent on the live and participatory formats that are actually hard to replace.
What this frees up.
None of these four cuts require new capital. They require redirecting capital already being spent defending things that don’t defend you back. License compliance trimmed to the legal floor. Quota content redirected toward relationship-building formats. Reach spend stripped of its status as an asset and pointed at list conversion instead. Voice-tracked margin re-labeled honestly and no longer funded at the expense of live local programming.
Every dollar that comes out of those four lines is a dollar available for the one asset most Canadian operators are actually missing: an owned list - first-party audience data nobody can take away with a single platform decision. That’s not a coincidence. You don’t get to invent new money for the thing you’re actually missing. You get to stop spending the money you already have on things that only look like assets because they’ve been on the P&L long enough that nobody’s asked the question this year.
The question.
For every line item on your P&L that exists to defend something rather than build something, ask what it’s actually defending. A relationship, or a permission slip. An audience, or a report about an audience. If the honest answer is the second one in each pair, that line item is a liability the P&L has been calling an asset out of habit.
An asset defends itself by producing leverage nobody can take from you. A liability needs you to keep paying to defend it, indefinitely, at a return that keeps shrinking. Cut the second kind. Fund the first kind with what you free up.
Next week: a break from the operator playbook. The CRTC wrote three rules with real teeth this year — a streaming contribution, a wholesale rate, a fee ban - and enforced none of them. What zero-for-three says about who this regulator can actually compel, and who it can only compel to file paperwork.
Low frequency. High signal.