Bastion Insights has launched Bastion Vantage, a digital intelligence tool that pairs AI analysis of online conversations with human interpretation.
It was built to address a problem that managing director Joanne Norton told B&T, has become endemic among marketers: too much data and not enough, well, insight.
“We’re focused on turning data into actionable outcomes and ensuring that instead of having a plethora of data at your fingertips, you have insights that unlock action for the business,” Norton told B&T.
In other words, marketers are “starving for meaning”.
Bastion Vantage can be purchased as a standalone product, separate from the rest of Bastion’s agency services. But Norton was direct about where she sees it delivering the most value: not replacing the agency’s existing qualitative and quantitative research, but sitting alongside it.
Used on its own, Norton said, Vantage is well suited to narrower and more specific questions that clients want answered quickly. The deeper payoff comes when it is combined with direct conversations with consumers.
“The benefit is when we not only understand that social discourse, but are then able to speak directly with consumers to understand other questions and then dig deeper to pull both elements together,” she said.
That distinction underpins Vantage. It is unlike social listening, according to Norton, which falls short of actually explaining anything.
“Social listening tells you what people are talking about. Bastion Vantage actually tells you what those conversations mean,” she said.
The tool is pitched more broadly rather than at a specific type of client. Bastion believes it has applications across brand strategy, campaign development and reputation management. It will operate at levels ranging from category-wide behaviour down to individual products.
AI-driven listening tools have long been criticised for missing nuance and raising privacy concerns, a weakness borne out In a 2025 Findings of ACL paper. Researchers introduced BESSTIE, a benchmark for sentiment and sarcasm classification across Australian, Indian and British English. It found that large language models performed better at sentiment than sarcasm. The models detected sarcasm in Australian English only 62 per cent of the time.
Norton’s position wasn’t that the technology has closed that gap, but that it isn’t meant to. The read still has to come from humans.
“The human interpretation is still absolutely paramount. We need to understand it from a human perspective in order to pick up nuances, differences and the things that are more leading edge versus mainstream,” she said.
Vantage pulls its information from publicly available sources including social media, search, news and blogs rather than anything gated or scraped from private data.
Asked what separates Bastion Vantage from an already crowded field of AI insights and listening platforms, Norton didn’t point to the technology itself but to a shift she sees coming in how insights get built at all.
Structured research including surveys and interviews, captures what people say when asked something directly. What she calls unstructured behavioural data included everything people post, react to and discuss without anyone prompting them. Norton argued the two haven’t been combined properly until now.
“It’s putting many different data sources into one synthesised location, from when insights can then be drawn on for human interpretation,” she said.
It’s a bet that the category’s next advantage won’t come from better AI, but from what the AI is being asked to sit alongside.
There’s a growing pressure for agencies to diversify their business models as the market contracts or concentrates, depending on which side of the fence you sit. Howatson+Company’s Plus Also Studios, for instance, recently launched a standalone SaaS product which seeks to automate campaign rollout.
