AI Health Search Study: Phase I
This study explores how AI Search is shaping both Patient and HCP journeys across a wide range of health conditions. In addition, the study examines the impact within rare diseases.
AI Health Search Study: Search and AI are Co-Dependent
November 5, 2025 – AI has been transforming business and consumer behavior over the past few years. In addition to AI powered search, many standalone AI models access search indexes—to perform live retrieval of top-ranked search results, which allows AI to incorporate recent, relevant content into its responses. Since Google released AI Overviews (AIOs) in May 2024, the prevalence of AIOs in the search results has steadily increased. The saturation of AIOs varies by industry. However, since September 2024, the healthcare category has seen a significant increase in AIOs (Source: BrightEdge 2025). Currently, healthcare-related searches arguably show the highest frequency of AIOs. As a result, questions have emerged regarding their impact on paid ads, click-through rates, and the importance of organic search rankings on both AI powered search and standalone AI models. Sosemo conducted a study to address these questions. The study explores how AI Search is shaping both Patient and HCP journeys across a wide range of health conditions. In addition, the study examines the impact within rare diseases.
Methodology
- Sosemo conducted an internal study to assess the impact of AI Overviews (AIOs) on healthcare searches. The study focused on keyword searches across the patient journey, in addition to HCP and rare disease searches.
- Ten of the top healthcare conditions with available treatment options were identified for inclusion in the study (namely: hypertension, diabetes, asthma, COPD, heart disease, depression, prostate cancer, Alzheimer’s disease, allergies, and obesity).
- Keyword research was conducted to identify phrases commonly searched at each phase of the patient journey (i.e., signs and symptoms → condition and diagnosis → treatment options → branded treatments), as well as phrases frequently searched by HCPs and within the rare disease category.
- Ten keyword phrases—one for each of the ten conditions—were mapped to each patient journey category, HCPs, and rare diseases.
- A Google search was performed for each keyword phrase and repeated ten times.
- For every search, the presence of AIOs and paid ads was recorded, along with the presence of pharmaceutical brand sites, pharmaceutical-sponsored disease awareness sites, and pharmaceutical corporate sites within the source list of each AIO.
- The study also counted the number of AIO sources that appeared within the top ten organic search results.
- This methodology was then replicated on Bing.
Get the Full AI Health Search Study.
Contact Us
AI Overviews for Health Searches
The study revealed that AIOs were consistently displayed in Google’s search results for health-related queries across the patient journey, as well as for HCP and rare disease searches. AIOs appeared slightly less frequently in Bing’s search results.
AIOs increase disease and brand awareness as a form of earned media, enabling brands to reallocate budgets toward lower funnel paid ads (such as treatment options), which continue to be regularly displayed.
AI Overviews Impact on Paid Ads from Health Brands
Paid ads appeared in a lower percentage of searches, with the majority targeting keywords related to treatment options; however, Google has recently stated that overall search volume has been trending upward (source: https://blog.google/products/search/ai-search-driving-more-queries-higher-quality-clicks/), indicating that paid ads are and will continue to deliver clicks.
Advertisers seem to be investing less in Bing, with ads appearing in the study only for treatment (branded) and rare disease searches. Branded searches typically deliver strong ROI. Furthermore, given the nature of rare diseases, it is crucial to target audiences wherever they self-identify, including on Bing. Therefore, these insights are not surprising.
The Pharmaceutical Industry’s Influence on AI Answers
The study revealed that pharmaceutical brand sites and corporate websites have not been used effectively to influence health-related AIOs outside of branded searches, at least partly due to a lack of SEO. However, disease state awareness (DSA) sites were shown to be more influential, especially for rare diseases where these sites showed significant impact. DSA sites also showed some influence over AIOs generated for searches related to signs and symptoms.
DSA sites should be considered a “must-have,” especially for rare disease brands. Brand sites have the potential to influence AIOs but will require greater investment in SEO to be favored as an AIO source.
Unlike Google, Bing typically places more importance on content relevance and less on website authority when determining organic search rankings. This may explain why brand and corporate sites have even less influence on AIOs on Bing, while DSA sites have even more.
Search Engines’ Role in Source Citations for AI
Google considers substantially more sources for AIOs than Bing, which may be related to Google’s larger site index. Furthermore, there is greater variation in the number of sources referenced on Google, especially for treatment options, suggesting that Google incorporates additional validation when user intent involves key decision-making. Therefore, there are more opportunities to influence treatment decision-making by optimizing a broader range of sources through SEO and by updating sourced third-party sites as a GEO tactic.
AI powered search and many standalone AI models cite sources that were used to generate the AIOs and AI answers that were gathered from the organic search results. On both Google and Bing, approximately half of the sources were ranked at the top of the organic search results, implying that SEO rankings are highly influential and serve as the basis to ground the overviews.
The study evaluated sources against the top 10 organic rankings on Google, which was found to be a flawed approach, as it was revealed that Google sourced between 9 and 41 sites (with an average of 18.1) for AIOs. If organic search rankings within the number of sources referenced were used as the study criteria, a significantly higher percentage of sources would have been reported as ranking within Google—estimated at 50%.
Bing sources considerably fewer sites for its AIOs (average of 5.48); however, a similar percentage of top organically ranked websites—compared to Google—are used to ground the AIOs. There is greater variation across health categories in the percentage of organically ranked sites referenced as sources on Bing, which suggests additional filtering.
As with Google, brands can influence AIOs on Bing (ChatGPT, Copilot, OpenAI, etc.) by securing top rankings in the organic search results through investment in SEO and GEO. Specifically for Bing, it may be more advantageous to focus on achieving organic rankings for condition and diagnosis, treatment (branded), HCP, and rare disease keywords.
Conclusion
Health brands should consider SEO and GEO as a vital component of their marketing mix due to the high prevalence of AIOs within the search results, as well as the role of search results within standalone AI models. Generally, brand sites are more effective at influencing AI for lower funnel prompts, while DSA sites have more influence over the upper funnel. Paid search remains a key marketing channel. However, investment budgets may be re-allocated to lower-funnel searches. DSA sites are essential, especially for rare diseases brands, due to their influence on AIOs and AI answers.
Given that AI source ranking factors are new and evolving, there is an opportunity for brands to increase their visibility within AI more easily with on-page optimization strategies. Reflecting on how the search algorithms evolved over time. Earlier algorithms focused more on on-page optimization but later shifted to off-page factors that were more challenging to influence as competition increased. This is likely to occur with the AI source algorithms as well. Therefore, brands that implement GEO strategies now will likely benefit more in the long-term.
What’s Next?
Sosemo is conducting a Phase 2 study that will provide additional insights into how AIO sources are determined. The Phase 2 study will be used to identify the key ranking factors for AI source citations. These insights will be synthesized into optimization best practices and included within Sosemo’s GEO service offerings to help health brands elevate their coverage with AIOs and AI Answers across the health sector.
Get Early Access to The Phase 2 Study.
Contact Us
What is GEO?
GEO or Generative Engine Optimization is the process of refining and structuring a brand’s content so it is effectively recognized, prioritized, and surfaced by AI-powered search engines and generative models, increasing visibility in AI-generated answers and overviews.
Learn More About Our GEO Services.
Contact Us
The core components of Sosemo’s GEO service offering includes the following:
AI Visibility Reporting
AIOs and AI-generated answers vary considerably with each prompt and are not static. Additionally, numerous models prioritize and source information differently. Repeat simulations are run across relevant AI models, and the results are used to compute brand visibility scores, assess sentiment, aggregate sources, and identify hallucinations or inaccurate information.
AI Measurement Plan & Goal Setting
Standardized AI metrics such as brand visibility scores, answer sentiment, click-through traffic, traffic attribution, source rankings, and hallucination frequency are included in the measurement plan, along with custom metrics aligned to strategic imperatives like target audience sentiment. AI visibility reporting provides baseline data to support goal setting.
AI Prompt Keyword Mapping
Prompts and natural language questions are common inputs for AI models, including AI-powered search. Additionally, search engines use natural language processing (NLP) and semantic search to interpret intent, followed by query parsing to determine search results that feed AI responses. Therefore, optimizing for filler words is unnecessary for achieving visibility.
GEO Content Recommendations
Keywords that trigger AIOs, along with those parsed from prompts and natural language questions and prioritized based on AI visibility and analytics reporting, are targeted. Content optimization recommendations incorporating SEO best practices are provided for relevant new web pages and/or owned AI-sourced content.
GEO Technical Enablement
Due to the limited reasoning and processing speed of LLMs, inaccurate outputs often occur. To improve accuracy, technical enhancements such as llms.txt files, ai.txt files, and schema markup/structured data are essential to help LLMs efficiently locate, interpret, and prioritize key site content.
AI Visibility Monitoring
AIOs and AI-generated answers rely on the most current information available, causing responses to change and evolve as new data is incorporated. The underlying technology and algorithms also evolve rapidly, further driving these shifts. Ongoing AI visibility monitoring and monthly reporting keep brands informed about their AI footprint.
Monthly Dashboard Reports
Monthly dashboard reports include data from both analytics and AI visibility monitoring. Insights and recommendations are provided for all metrics outlined in the measurement plan. These reports track performance against goals and inform ongoing optimizations.
Quarterly Deep Analysis & Strategy Development
Each quarter, a deep analysis is conducted based on an assessment of metrics and performance to date. This analysis examines the sources most influential in AIOs. Strategies to address hallucinations and inaccurate information—whether through further optimization, content suppression, or outreach—are then developed.
