Most "SEO trends" articles list things that changed last quarter, half of which will not matter next quarter. If you buy links or sell placements, that churn is expensive: you chase tactics instead of building the few signals Google has rewarded for years.
This post covers the shifts that have already reshaped search and are still compounding: AI answers and how sources get cited, E-E-A-T and entity signals, the helpful-content and scaled-content-abuse policies, Core Web Vitals, intent changes, video, and zero-click results. For each, you get what changed and what to do about it, plus a "what to stop doing" list at the end.
AI answers and the new value of being cited
Google's AI Overviews, plus answers from ChatGPT, Perplexity and similar tools, now sit between the searcher and the ten blue links. Those systems assemble an answer and then cite the sources they leaned on. That changes what a "good link" is.
Classic links still pass authority and matter for rankings. But the sources cited in AI answers tend to sit on domains already treated as reliable for the topic, with the brand mentioned in the same context across several of them. An unlinked mention of your company on a respected industry site is no longer a missed link opportunity; it is a signal in its own right.
What to do
- Track brand mentions, not just backlinks. Set up alerts for your brand and founders' names.
- Prioritize placements on sites genuinely read in your niche over sites that merely have a high domain score. Topical fit gets you cited.
- Answer the core question completely in the first two hundred words, then expand. AI systems extract; make the extractable part correct and specific.
E-E-A-T and entity signals
E-E-A-T stands for Experience, Expertise, Authoritativeness and Trustworthiness. It comes from Google's Search Quality Rater Guidelines; it is not a single ranking factor, but Google's systems are tuned to reward pages that show those qualities. The "Experience" part is the one most sites ignore: content by someone who has actually done the thing, with evidence.
In practice, E-E-A-T means entity clarity: real author pages with credentials, an About page that says who runs the site, and consistent bylines across the web, so Google can connect author, brand and topic.
What to do
- Give every article a named author with a bio page linked to their LinkedIn or professional profile.
- Add first-hand detail: your own screenshots, your own test numbers, photos you took. Generic advice is what the helpful-content system demotes.
- Add Organization and Person structured data; the technical audit checklist covers validating it.
- When you buy a guest post, publish it under a real expert's byline, not "Admin" or "Guest Author."
Helpful content and scaled content abuse
Two policies define the line. The helpful-content signals, now part of Google's core ranking systems, demote sites whose content exists mainly to rank rather than to help. The scaled content abuse spam policy targets large volumes of pages produced, by any method, primarily to manipulate rankings. Google is explicit that the method does not matter: mass AI generation, spun articles and scraped content are treated the same way.
Neither policy bans AI-assisted writing. They penalize publishing at volume without adding anything a searcher could not get from the result already ranking first. The tell: if you cannot name the person who checked the page for accuracy and added original insight, the page is at risk.
Every page should be able to answer one question: what does a reader get here that they cannot get from the page currently ranking first?
What to do
- Audit for pages with no clicks in Search Console over a long window; consolidate, improve or remove them.
- If you use AI drafts, add an editorial step where a subject-matter expert changes the substance, not just the wording.
- Publishers: reject guest posts that read like a template with the keyword swapped in. They put your own rankings at risk; the publisher's guide to guest posting shows how to set standards buyers respect.
Core Web Vitals: a baseline, not a differentiator
Core Web Vitals (Largest Contentful Paint, Interaction to Next Paint and Cumulative Layout Shift) are part of Google's page experience signals. Failing them can hold a page back; passing them does not lift a page above a competitor with better content.
Fix what Search Console's Core Web Vitals report flags as "poor," get the "needs improvement" URLs green where it is cheap, then stop. Hours spent shaving another hundred milliseconds off LCP are better spent on content and links.
Intent, video and zero-click results
Search intent keeps shifting
The same keyword can carry a different intent than it did a few years ago. "Best CRM" used to return listicles; increasingly it returns comparison tables, AI summaries and video. Informational queries often turn transactional because the informational part is answered on the results page itself.
What to do
- Re-check the live SERP for your top twenty keywords twice a year. Note the dominant format (guide, list, tool, video, product grid) and match it.
- Mine "People also ask" and the follow-up questions AI Overviews suggest; those are the sub-intents your page must cover.
Video and short-form
Video results appear in a growing share of how-to and product queries, and YouTube is a search engine in its own right. A two-minute screen recording that answers one specific question, with a transcript on the matching page, can rank in both places and gives AI systems text to cite. Add VideoObject structured data and title the video to match the query.
Zero-click results
Many searches now end without a click because the answer is on the results page. You cannot reverse that; you can decide which queries are worth fighting for. Purely factual queries ("how many ounces in a pound") are gone. Queries where the searcher must evaluate, compare or buy still send clicks, often better qualified than before. Focus your budget there, and treat snippet and AI Overview visibility as brand exposure rather than traffic.
What links look like now
Put the trends together and the shape of a valuable link is clear:
| Link characteristic | Still matters | Why |
|---|---|---|
| Topical relevance of the linking site | Yes, more than ever | Drives both classic authority and AI citation likelihood |
| Real traffic on the linking site | Yes | A site nobody reads is not a source anyone cites |
| Raw domain score (DA/DR) | Somewhat | Useful filter, poor sole criterion |
| Exact-match anchor text | Less | Over-optimized anchors are a risk; see the anchor text guide |
| Brand mention with or without a link | Yes | Feeds entity recognition and AI answers |
When you evaluate a placement on BacklinkCAT, apply the niche and Semrush traffic filters before the DA or DR filter. A listing in your exact category with steady traffic and an ownership-verified badge beats a higher-scored general site.
What to stop doing
- Stop publishing to hit a quota. Ten pages a week with nothing new in them is exactly what scaled-content-abuse policy describes.
- Stop chasing domain score alone. A DR 70 site with no readers in your niche is a weak source.
- Stop using exact-match anchors on paid placements. Use brand and natural phrases.
- Stop targeting pure-fact queries. The results page answers them; you will not get the click.
- Stop hiding who wrote your content. Anonymous articles lose on E-E-A-T.
Where to start
- Add real author pages and Organization/Person schema this week.
- Run a thin-content audit and consolidate anything with zero clicks over a long window.
- Re-check the SERP format for your top keywords and rewrite the mismatches.
- Shift link budget toward topically relevant, actually-read sites, and count brand mentions as wins.
- Clear the "poor" Core Web Vitals URLs, then move on.
These are not disposable trends; the sites that do the boring parts well are the ones AI answers keep citing.