It's a common observation, and a genuine frustration: AI-generated answers often seem to lean on Wikipedia, major press outlets and established reference sites more than a business's own website, even when that business's own content on a topic is thorough, accurate, and genuinely well researched. There's real substance behind this observation, and understanding why changes what a realistic response actually looks like.
Why these sources genuinely do carry disproportionate weight
Wikipedia, major press outlets and established reference sites, in a way worth acknowledging honestly rather than resenting, have years or decades of accumulated trust signals, extensive editorial review processes, and a demonstrated track record of accuracy that a typical business website, however good, simply hasn't had time to build regardless of content quality. This isn't a flaw in how these models work, it's a reasonable reflection of which sources have earned trust over a longer track record than most individual businesses have existed.
This doesn't mean a business's own content is worthless
For specific, detailed questions about a business's own products, services or specific expertise, a model still needs to draw on that business's own content, since no third-party source will have that specific level of detail. The Wikipedia and press advantage is strongest for general, background or definitional questions, not for specific, business-particular ones a genuinely interested buyer would ask.
The realistic strategic response, not chasing an unrealistic goal
Trying to outrank Wikipedia for a general definitional question is rarely worth the effort for most businesses. A more productive focus is winning the specific, detailed, business-particular questions where a business's own content is genuinely the best and only source available, while treating broader background credibility as something that benefits from third-party mentions over time rather than something a business's own website can fully control.
Working with these sources rather than only against them
If genuinely eligible, a well-sourced, neutral Wikipedia presence, and consistent, accurate press mentions over time, contribute directly to the third-party validation layer discussed in a broader GEO framework, rather than being a competing, adversarial source to somehow outrank. Earning a mention in a source these models already trust is often a more efficient path to visibility than trying to independently build the same level of trust from a standing start.
What this looks like in practice
Build content depth around the specific questions only a business itself can answer authoritatively, pursue genuine press and industry mentions as a trust-building layer rather than treating them as separate from GEO work entirely, and accept that broad, general topic questions will likely continue favouring established reference sources for the foreseeable future regardless of how much content any individual business publishes. This realistic framing is part of what makes GEO work sustainable rather than an endless, ultimately unwinnable fight against sources with a genuine, earned trust advantage.
The bigger picture worth keeping in mind
This dynamic isn't unique to AI search, it mirrors how traditional Google results have long favoured established, high-authority sources for broad queries while still surfacing specific business content for narrower ones. Recognising that AI search largely inherited this pattern rather than inventing it makes the strategic response feel less like a new problem and more like a familiar one applied to a new interface.