The WebStream
Media blog.

Four days after the launch we wrote up, a news.com.au piece by the journalist Mary Madigan catches the second failure. The Friday livestream of Kyle Sandilands Live did not go the four-hour distance. Producers pulled the plug after roughly two hours. The headline on the piece is "A flop."

What is on the record is worse than the headline. Executive producer Jaimee "Mayo" Blazquez and newsreader Brooklyn Ross recorded an apology video for the show's app after the failure. Ross opened with the line that should not have needed to be said: "We were scared this day might come." Blazquez followed with the explanation subscribers had already guessed. "We're so sorry that the tech has failed us today because the show that we had planned would have been incredible. We've tried a few times, and I think we're just going to upset you more if we keep coming back and then disappearing again. So we've decided today to not take the show to air." The plan is to use the weekend to "properly get everything sorted" and return Monday with "new equipment because we've ordered everything."

Sandilands's own contribution to the postmortem is the most technically revealing line of the day. Speaking in a video shared exclusively with news.com.au by Madigan, the host walked through what he had been told three days earlier. "Our mainframe computer... I was told three days ago, 'This is starting to feel the pressure', and I just assumed they'd buy a new one. But they didn't. So now someone is down buying a new one. Apparently, back on Monday. Fingers crossed." That is a production manager hearing a server warning, a host assuming the warning would be acted on, and nobody escalating the warning to a purchase order for seventy-two hours. By the time the order was placed, the Friday show was already off air.

The subscriber experience during the failure is the part the technical column would actually want to dwell on. Fans who had paid the $99 annual fee were in the live chat, which Madigan reports as the only function on the site that appeared to be working without a hitch. The livestream itself was plagued by no audio, frozen screens, bad camera angles, and no images altogether. "This is worse then Monday," one subscriber wrote. Another: "My god this morning is a flop." When the show was pulled, a "back to catalogue" button appeared guiding desktop listeners back to the landing page. The link to listen in to the Friday show had disappeared. The end-of-stream card read "Live stream finished," and that was the subscriber's parting view of a product they had paid for that morning.

The other revealing line is the one the host will most regret. Asked by subscribers about refunds, Sandilands cut in with, "No refunds. It's not Harvey Norman here, guys." The show's own FAQ does outline a refund policy, with a thirty-day window from the transaction date and a clause that services or products already used or accessed may not be eligible for a full refund. A show that was pulled two hours into a four-hour paid livestream is, on the plainest reading, a service that was not fully delivered. The host has now told the subscriber base, on the record, that the policy will not be applied generously. The retention lever is cut off, the chat is the only thing still working, and the people watching the chat are the same people being told no.

The internal signal in the piece is the friend-of-the-show text. John Ibrahim, the former Kings Cross operator and a known Sandilands contact, texted the host during the failure to ask what was happening in the studio. Sandilands read the text out on air. "I just got a text from John Ibrahim saying something's wrong with the sound." When the people texting you to ask what is wrong with your broadcast are not in the building, the broadcast has failed in a way the building cannot talk its way out of. The host's earlier line, delivered while the audio was still cutting, is a different tell. "If you've just joined us on livestream, we had someone trip over a cord or something weird happened." That is the line a host reaches for when the on-air explanation has not been written yet. It is the line a producer whose only job was talkback would have pre-empted.

The host stayed in the chair, defended the team publicly, and told listeners he "feels responsible for everything." The defence is fair. The team is small, the timeline was sixty days, and the equipment that was on the warning list three days before the show is the equipment that failed. None of that is the team's fault in the moment. It is the fault of the procurement decision that did not get made, and the decision not to delay the Friday show while the new equipment was being sourced. A broadcast operator reading the same timeline would have done two things differently. The procurement would have happened the moment the warning was filed, and the Friday show would have gone out as audio-only, pre-recorded, with a banner across the player saying the video chain was offline for maintenance. Neither option is exciting. Both options keep the subscription product live.

This is the second column in a week on the same launch. The 10 August piece walked through the Monday failure and argued the audio chain was the actual story. The Friday failure is the column writing itself. The producer who knew this was coming, the mainframe that was on warning for three days, the host who stayed in the chair while the equipment collapsed around the show, the apology video that had to be recorded by the team that was on the warning list, the subscribers who watched the only working function of the product argue with the host who was telling them there would be no refunds. The series has been tracking personalities outgrowing their hosts, and the structural trap of building distribution without the production layer to match. This is the production layer catching up to the distribution, in real time, on the record, in the chat of the people who paid.

Monday's return is the next test. New equipment, the same team, the same chat, the same $99 fee, and now a subscriber base that has watched one launch and one cancellation inside a single working week. The technical column's expectation, for the record, is that Monday goes out clean or does not go out at all. The third failure in a row is the failure the platform does not recover from. The show's own producer, on the record, said the team was scared this day would come. They did not need to be scared. They needed a procurement meeting.

---

Adjunct: what a recurring engagement is actually worth, and why we only ship tested solutions.

WebStream Media does not run the kind of operation that ends up in the previous column. That is not bravado. It is a structural difference, and it is worth spelling out because the difference is the entire reason clients come back.

Our work is grounded in reliability. A livestream that buffers or drops mid-event undermines everything the production set out to do, so we design every signal chain with redundancy in mind, and we run a documented rehearsal before anything goes to air. The list is short and it is non-negotiable. The camera chain is built the night before. The audio path is checked end to end with a real source and a real destination. The streaming encoder is fed a test pattern for at least ten minutes. The backup encoder is hot and switching. The talkback channel between the director, the camera operators, the audio op, and the talent is open and tested with every voice on it. The fallback, if anything in the primary chain falls over, is rehearsed, not improvised. If any of that fails the test, the show does not go live. It goes out audio only, or it gets pushed, or the equipment gets swapped. The audience never finds out which option we picked, because the audience only ever sees the version of the show that holds.

That discipline is the entire reason clients come back. Recurring engagements are not won on price and they are not won on personality. They are won on the slow accumulation of events that went out clean, where the client never had to think about the technical layer during the show, and where the post-event report was a list of green ticks. A production company that has not yet built that record cannot fake it on a single high-stakes launch, and the record cannot be bought in a sixty-day procurement window. It is built across years of small events, every one of which was treated as if it were the broadcast that had to hold.

The credentials that pay for that record are unglamorous. Our senior engineer has production credits across Channel 7, the Queensland Government, the Federal Court of Australia, and Queensland's major performing arts venues. Those rooms do not give second chances. The camera goes on, the microphone goes on, the program goes to air, and the audience, whether a courtroom, a press gallery, a public theatre, or a national broadcast, sees a chain that holds. The same standards we brought to those rooms are the standards we now apply to a corporate briefing for thirty executives or a hybrid conference for three thousand remote attendees. We scale the production to match, and we do not compromise on the technical quality to do it.

This is what we mean when we say we deploy only proven and extensively tested solutions. The word proven does not mean new. It means it has been used in the field, in a room that mattered, in front of an audience that would have noticed if it failed. The word extensively tested does not mean a quick check on the morning of the show. It means the full chain has been through a documented rehearsal, the failure modes have been walked, the backup paths have been hot-swapped, and the operators have worked together long enough to know what the other person's voice sounds like on talkback. Anything less than that is, in our trade, an experiment, and we do not experiment on paying audiences.

The Kyle Sandilands Live column is not a piece about a single bad launch. It is a piece about what happens when a subscription product is built around a brand and a personality before the technical layer is built around an audience. The chat worked. The microphones did not. The login did not. The mainframe was on warning for three days and the warning was not acted on. None of those failures are mysteries. Each of them has a rehearsal that would have caught it. The cost of running the rehearsal would have been a fraction of the cost of refunding the subscribers who asked, who were told no, and who will not be back on Monday.

Recurring engagements are the only honest proof a production company can offer. A flashy showreel can be cut from any operator's best half hour. A list of gear on a website can be assembled in a week. The only thing that cannot be faked is the calendar of clients who came back the next year, and the year after that, because the last event went out clean and the one before that went out clean and the one before that went out clean. That calendar is what we trade on, and it is the reason a brief from us will not contain a single piece of equipment that has not been used in the field, a single workflow that has not been rehearsed end to end, or a single client commitment that we are not confident the chain will hold.

If you are planning a launch, a conference, a hybrid broadcast, or a recurring series of events and you want the technical layer to be the part you stop thinking about, that is the work we do. The brief starts with a conversation about what the audience has to see and hear, and it ends with a broadcast that goes out the way the rehearsal said it would. Everything between those two points is where the testing lives, and it is the part we are not willing to skip.

Three pieces from today's coverage walk through the first day of Kyle Sandilands Live, the subscription-only podcast that replaced the KIIS FM breakfast show after Sandilands's February exit. An ABC first-episode review calls the opening "clunky." A news.com.au four-hour piece is headlined chaotic. A Brisbane Times account by Kayla Olaya adds the cleanest technical-layer detail of the three: at about 7am, producer Bruno Bouchet had to reset the entire show. "Everything is going to go black here," he announced on air. "We are going to lose all audio, all lights. Screens are going to shut down." That is the kind of failure mode that should not happen on day one of a paid product.

ABC catches the opening. Five seconds of dead air. Pre-recorded audio ran hot. Sandilands pushed back that he could not hear anything in his own cans. The Brisbane Times reporter walks the next three hours and finds the same problem repeating. The producer, who had skipped sleep over a week, told listeners plainly: "it all went to shit." Celebrity psychic John Edward joined via Zoom with technical issues and presenters talking over him. Dave Hughes was dialled in the last ten minutes as a substitute for Anthony Albanese, who was meant to call and did not. His voice came in crackly. "Why do you not understand how to use the microphone?" the host fumed at his team.

Sandilands's defence, quoted in the Brisbane Times, is worth dwelling on. "It is a brand new, never before done anywhere in the world live app radio program. We put it together in 60 days, and the team have worked tirelessly behind the scenes." Sixty days is not a technical timeline for a daily live audio product. It is a marketing timeline. The team were not given the rehearsal window a broadcast operator would have demanded, and the host is now on the record defending that call.

Listeners noticed in real time and flooded the show's Instagram. "Bruno's mic connection to Kyle is cooked, massive echoes." "Trouble with the sound system Kyle." "Audio levels are terrible." News.com.au adds a different complaint category: several paid subscribers could not log in to watch at all. "Can't sign in to the app. Link to sign in isn't not being sent to me." The launch of a paid subscription product could not deliver a sign-in flow on day one. That is a separate engineering failure, but it lands in the same column.

The financial structure is the other half of the story. Sandilands launched the show after settling his wrongful termination case against ARN for $12.09 million, well under the $85 million he originally sought. The settlement included a provision under which ARN takes 19.9 per cent of revenue from the show for three years, in exchange for $1.5 million in advertising. The subscription itself costs $99.99 per year, upfront only, no monthly option. Sandilands told the Brisbane Times the show runs no advertisements "because of the female militant activist group, the Mad F---ing Witches," who he said spooked advertisers. KIIS FM's counter-move on day one was blunt: every caller who made it to air on the old station got $1000. The first song picked for the new era is Josh Fawaz's AI-assisted remix of Like A Prayer.

What this column is actually interested in is the audio chain. Six months on from the radio exit, a multi-million-dollar operation went to air without a working mic on the host, without a tested phone-in chain, without a working subscription login, without off-camera staging under control, and with a producer on air announcing a full system reset to subscribers who had just paid $99.99. The failure mode is not the personality, not the politics, not the platform. It is the technical chain.

This is the pattern we have been tracking in this series. The 23 June piece argued networks build personalities they cannot control. The 26 June piece walked through Stefanovic's exit. The 30 June piece closed the loop on the incentive structure. Today's three pieces extend the series sideways rather than forward, into the part nobody wants to talk about. The talent exits. The capital stacks up. The platform of choice is announced. And then someone has to actually run a show, on the day, on the air, with a mic that works, a phone chain that carries a voice, and a sign-in flow that lets subscribers in.

For a Brisbane production company that runs the technical chain for a living, this launch is the default. Independents who walk away from networks and into their own studios inherit the entire production layer at once: signal, scheduling, switching, captions, uploads, ad insertion, redundancy, monitoring. Networks used to absorb all of that behind a wall the talent never had to see. The wall is gone. The ones who can afford to hire a team that has, often, in our observation, have not built that team in time. Half up and running is the most accurate line Sandilands has said in years.

The postmortem we would write is straightforward. Episode one is a stress test, not a launch. Rehearse on a private stream the night before. Have a producer whose only job is the talkback channel. Test the phone-in chain with two real callers. Test the sign-in flow the day before. Start from cold on a different day if the chain is not clean. The first hundred paying listeners are not forgiving. They are taking notes, screenshots, and posting on Instagram before the first ad break runs.

The series will keep tracking this. Every high-profile independent launch we can find is now a case study in how the technical layer actually gets built, and how often it does not. The names will change. The audio chain does not.

A 30 June Guardian opinion piece by Ed Coper puts a name on the thing we kept circling this week. It is not the personality, not the politics, not the podcast. It is the algorithm. TV Karl was governed by TV incentives; podcast Karl is governed by social media incentives. One rewards broad appeal within socially acceptable bounds. The other rewards contrarian, outrageous, attention-grabbing content, because that is what the engagement model was built to surface.

The piece is the cleanest articulation we have seen of the structural trap Stefanovic walked into. On Today, the editorial layer filtered what got to air. Advertisers paid for safe reach. Regulators watched. The host had to remain, at minimum, plausible inside a newsbrand. On a podcast distributed across YouTube, Spotify, Apple and Instagram, none of those filters apply. The distribution layers are the same platforms that have, over a decade, been tuned to map onto our emotional wiring. Coper's line is the right one: "when you migrate our entire news media ecosystem on to platforms that map on to our emotional wiring, you are inadvertently building Karl's podcast."

The case studies Coper lines up are the same ones Nine staff raised internally. Megyn Kelly, who took her Fox News audience and tripled it on YouTube after leaving the network. Candace Owens, who now regularly pulls a larger YouTube audience than Fox, CNN and MSNBC combined. Joe Rogan, whose 2024 Spotify deal was reportedly worth around $US250 million, and who platforms fringe ideas under the cover of long-form, unscripted, just-asking-questions neutrality. These are not aberrations. They are the template. Stefanovic followed the template.

This is where the two previous pieces in this series connect. The 23 June post argued networks build personalities they eventually cannot control. The 26 June post walked through the Stefanovic departure as the case study landing. Today's Guardian piece closes the loop by naming the incentive structure that made the case study inevitable. The network did not lose Karl Stefanovic to the far right. The network lost Karl Stefanovic to a distribution layer that pays better for outrage than for journalism, and his audience came with him.

There is a useful corrective in the piece for the broadcast industry. Coper's prescription is twofold: support the outlets doing quality news, and regulate the platforms so the incentives reward consensus, facts and balance rather than anger. We would not go that far. The platforms are not going to be re-engineered for consensus any time soon, and the editorial layer that once bounded a network host is gone regardless of what regulators do. The honest version is the harder one. Audiences are sovereign now, and sovereign audiences reward whatever the algorithm surfaces, and the algorithm surfaces whatever holds attention, and attention is held by things that make us angry, frightened, or morally certain.

For a Brisbane production company sitting in the middle of this, the practical reading is straightforward. The businesses that will thrive are the ones that pick their incentive layer deliberately. If you build for a network's editorial standards, you will inherit a network's editorial ceiling. If you build for an algorithm, you will inherit whatever the algorithm is being optimised for this quarter, which is usually outrage. The path that holds is the slow one: build a craft, build an audience, pick the distribution that does not require you to become someone else to keep the lights on. Coper would call that restoring quality information systems. From the broadcast desk, we would just call it making work you can stand behind in five years' time.

Two days after we wrote the piece on personalities outgrowing their hosts, the case study caught up. Mediaweek confirms what the 23 June coverage anticipated: Karl Stefanovic has left Nine immediately, with the network and his representatives mutually agreeing the dual role of hosting Today and running his independent podcast was no longer tenable. The departure was originally scheduled for the end of 2026. It has been brought forward.

Two internal emails to staff, both seen by Mediaweek, frame the announcement. CEO Matt Stanton acknowledged the volume of coverage and framed the call as right for both parties. “After more than 20 years it’s the right time for Karl to move on from Today and from Nine,” he wrote. Director of News and Current Affairs Fiona Dear acknowledged the strain on the Today team and signalled more announcements on the show’s future are imminent.

The 48 hours before the announcement were the unstable part. Wednesday brought crisis meetings, the Australian breaking the story, Nine denying it on the record, and Pauline Hanson posting in Stefanovic’s defence on X and accusing the network of trying to sack her “good friend.” By Thursday morning the podcast interview with Tommy Robinson that triggered the cycle had been pulled from every platform it had been posted to, with nobody on the record about who asked for the takedown. The most likely explanation, as we noted, is that Stefanovic’s own advertisers blinked first.

The ripple reached ARN, where Stefanovic co-hosts The Long Weekend with Eddie McGuire. He stood aside from this week’s show to give all parties time to review the situation. ARN moved quickly to separate the network from the controversy, telling Mediaweek that Stefanovic’s external media activities are “undertaken in a personal capacity and are entirely separate from the network.” The phrasing is the same distancing language Nine used at the start of the week, which suggests the lawyers have agreed on a template.

The pattern we flagged on 23 June holds. The personalities networks build eventually become the thing networks cannot control. Joe Rogan and Megyn Kelly are the comparison cases Nine staff reportedly raised when they started calling Stefanovic “Karl Bogan.” Nine has now lost the gamble, ARN has spent a week publicly disclaiming its own co-host, and Stefanovic is, by Friday morning, an independent operator with a podcast, a radio audience that will follow him or not, and a set of backers who are louder than the network that just let him go.

The dream is seductive. Shoot video, upload it, watch the algorithm bless you, retire on ad revenue. For every creator who makes it, thousands pour years into the same gamble and walk away with a few hundred dollars and a hard drive full of dead content.

The math is brutal. YouTube's partner program needs a thousand subscribers and four thousand watch hours before a cent lands. CPM rates swing wildly by niche and season, and a brand-safety scare can spook advertisers off your topic overnight. A video that takes a week to make might earn eleven dollars. One that takes an afternoon might earn eleven thousand. There is no predicting which.

The algorithm is a rumour. Creators who study it speak in conflicting certainties, while attention fractures across TikTok, Shorts, Reels, podcasts, and Twitch, so a hit video earns less than it would have five years ago.

The trap is treating YouTube as a career plan rather than a craft. Those who make full-time money from it usually built elsewhere: a production background, niche expertise, a community they already owned. Walking in cold with monetization as the only goal is the surest way to burn out before the money arrives.

Then there is the income that never appears in the YouTube dashboard.

Affiliates are the first surprise. A creator with fifty thousand subscribers might earn more from a single Amazon link in a description than from a month of ads. Camera gear, software, books, even dog food; commissions stack quietly behind the videos, and disclosure rules are the only thing keeping the practice honest.

Training courses are where the real money often hides. The successful creator stops selling hours and starts selling knowledge: how to edit, how to grow, how to monetise. A two-thousand-dollar cohort, run twice a year, can dwarf ad revenue without needing a single new viewer.

Consultancy sits on top of that. Brands pay for strategy calls. New creators pay for channel audits. The YouTube channel becomes a reel, not the product.

The uncomfortable part is how these streams invert the model. The audience funds the dream; the side hustles fund the business. Anyone planning to live off YouTube alone is mistaking the stage for the whole theatre.

The first piece, a Brisbane Times opinion column, frames a paradox every broadcaster eventually meets. Networks pour years and millions into nurturing a personality, only to watch that personality become the thing they cannot control. The column uses two Australian names: Kyle Sandilands and Karl Stefanovic.

Sandilands is the cautionary tale ARN has now lived through. He was paid for years to push the limits of morning radio, until advertisers walked under pressure from activist campaigns and his falling-out with co-host Jackie O gave the network its exit. ARN paid a reported twelve million dollars to settle, rid itself of the contract, and watched advertisers quietly return. Sandilands is now independent, with his own online show and a partnership with Pauline Hanson.

Stefanovic was the unfolding case. The second Brisbane Times piece, published the same evening, confirms Nine is negotiating his exit. The trigger was a podcast interview with British far-right activist Tommy Robinson, posted to YouTube, Spotify, Apple Podcasts, and Instagram, then quietly pulled from all four within twelve hours. Nine denies any involvement. Nobody is owning the takedown. The most likely explanation is that Stefanovic's own advertisers asked for the episode to disappear, with the activist group Mad F---ing Witches preparing to relaunch its #KancelKarl campaign.

The pattern is structural, not personal. Free-to-air television rewards charm and broad appeal. Subscription content, podcasts, social platforms, reward conviction, conflict, and an audience that actively chooses to follow. Joe Rogan and Megyn Kelly are the modern business models Stefanovic appeared to be reaching for, with Nine staff reportedly calling him "Karl Bogan" since the podcast launched in January.

For broadcasters, the lesson is uncomfortable. Build a personality, and you build a liability. The audience those networks spent decades cultivating now follows the person, not the channel. Nine can distance itself from Stefanovic in public, but his politics, his podcast, and his backers (including Hanson, who has reposted the deleted interview on her own channel) will keep speaking regardless.

A 25 June Mediaweek piece reports that Shameless Media has signed its first licensing deal and launched a dedicated Shameless Media Studios arm to handle future partnerships. The deal poaches KICPod and KICBump from LiSTNR. Both shows, hosted by Laura Henshaw and Steph Claire Smith, will move to Shameless Media in late September.

The story is bigger than one podcast swap. Shameless Media now commands ten per cent of Australia's total podcast advertising revenue, has logged 155 million listens, sits on 1.8 million social followers, has been profitable since launch, and has never taken external investment. The founders, Zara McDonald and Michelle Andrews, built that without a radio network behind them, and they are about to take share from the biggest one in the country.

The shift underneath the headline is structural. Under the new arrangement, KICPod and KICBump will be available as full-length videos on both YouTube and Spotify. The piece flags this as a deliberate departure from the closed proprietary infrastructure of radio-backed networks. In plain terms: the big networks built walls around their distribution. Independent networks are now building bridges to where audiences already are, which is on video platforms, in feeds, and in subscription apps.

The licensing arm matters too. Shameless Media Studios exists to onboard shows like KICPod into the network without the founders having to produce them. They get revenue share, brand alignment, and audience overlap, while the original hosts keep ownership of their own product. It is the model every independent network eventually reaches for, once the original slate is full and the back catalogue is paying its own way.

There is also a deliberate parenting push. A Shameless Media original parenting podcast launches in October, the month after KICPod arrives. By year end the network will run five always-on original shows and two licensed ones, with a chief commercial officer framing the move as unlocking new revenue in health, wellness, and parenting categories. Three of the five original slots will land in the same vertical, by design.

For broadcasters watching from the outside, the lesson is uncomfortable. The independent network built without a parent company is now licensing content from shows that used to belong to Australia's largest podcast network. The future of audio is not the biggest distribution. It is the smartest stack of original shows, licensed shows, and platforms that meet the audience where they already live.

The growth industry angle is the part worth dwelling on. The piece notes Shameless Media has been profitable from launch and never raised external capital. That is rare in media, where most operators chase scale first and profit later, if ever. The licensing model extends that discipline. Instead of producing every show, the network takes a cut on shows already producing themselves. Revenue grows without headcount, without studio expansion, without the producer-of-the-month overhead. Each new licence adds margin, not cost.

For Australian creators, the path that opens up here is more interesting than the corporate ladder ever was. Build an audience. Build a brand. Stay independent. Then let a network like Shameless Media licence your show the way a record label once licenced an album, except this time the artist keeps the masters and the network gets a slice. That is a working template for anyone treating audio as a growth industry rather than a content side hustle.