I woke up with a question the other morning. Why is everyone talking about Claude? I'd made the switch myself recently, and I wanted to understand what had actually happened. Was it the Super Bowl stuff? The Pentagon drama? Could any of that really drive purchase intent?
So I went to the data. I pulled ten months of Australian media monitoring from Medianet and cross-referenced it with Google Trends. What I found surprised me and I think it should matter to everyone working in communications, PR and marketing.
For most of 2025, ChatGPT dominated Australian media coverage. Around 2,500 online news articles a month. Claude was sitting at roughly 700. Not even close.
Then February 2026 happened.
Claude's monthly coverage jumped to 2,236 articles — nearly matching ChatGPT's 2,820. By March and April, Claude was consistently generating over 80% of ChatGPT's coverage volume. In Google Trends Australia, Claude's relative search interest climbed from a near-zero baseline to 28, a 14x increase, while ChatGPT held flat at 85.
What events could have driven this? Anthropic's Super Bowl ad and the Pentagon contract standoff.
This validated a few things.
According to Appfigures and BNP Paribas, Anthropic's Super Bowl ad mocking the inclusion of ads on ChatGPT conversations triggered a 32% spike in app downloads and an 11% jump in daily active users in the days after the game — versus just 2.7% for ChatGPT and 1.4% for Gemini. Claude climbed from number 41 to number 7 on the App Store within days.
But the paid ad was only the beggining. The story spread through earned media for weeks. And when the standoff with the Pentagon hit two weeks later, Claude made it to #1 on the App Store. The paid placement planted the seed. The media coverage made the narrative bloom.
Sam Altman called the Anthropic ads "deceptive" and "clearly dishonest" in a public post on X. That response became its own major news story, extending the campaign's reach far beyond what Anthropic had paid for.
If you ever do any of Medianet’s Editorial sessions, you will learn that conflict is one of the seven news values.
The media LOVES conflict. It loves a David vs Goliath narrative. And it especially loves it when Goliath attacks the underdog.
Every time OpenAI defended itself, it fed the beast and another headline was written that favoured Anthropic.
The Medianet data and Google Trends tell the same story in parallel. Claude's Australian media mentions went from flat to surging in February, alongside the Super Bowl and the Pentagon crisis. Google Trends shows Claude's search interest making the exact same move, with a sustained upward trend from February onwards.
Correlation is not causation. But it is very hard to ignore that earned media coverage about both events directly aligns in timing and in shape with the spike in consumer search behaviour.
This is perhaps the most important lesson. ChatGPT generated roughly 2.3x more Australian media mentions than Claude over the full period. But what is Claude's coverage actually telling us?
Sentiment (albeit at the article level and not entity level sentiment which would be ore accurate) for Claude was 24.6% positive versus 17.2% for ChatGPT. Negative coverage sat at 17.2% for Claude versus 23.6% for ChatGPT. That is a meaningful gap in both directions — more positive, less negative, across thousands of articles.
More striking is who appears alongside each brand. In Claude's coverage, the Pentagon is a co-mentioned organisation in 16.7% of stories. ChatGPT's co-mention map is dominated by Microsoft, Nvidia and Meta, which indicates a more commercial and competitive narrative. Claude's is an ethics, governance and values story. And those stories align with the brand identity and values Anthropic needed to differentiate itself from their giant competitor.
Sentiment of mentions of Claude and/or Anthropic in Australian online media and other organisations mentioned. Source: Medianet
I didn't switch because of a single ad or a single headline.
I switched because over several months, I was exposed to an accumulating set of stories and narratives about each company. Beyond features and the latest model release, the stories were about what they stood for, what they refused to do and how they behaved under pressure.
My network was obviously exposed to the same stories. I started noticing more of my LinkedIn connections talking about Claude as a company, not a tool. I got curious. I explored. Eventually, I paid for an annual subscription.
So the journey started with stories and ended with a commercial transaction. That is brand building working exactly as it should: flying under the radar and kicking long-term objectives that end up in open wallets.
I didn't switch because of a single ad or a single headline.
I switched because over several months, I was exposed to an accumulating set of stories and narratives about each company.
As marketers, we often talk about the long and the short of it. The long-term goals vs the short-term results. The short is the Super Bowl ad — measurable, immediate, planned and calculated. The long is the ten months of narrative accumulation. You need both. But most brands only invest in one.
If you work in communications or PR, this is your argument for the boardroom. Earned media is not a vanity metric, but it can be a purchase funnel. It's not immediate enough sometimes to fit in the quarterly report. But that doesn’t mean it doesn’t work; in fact, if done well, it makes a significant difference that would be nearly impossible to accomplish with marketing dollars.
And if you're a marketer, it’s time to learn from comms as we move to build brands rather than sales campaigns. To focus on the message, longer-term objectives and value. Brands that carry meaning and open hearts. Because the narratives around your brand exist whether you invest in them or not.
Media monitoring data sourced from Medianet, covering Australian online news, social media, blogs and press releases from July 2025 to April 2026. Search interest data from Google Trends Australia, past 12 months. All figures cited relate to the Australian market unless otherwise stated. And of course, Claude helped with part of the data analysis, which was verified by a human.