The way we search for information is undergoing one of its most significant transformations to date, a shift we will undoubtedly continue discussing in the years to come. We are entering the era of generative or conversational search, where users engage in a natural, dialogue-like interaction with an AI system that responds with complete, contextualised answers rather than a list of hyperlinks. This experience feels increasingly human, highly personalised, and remarkably efficient.
It is precisely this seamless, conversational quality that is accelerating the widespread adoption of AI-powered search. The current boom has prompted intense competition among major technology companies, each racing to offer the most capable AI solutions. The numbers speak for themselves: according to OpenAI CEO Sam Altman, ChatGPT reached 800 million weekly active users as of October 2025, while Google Gemini reports around 450 million monthly active users, according to CNBC. For many people, turning to an AI search tool is already becoming a first choice, thanks to its ability to synthesise information instantly and significantly reduce the time spent navigating traditional search engines.
The new Google search?
Although AI search engines have not yet overtaken Google in terms of influence or scale, their growth trajectory is undeniable. At OTT, we are already observing this shift. Our monthly analytics reveal a sharp rise in website visits originating from AI search tools. In fact, the numbers have doubled in just two months. This is clear evidence that audiences are increasingly relying on AI assistants to surface relevant content.
However, this transition also brings challenges. Not all websites permit AI crawlers to access their content, and there is growing concern about the accuracy and transparency of AI-generated responses. These issues highlight a crucial point: organisations can no longer afford to ignore how AI models discover, interpret, and present information.
For think tanks in particular, the implications are profound. If your content is poorly structured, not machine-readable, or not easily discoverable, AI systems may overlook it entirely or, worse, misrepresent it. Preparing for this new landscape is no longer optional. The sooner organisations begin adapting by improving their website architecture, enhancing metadata, and rethinking how content is displayed, the sooner they will become visible and trustworthy sources within AI-generated search environments.
This shift also reinforces the need to evolve our digital strategies. Relying solely on SEO is no longer enough. While SEO will remain relevant, organisations must also invest in SEA (Search Experience and AI optimisation) and broader visibility tactics to ensure their work reaches the right audiences through both traditional and AI-powered pathways.
Generative search is not just reshaping how we find information; it is reshaping how we must present information. Those who adapt early will strengthen their visibility, credibility, and long-term reputation in an AI-dominated information ecosystem.
What we’re learning about optimising content for generative search
As generative AI tools increasingly influence how people access policy knowledge, many think tanks are still in an early phase of understanding what “visibility” really means in this new environment. At OTT, our communications work over the past year — across major publications, learning programmes, and collaborative initiatives — has surfaced a set of emerging lessons. These are not yet fully systematised practices, but insights we are actively reflecting on as we plan future cycles of work.
We share them here as considerations that other think tanks may find useful as they navigate similar transitions.
- Clarity and structure matter more than we expected
One early lesson from working on initiatives like the State of the Sector and the School for Thinktankers is how effectively generative tools respond to content written in clear, accessible language and organised around explicit themes.
In several instances, we noticed that AI-generated summaries of our work were more accurate when the original content included:
- clear headings,
- short introductory paragraphs that defined the scope of the piece, and
- unambiguous descriptions of who the content was for and why it mattered.
This has prompted us to reflect on how future outputs might benefit from stronger upfront framing and more intentional use of metadata — not only for human readers, but also for AI systems that rely on contextual signals to interpret expertise.
- Distilled formats travel further in AI-mediated spaces
Another emerging insight is the value of distilled formats — executive summaries, FAQs, and explainer sections — in shaping how generative tools surface think tank work.
While working on programme pages and campaign content, we observed that when information was broken down into direct questions and concise answers, AI tools tended to reproduce those explanations more clearly. This was especially evident in content related to learning programmes and events, where FAQs helped reduce ambiguity in how the initiative was described by AI assistants.
This has led us to consider whether future research outputs and flagship reports should more consistently include:
- standalone executive summaries written in plain language, and
- structured FAQs that anticipate common questions from non-expert audiences.
- Thought leadership positioning is cumulative — and AI notices
Another important learning is that thought leadership in generative search is built through repetition and focus over time.
At OTT, we consistently write, curate, and share content around specific themes — such as think tank management, funding models, policy influence, and evidence use. Over time, this thematic consistency appears to influence how AI tools associate OTT with these topics. When users ask broad questions about the think tank sector, governance, or evidence-informed policymaking, generative responses increasingly draw on ideas and language that echo this recurring body of work.
This has reinforced for us that “showing up” in AI-generated responses is less about one-off viral pieces and more about sustained, coherent positioning. For think tanks, consistently engaging with a defined set of topics may be one of the most effective ways to signal thought leadership — not only to human audiences, but to AI systems learning who to trust on what.
- Keywords still matter — especially when tied to user intent
While generative search represents a shift away from traditional SEO, our experience suggests that keywords remain important, particularly when they reflect real user needs and behaviours.
A clear example is the OTT jobs board. By consistently using straightforward, high-intent terms such as “think tank jobs,” “policy research roles,” “international development careers,” and “research and communications positions” in job listings, landing pages, and promotional content, we’ve made it easier for AI tools to understand what the platform offers and who it is for.
As a result, when users ask AI assistants questions like “Where can I find think tank jobs?” or “What careers exist in policy research?”, the jobs board is more likely to be referenced or surfaced. This has highlighted the value of being intentional about language — not for gaming algorithms, but for clearly aligning content with the questions people are actually asking.
Monitoring and reporting
Applying these actions will help think tanks remain visible within AI-powered search environments. Yet visibility alone is not enough; it is equally important to understand the outcomes of such optimisation. Once measures are in place, organisations can begin to monitor how their work is represented and referenced by AI tools.
Think tanks should regularly review the way AI platforms summarise or describe their work. This can be done by testing queries related to their publications or experts and comparing the generated outputs with the original material. Such monitoring provides insight into the accuracy and reliability of information reaching users through AI-driven searches.
Another valuable indicator is the analysis of traffic patterns. Sudden increases in visits from AI browsers or platforms may be linked to specific triggers, such as the release of a new report, coverage of an event, or even job searches in the sector. OTT’s experience shows that employment-related queries often produce noticeable spikes. Similarly, many think tank profiles are discovered through the OTT Open Think Tank Directory, which is frequently surfaced by AI search engines.
A call to experiment, learn, and build capacity together
There is no single blueprint for optimising think tank work for generative search. The ecosystem is evolving rapidly, and most organisations are still navigating it through experimentation. This makes now a moment not for rigid rules, but for learning by doing.
We encourage think tanks to:
- Test how generative AI tools are currently representing their work, and where key messages may be lost or distorted.
- Learn from small, intentional changes in structure, language, keywords, and formats — and reflect on what strengthens accuracy and visibility.
- Share emerging practices and lessons with peers to help the sector collectively adapt to this shifting knowledge environment.
At OTT, we are beginning to integrate these questions into our research, learning programmes, and advisory support — working alongside think tanks to strengthen not just what they publish, but how their expertise travels in an AI-mediated world. For organisations looking to explore this space more deliberately, this is an area where collective learning and targeted support can make a meaningful difference.
As generative AI continues to shape how policy knowledge is accessed, think tanks that invest early in visibility, credibility, and clarity will be better positioned to ensure their evidence informs the debates that matter most.