Links October 14, 2026
Brand mentions and digital PR: what confirms authority in the AI era
Google’s 2026 guide says fake mentions do not work. What works is what the media actually quote: original data and expert opinion.
Brand mentions and links are two sides of one question today: does someone trustworthy talk about you on their own initiative. The AI era did not change that principle, it only made deviations easier to spot.
What counts as a mention and what is just noise
A mention is third-party content that names your brand in context, with or without a link. Noise is a mention you ordered, wrote and placed yourself, which represents nobody’s independent judgement.
Google’s May 2026 guide addresses this directly: hunting for fake mentions across blogs, videos and forums is not as helpful as it seems, because quality systems and spam filters work against it.
A live example: how data becomes a citation
Example: in July 2025 Pew Research Center published a short study showing that people clicked a link in 8% of visits to searches with an AI summary and 15% without. Many publications and search and SEO firms quoted that number. Those mentions arose because the data was original, the method described and the sample clearly stated.
That is the core of digital PR. Not a press release, but a number another author needs to support their own claim.
What works for a small or mid-size site
- Original data: a survey, audit results, an anonymised comparison across your clients, a consolidated public table
- Expert opinion: a named author with experience and contact details a journalist can call
- A tool or calculator people use and reference
- Local context: Estonian and Nordic data that global sources do not cover
- A clear author and brand on every page: Google’s guidance asks you to show who created the content and how
What not to do
- Buy dozens of “mentions” in low-value directories
- Write about yourself on a forum under three different names
- Distribute the same press release in bulk with no editorial layer
- Change dates without changing content to look fresher
How to measure
- Track brand-query trends (searches by your name) in Search Console
- Once a month, collect references that came from your data rather than from a press release
- Ask AI assistants questions on your topic and see which sources they show
- Rate each mention: is the source independent, on topic and read by real people
In May 2026 Google also noted the rollout of “Preferred sources” to AI Overviews and AI Mode, which lets users favour specific publishers. The conclusion is simple: lasting trust with one readership beats a one-off mention.
What not to measure
The count of mentions and “domain authority” say nothing when there is no reader. Measure instead whether your data is quoted, whether brand queries grow and whether AI answers cite your page. These three signals are harder to fake.
A practical note
In a small company one well-made piece of data work per quarter is enough. Better one overview that gets quoted than ten press releases that go unnoticed.
30 days: how to start
- Week 1: pick one question in your field that lacks a good data-based answer
- Week 2: gather data (a survey, an audit, a public dataset) and document the method
- Week 3: write a short overview with a table or chart and a named author
- Week 4: offer it to three publications in the field that have written on the topic before, and track the replies
What a journalist wants
A journalist needs a number, a quote and a source they can cite. Instead of a press release, send three sentences, a table and a contact. If your data is unique, the rest is easy.
