For most of the time I have worked in search, which is longer than Google has existed, the conversation with clients has come down to one thing. Rankings.
Can we get this page onto page one. Can we move from position six to position three. Are we going after the right keywords.
Google’s move towards AI-generated search is quietly changing that conversation, and a feature called Preferred Sources is a useful way to see how.
In May 2026, Google extended Preferred Sources into AI Overviews and AI Mode. People can now tell Google which websites they value, and content from those sites can be given extra prominence, and a visible “Preferred” label, when it appears inside Google’s AI answers.
For pharmaceutical and medical-device teams, this is worth a look. Not because Preferred Sources is a clever new SEO shortcut. It is not. It is worth a look because of what it signals about where search is going. Being recognised as a credible source is starting to matter alongside being well optimised for a particular query.
What Preferred Sources actually is
Preferred Sources lets an individual Google user choose the websites they want to see more of. When content from one of those sites turns up in results, Google can give it more prominence, and inside AI Overviews and AI Mode it can carry that “Preferred” badge.
Google says people are around twice as likely to click through to a Preferred Source, though that figure comes from Google’s own data, so treat it as directional. By August 2026 more than 600,000 unique sources had been selected, up from around 345,000 at the May launch. Google has also released an embeddable button that lets a site invite readers to add it in one click.
There are limits worth being honest about. Preferred Sources does not lift your ordinary organic ranking, and it only helps you with people who have already chosen you. So this is not a replacement for SEO. It is closer to a reader-loyalty signal, and that difference is the whole point.
Why any of this matters now
Search behaviour itself is shifting. Google says AI Mode has passed one billion monthly users, with AI Mode queries more than doubling every quarter.
At the same time, the evidence is building that AI answers reduce clicks through to websites. A 2025 Pew Research Center study of US browsing found people clicked a traditional search result on 8% of visits where an AI summary appeared, against 15% where there was no summary. Clicks on the links inside the summary itself ran at just 1%.
Health is particularly exposed, because so many health searches are informational and have short answers. An Ahrefs analysis published in August 2026, looking at 963 domains after AI Overviews launched in France, found health was the most heavily affected sector in its dataset, with AI Overviews appearing on 22.2% of queries and click-through rates falling around 20%. That study only covered the first nine days after the French rollout, so I would treat it as a direction of travel rather than a fixed number. But it points the same way as everything else.
For healthcare organisations, a good ranking starts to matter less if the searcher gets their answer before they reach the website. The question moves from “how do we rank for this query” to something harder: “how do we become one of the sources people, and search systems, treat as worth consulting”.
Why this is more complicated for pharma
Pharmaceutical search carries a layer most sectors never have to think about.
All pharma marketers know that prescription-only medicines (POM) must not be advertised to the public in the UK. That is not softening because Google has added AI. The ABPI Code of Practice is explicit on it at Clause 3.2, and it also requires, at Clause 6, that information and claims are accurate, balanced and capable of substantiation. Clause 8 requires promotional material to be certified before use, and its supplementary information is clear that this includes promotional material on websites and other internet-based systems. None of that changes because an answer engine is doing the summarising.
What it does mean is that pharma teams need to be deliberate about what they want to be known as a source for, and for whom. A pharmaceutical company might hold corporate and pipeline information, disease-awareness content, patient support resources, medical and scientific content, HCP education and promotional HCP content. These are very different information environments, with different audiences and very different governance.
So Preferred Sources raises a more interesting question than whether to bolt Google’s button onto the site. What does your organisation genuinely want to be regarded as an authoritative source of and to which audience.
Site architecture matters more than it looks
There is a detail in how Google has built this that is directly relevant to pharma. Preferred Sources works at domain or subdomain level. Someone can select examplepharma.com or hcp.examplepharma.com, but they cannot separately select examplepharma.com/hcp/.
For any organisation that has piled corporate content, patient information, disease awareness, HCP resources and product information into one large domain, that is worth sitting with. It also lines up with the Code’s own thinking on website access at Clause 16, which expects public information and HCP promotion to be clearly separated so the public are not steered towards material that is not meant for them.
This does not mean every company suddenly needs separate domains. It means decisions about domain structure, audience separation and content architecture now have consequences well beyond site navigation.
Med-tech has the simpler opportunity
For medical-device businesses selling into clinicians and clinical organisations, the opportunity is more straightforward. A lot of med-tech sites are still mostly product pages.
Buyers and clinicians tend to arrive with broader questions. How does the technology work. Where does it sit in the clinical pathway. What evidence supports it. How does it compare with what they use now. What does implementation involve.
Those are exactly the more complex, informational searches where AI-assisted answers become part of the research. A company that consistently publishes clear, useful, properly evidenced answers has a far stronger claim to be a source than one whose website is a digital brochure.
The button is not the strategy
It would be easy to see Google’s new button and decide every healthcare site should add one. That misses the point. People need a reason to choose you.
Google’s own guidance on generative-AI search keeps saying the same thing: create genuinely useful, distinctive content built on real knowledge and experience, rather than rewriting what is already out there, and keep doing the SEO fundamentals properly (think E-E-A-T), because there is no separate trick that makes a site “AI optimised”.
That is squarely relevant to pharma and med-tech, because these are the organisations with the things generic publishers cannot easily replicate. Clinical expertise. Specialist knowledge. Original data. Experienced medical teams. A real understanding of specific disease areas. The opportunity is to turn more of that into useful, accessible, properly governed information. Not more content for its own sake. Better source material.
A few practical starting points
If you want to act on this, these are the sensible places to look:
- Check whether your domain is eligible for Preferred Sources and for Google’s generative-AI features using its source preferences tool
- Work out what your website is genuinely authoritative about, beyond keywords, so you know where you have something distinctive to say
- Review your audience and domain architecture, particularly if corporate, patient and HCP content all sit on one domain
- Strengthen your evidence and authorship signals, so it is clear who produced or medically reviewed key information, when it was last reviewed, and what supports it
- Keep the SEO fundamentals in good order, since crawlability, structure and clear information architecture still underpin visibility in AI features
- Start measuring AI-search visibility separately, using the generative-AI reporting Google has rolled out to Search Console worldwide
Where this leaves us
Preferred Sources is still new, and it would be a mistake to treat it as a ranking breakthrough. Its importance is strategic. Google Search is becoming less about handing you ten links and asking you to choose and more about reading multiple sources and building the answer itself.
In that world, keyword position is only part of the job. The harder question is whether your website holds information clear, useful, trustworthy and distinctive enough to be used as a source at all. And increasingly, whether the people you want to reach see you as a source worth coming back to.
For pharma and med-tech teams reviewing digital strategy, that makes search visibility as much a content, authority and architecture question as an SEO one
Need to know how AI search is treating your content?
Genetic Digital helps healthcare, pharma and life sciences teams review website structure, content visibility and digital governance in regulated environments.
If you want to understand how AI search is evolving and what AI search is citing about your brand, and whether your own UK content is visible and structured well enough to be the source it uses, let’s talk. As part of a first step we can run a practical AI visibility audit across a brand, therapy area, product category, platform, pipeline or the whole of your digital estate.
You can also read more about our Healthcare and Pharma SEO, GEO and AEO solutions.
