AI search already has vast amounts of established knowledge to draw on. Repackaging that material gives it little reason to cite one business over another.
What creates value is information that adds to what is already known: new findings drawn from proprietary data, first-hand experience or original analysis. In search, this is known as ‘information gain’.
As generative AI makes generic content cheaper to produce and easier to reproduce, the value of that original evidence rises.
Why Information Gain matters in AI search
AI search engines do more than direct users to websites. They retrieve information from multiple sources and use it to construct an answer.
A page that repeats the accepted view may still be relevant, but it is also one of many substitutes. A source with original evidence gives the system a distinct fact, finding or conclusion to draw on and attribute.
The same evidence matters to prospective clients. It can answer a question they are actively considering, help them make a decision and demonstrate the quality of thinking behind the business.
Google’s current guidance for generative AI search advises publishers to produce valuable, non-commodity content based on unique viewpoints and first-hand experience, rather than recycling material already online or easily generated by AI.
Its patent for ‘contextual estimation of link information gain’ describes assessing what a document adds beyond material already presented to a user. The patent does not confirm ‘information gain’ as a live ranking signal, but it shows Google has developed a way to evaluate the same underlying quality.
For publishers, visibility is no longer just about covering a subject. It increasingly depends on contributing something useful to it.
‘Information gain’ starts inside the business
Many content strategies begin with a topic or keyword, followed by a brief to produce a page about it. The result may be well written and technically optimised, but it will not add new information if it draws on the same public sources as everyone else.
Information gain starts earlier, with knowledge the organisation already holds:
- proprietary customer, operational or market data
- patterns its people observe through their work
- original research, interviews or analysis
- results and lessons from completed projects
- comparisons or conclusions that resolve an unanswered question
Much of this material is never published. It remains in spreadsheets, internal reports, client conversations and the accumulated experience of staff.
The editorial task is to find the questions that matter to the audience, identify what the organisation can credibly establish and turn that material into clear, verifiable findings.
Client case study: From internal knowledge to primary source for AI
An Australian advisory firm had proprietary market data and years of direct experience that could help prospective clients make specific property investment decisions.
The campaign started with those decisions, then combined the firm’s data with the patterns its advisers were observing on the ground. From that material, we developed original findings supported by clear evidence and reasoning.
The research helped prospective clients make better-informed decisions, demonstrated the firm’s analytical capability and contributed findings that other organisations could reference.
It was published as a detailed web page and a downloadable report. The web version made the findings accessible to readers and discoverable by Google and AI systems, using clear sourcing, authorship, purposeful links and structured data. The downloadable report provided greater depth, while report downloads and enquiry forms connected the research to measurable actions.
The results:
- 28.6% of all site conversions from one page – converting at 4x the site-wide rate
- Site-wide AI traffic at 9x the Australian benchmark
- The firm achieved 53% share of voice across tracked AI prompts – more than all its business competitors combined
- A third of all AI referral traffic landed on this single flagship report
- Across all traffic sources, the flagship report was the site’s second most-visited page, behind only the homepage
- AI-referred sessions across the website converted at more than 2x the site-wide rate
- The flagship report’s Google organic conversion rate was 5x the channel average
Its advantage was not length or an AI writing formula. The report established evidence that was useful to prospective clients and unavailable from competing sources, then made it easy to find, understand and act on.
That is information gain in practice.
The full approach and results are set out in our original research and AI search case study.
What can your business establish?
Most businesses already hold knowledge that could become original evidence. It simply has not been identified, tested and shaped for an external audience.
The opportunity begins with two questions.
What can your business establish that others cannot?
Which decisions could that knowledge help your clients make?
If an article can be produced without speaking to the organisation’s people or accessing its data, it will probably contain little that only that business can contribute.
Original evidence still needs the infrastructure around it: search strategy and technical structure to make it discoverable and attributable, distribution to extend its reach, and conversion design to give interested readers a clear next step.
Media Collateral helps organisations turn their knowledge into original research, then builds the search, AI and conversion pathways around it.
AI has dramatically increased the supply of content. Genuine first-hand knowledge remains scarce.
