A new Australian consultancy launched this week with a stark warning for retailers: most brands are practically invisible to AI assistants like Microsoft Copilot and ChatGPT. Founded by former Yahoo and retail media executives, Obsessd.ai says its audits of major Australian brands found that 60 percent of AI discovery queries returned no recommendation at all.
That figure means that for the majority of product searches a customer might throw at a chatbot—think “best wireless mouse for Windows” or “reliable Australian corporate gift supplier”—the assistant simply didn’t name a specific company. The launch signals that AI discoverability is becoming a boardroom issue, not just an SEO side project.
A New Consultancy Emerges, but the Stakes Are Highest on Copilot
Obsessd.ai has opened offices in Sydney and Melbourne, founded by Dan Richardson, previously an executive at Yahoo, and Bec Penn, who led product at retail media networks Cartology and Coles360. The consultancy brands itself as an “AI intelligence” firm, distancing itself from traditional digital marketing. Its services range from AI discoverability audits and agentic media strategy to frameworks for buying AI-based media.
Central to its pitch is the Obsessd Visibility Index (OVI), a proprietary metric designed to measure a brand’s presence, accuracy, and competitive position across five generative AI platforms: ChatGPT, Google Gemini, Anthropic’s Claude, Perplexity, and Microsoft Copilot. For Windows users, Copilot’s inclusion is critical. Unlike standalone chatbots, Copilot lives inside Windows, the Edge browser, Bing, and Microsoft 365. That gives it a permanent seat at the table when a consumer researches a product or an employee evaluates a vendor.
“When we sit down with local marketing teams and ask how they are performing in paid search or SEO, they have dashboards ready,” Richardson told retailbiz. “But ask what ChatGPT versus Gemini or Claude says about their brand and most have no idea beyond screenshots from ad-hoc prompts.”
The 60% Gap and Why It Hits Home on Windows
The consultancy’s headline claim—that 60% of AI discovery queries yield no brand recommendation—deserves scrutiny since the full methodology hasn’t been published. Yet it points to a tangible blind spot. Most organizations have mature dashboards for Google rankings, social engagement, and conversion funnels. But when a real customer asks Copilot “Which antivirus works best with Windows 11?” or “Where can I buy a sustainable laptop bag in Melbourne?” the assistant often returns generic advice or a blank slate.
For Windows users, this matters in two ways. First, Copilot’s recommendations can directly shape software and hardware purchases that integrate with the operating system. If a user asks for a backup utility and Copilot never mentions your tool because your documentation is weak, you lose a sale. Second, for IT decision-makers, Copilot in Microsoft 365 can influence enterprise procurement. An admin researching cloud migration partners or endpoint management vendors might receive an AI-curated shortlist without ever seeing a conventional search results page.
Penn, the co-founder, captures the disconnect: “The AI layer does not care how much you have spent on the brand. It surfaces what it can read, what it trusts, and what has been structured to answer the question being asked.” That shift—from bought attention to machine-readable trust—is the core of the discoverability challenge.
The Old SEO Playbook Won’t Work on AI Answers
For two decades, search engine optimization meant keywords, backlinks, and technical tweaks to climb the blue links. Generative AI overthrows that model. Instead of ten competing entries, you get one synthesized answer. The assistant has already compared products, applied filters, and summarized reviews before the user ever reaches a website.
Copilot accelerates this trend. Because it’s embedded in the OS, it can intercept research well before a browser opens. A Windows user might hit the Copilot key and ask, “What’s the best monitor for graphic design under $500?” The resulting answer may combine data from Bing Shopping, user reviews, and manufacturer spec sheets. If your product isn’t cited, the customer moves on—often without even knowing you exist.
Obsessd.ai’s OVI tries to capture this nuance. It doesn’t just count mentions; it weighs accuracy, recommendation prominence, and prompt coverage across unbranded queries. For a brand, appearing in a branded search (“tell me about Acme Corp”) provides little competitive insight; winning on a discovery prompt (“which project management tool has the best Gantt charts?”) is what drives new business.
The consultancy’s initial clients—personalised gift retailer Personalised Favours and Sydney fragrance brand By Yuliya—illustrate the point. Both operate in categories where a well-timed conversational recommendation can seal a purchase. Their early adoptions show that even small brands recognize the need to be legible to machines.
Your Action Plan: 5 Steps to AI Visibility
Whether you’re a solo developer selling a Windows utility or a large retailer with thousands of SKUs, you can improve your chances of being seen by Copilot and its peers. Here’s a practical sequence drawn from Obsessd.ai’s methodology and emerging best practices.
1. Run a diagnostic audit. Don’t rely on a few casual prompts. Build a controlled library of real customer questions—including location, price, use case, and comparison phrases—and test them on multiple AI platforms. Record not just whether your brand appears, but where it ranks, how accurately your products are described, and which sources the assistant cites. For Copilot, test both in-browser (Edge) and in-Windows scenarios if possible.
2. Fix your data foundation. AI models pull from your website, product feeds, and business listings. Ensure product names, descriptions, prices, availability, and policies are consistent across all pages. Use Schema.org structured data to help assistants understand product entities, reviews, FAQs, and organization details. For software developers, maintain clear version histories, compatibility notes, and support documentation in plain text—not buried in images or JavaScript.
3. Build independent authority. Copilot and others weigh trusted third-party sources heavily. Encourage genuine customer reviews on Google, Bing, Trustpilot, and industry forums (yes, sites like WindowsForum.com matter). Earn press coverage and expert citations. Avoid fake reviews or spammy content; AI models are increasingly adept at detecting manipulation, and aggressive tactics can backfire.
4. Tailor for Copilot specifically. Copilot draws on Bing’s index, Microsoft Shopping feeds, and Bing Places. Verify your Bing Places listing is claimed and accurate. If you sell physical products, ensure your feed is uploaded to the Microsoft Merchant Center. For enterprise IT brands, consider publishing case studies that mention specific Microsoft technologies and partner badges; Copilot in Microsoft 365 may surface those when responding to business inquiries.
5. Monitor, don’t guess. AI recommendations are probabilistic. The same prompt can yield different answers depending on model version, browsing history, and even time of day. Set up a monthly audit cadence to track whether your brand remains in key shortlists. Use screenshots for anecdotal evidence, but aim for structured tracking—perhaps a simple spreadsheet noting date, prompt, platform, result, and any factual errors spotted.
Improving AI discoverability isn’t a one-time project. It requires ongoing content hygiene, reputation management, and technical consistency—much like traditional SEO, but with the added layer of learning how each AI platform “thinks.”
Beyond Audits: Getting Ready for AI Agents
Obsessd.ai’s scope extends beyond visibility into “agentic media strategy” and “AI buying frameworks.” That language hints at a coming wave: software agents that don’t just recommend products but can actually purchase them on a user’s behalf. In a Windows enterprise environment, Copilot agents might soon be able to compare software license terms, check renewal dates, and even execute orders—all without a human touching a keyboard.
For brands, that means discoverability is only the first hurdle. You’ll also need robust APIs, real-time inventory feeds, machine-readable commercial policies, and transaction safeguards that allow agents to interact safely with your systems. Think of it as building a digital storefront that is both appealing to humans and interpretable by machines.
This convergence of retail media, AI recommendations, and agentic commerce is where the founders’ background in retail media networks (Cartology, Coles360) could prove pivotal. Richardson and Penn understand both the advertising side and the data plumbing that retailers need to operate in programmatic ecosystems. Their move into AI intelligence suggests they see generative platforms as the next advertising frontier—one where the line between organic endorsement and paid placement will soon blur.
What to Watch Next
Obsessd.ai’s launch is likely the first of many such consultancies targeting the Asia-Pacific market. The immediate challenge for buyers is methodological transparency: how are OVI scores calculated, and can audits be reproduced independently? The company’s long-term value will hinge on whether it can turn probabilistic model outputs into reliable, actionable intelligence.
For Windows users and IT pros, keep a close watch on Copilot’s evolving commercial capabilities. As Microsoft expands Shopping integrations, local answer cards, and enterprise agent actions, the assistant’s influence over purchasing will only grow. Today’s AI visibility audits may soon become as routine as checking Google Analytics—but with far higher stakes if you’re the brand that doesn’t appear.