The rise of Large Language Models (LLMs) like ChatGPT, Claude Sonnet, and Gemini significantly advanced text generation, making them pretty good at drafting all kinds of text outputs.
But when it comes to research tasks like building reliable, targeted media lists, these tools fall short due to outdated training data and online research limitations, leading to inaccuracies and hallucinations.
We put the latest version of ChatGPT to the test for PR research tasks, comparing its media list-building abilities to Dazzle’s. To kick things up a notch, we also included Perplexity, an AI created specifically for online research.
From broad prompts like “Find journalists covering AI startups in the US” to niche topics like “generative AI music tools,” the results speak for themselves:
As you can see, Dazzle is leagues ahead of general-purpose LLMs for PR-specific research. It’s important to remember that Dazzle is not a media database and doesn’t use any training data. It researches information in real time to deliver the most accurate and timely insights.
Let’s have a closer look at the results.
Journalists found
General-purpose LLMs are surprisingly lazy when it comes to media research. They return between 3-10 journalists, depending on how specific your query is. That’s nowhere near enough, especially for a broad story angle that requires a more extensive list of journalists.
With its advanced research workflow, Dazzle consistently delivers 40-100 journalists per query (when expanded using Chat recommendations). While other AI tools scrape a handful of sources, Dazzle searches hundreds, ensuring a more comprehensive and reliable list every time.
Journalist relevance
To judge journalist relevance, we teamed up with PR experts from global PR firm SlicedBrand. They ran several queries and evaluated the results. The verdict? Only about half of the journalists recommended by ChatGPT and Perplexity were truly relevant. Common issues included:
- Wrong beat – Journalists didn’t really cover the requested topic.
- Not a journalist – External contributors or opinion writers were misidentified.
- Not a media outlet – Authors of company blog posts were mistakenly included.
Here’s an example where Perplexity fell short: it provided three results, none of which were journalists. Yan and Bernard are occasional contributors, while Andreessen is a founder writing on his VC firm’s blog.
Dazzle doesn’t just deliver more journalists—it delivers better ones. It provides 10 times more journalists per query than general AI models, with nearly double the relevance. PR experts rated over 90% of Dazzle’s results as relevant, compared to just 50% from general AI tools.
What’s the secret? Dazzle is purpose-built for PR workflows. It scans hundreds of articles, filtering journalists by their likelihood to cover your story based on audience, location, outlet and other criteria. This works just as effectively for national, regional or industry-specific outlets as it does for podcasts, blogs, newsletters, and guest posts.
Data accuracy
Building media lists isn’t just about finding the right journalists—you also need accurate, up-to-date details like their current beats, roles, and contact info. ChatGPT struggled here, with 20-30% of its results being inaccurate. The biggest issue? Outdated journalist occupations.
Here’s an example:
A solid list, you might say. But a quick check revealed that three of the six results had outdated information:
- Jennifer left MIT Technology Review in August 2023.
- Matthew left Insider in March 2023.
- Alex has been inactive since May 2024.
What’s behind the old info? ChatGPT's limited search capabilities. Instead of pulling info directly from news outlets, it often takes shortcuts, relying on third-party articles like “top journalists writing about AI.” Here’s a list of sources for the query above:
These articles are often outdated and poorly maintained. Worse, when no list is found, the AI usually outputs something based on years-old training data.
Dazzle takes a smarter approach, delivering over 95% accuracy by prioritizing reliable sources like outlet websites and journalist profiles. To provide enough results, it searches hundreds of sources—an intensive process that actually mirrors how PR pros build lists manually.
The difference? Dazzle automates this tedious task, compiling comprehensive media lists in seconds, saving days of effort.
Contact details
When tested against real media lists from our PR experts, ChatGPT and Perplexity fell short, finding accurate email addresses for less than 10% of journalists.
Dazzle, however, nailed it, delivering correct emails for 80% of journalists. How? Dazzle taps into a network of trusted data partners that we regularly verify against real media lists from our PR partners, ensuring the highest level of accuracy.
Ease of use
Dazzle excels beyond these measurable metrics, transforming your PR workflow with unique, user-friendly features:
Save journalists with one click
Add journalists and their key details to your media list instantly. No copy-pasting required.
Efficient media list management
Collaborate with your team, track outreach progress, share updates, and stay organized with ease.
Smart recommendations
Dazzle learns from your selections, automatically suggesting new journalists that perfectly complement your media list.
Easy info verification
Every detail comes with source links for quick verification. Plus, Dazzle generates article summaries—even for paywalled content—to save you time.
Conclusion
AI is already transforming PR, and PR professionals must embrace the right AI tools to streamline their workflows and stay ahead of the curve. While LLMs like ChatGPT and Perplexity have made great strides, they fall short in essential PR tasks like real-time research and accurate media list building.
That’s where Dazzle shines. Designed by PR pros for PR pros, it combines powerful AI with a vast network of reliable data sources to deliver more accurate, relevant and up-to-date journalist information—all in seconds. Our tests proved it: Dazzle consistently outperformed general-purpose AI models, making PR workflows faster, smarter, and more effective.
