Google Discover Dive Deeper introduces an important publisher question: what happens between someone noticing a story and reaching the source? In the announced test, a button opens an AI-generated topic overview with links to related coverage. The experiment starts with videos.
That creates a possible additional stop in the reader’s journey. It may also introduce supporting coverage that the reader would not otherwise have encountered. Neither outcome tells us how much traffic an individual publisher will gain or lose.
The useful SEO response is to understand the experience, strengthen the evidence behind your coverage, and measure the right outcomes. If you are comparing observations with other publishers, the Scale Xpert Discord community is a place to share examples with their context intact.
What is Google Discover Dive Deeper?
According to Search Engine Roundtable’s September 18 coverage, Google’s Robby Stein announced an experiment that opens a topic overview with links to stories, community reactions, and original reporting. Google said it would begin with videos and test multiple designs. That describes an experiment, not a confirmed universal rollout.
Discover is an interest-based feed, so it does not require the user to type a query first. General Discover eligibility requires indexing and compliance with its content policies; no special tags are required, and eligibility does not guarantee an appearance. Google’s Discover documentation.
Keep this separate from AI Overview links that open AI Mode in Search. The entry point, interface, and reporting questions are different. Treating every Google AI experience as the same product would lead to the wrong optimization plan.
Where does the overview get its information?
The public example gives us a view of the output, not a full explanation of retrieval. Search Engine Journal describes a weather video, an AI summary, publisher source cards, and further stories grouped by subtopic. It also notes that the announcement did not specify the test’s countries, languages, or audience size.
A visible link shows that Google offers that destination in the experience. It does not establish that every sentence came from that page, or reveal every source involved in generating the overview. The sources reviewed do not provide a complete selection formula for Dive deeper.
For publishers, the practical question is which contribution your page can make. Are you providing firsthand reporting, an explanation, a useful reaction, or supporting evidence? Decide that before trying to reproduce the wording of a summary.
| Possible contribution | Useful material to prepare | What it does not prove |
|---|---|---|
| Original reporting | Firsthand observations, documents, interviews | Guaranteed source selection |
| Topic explanation | Definitions, timeline, relevant context | Use of your page in generation |
| Expert reaction | Named expertise and a specific interpretation | Preference for roundups |
| Supporting evidence | Methods, source data, reproducible examples | Guaranteed traffic from a citation |
This table is an editorial planning tool. It is not a list of ranking factors confirmed by Google. Use it to identify what readers would lose if your page disappeared.
Does the test mean publishers will lose traffic?
There are at least two plausible outcomes. A reader may stop after the summary, reducing the chance of a website visit. Another reader may discover a related publisher through the overview and open a page they would otherwise have missed.
You need evidence to determine the balance. An extra screen is a change in the journey, not a measured percentage decline. A publisher traffic drop during the test period also does not identify the cause by itself.
Frame the risk as a testable question: are relevant readers reaching your pages, and are those visits useful? That keeps the discussion connected to your business. It also prevents a dramatic industry headline from becoming the explanation for every daily fluctuation.
How to improve your usefulness as a source
Give readers evidence they cannot get from another rewrite
Select one contribution you can genuinely substantiate. For an SEO publisher, that might be a documented interface observation, an anonymized audit example, or a reproducible troubleshooting procedure. State what you checked, when you checked it, and where the finding stops applying.
Build the article around that contribution instead of adding a token expert quote to a generic summary. An observation log or comparison worksheet can become one of your original, linkable assets. Its value should remain clear even if no AI system cites it.
Google’s helpful-content guidance encourages original information, clear sourcing, and evidence of expertise. That supports this editorial direction, but does not establish a Dive deeper citation recipe. Google’s content guidance.
Keep claims and their limitations together
Write an important finding so someone can understand its scope without hunting through the footer. Name the sample, observation date, and relevant conditions near the claim. If the example is hypothetical, label it before presenting numbers.
Consider the difference between “AI Discover destroys traffic” and “This recorded session required a second tap to reach the linked article.” The second statement is narrower and easier to verify. It becomes useful evidence when the recording and observation conditions are available.
Add expert interpretation when it changes the answer
Ask a contributor to explain a specific consequence or limitation. For example, an analytics specialist could review whether your comparison uses compatible denominators. A general statement that publishers should create good content adds little.
The same principle applies to useful expert contributions: the contribution should help the reader, not exist solely to obtain a mention. Identify the person’s relevant experience and distinguish opinion from measured results. Neither a famous name nor a large roundup guarantees selection.
Make the headline compelling and complete
Google advises Discover publishers to avoid misleading, exaggerated, or information-withholding previews. It also recommends relevant large images, at least 1,200 pixels wide, with large previews enabled. Those are general Discover practices, not special Dive deeper ranking signals. Discover publisher guidance.
For this story, “Google Discover Dive Deeper: Will Readers Reach Your Website?” identifies both the feature and the question. “Google Just Killed Publisher Traffic” claims an outcome the evidence does not establish. Curiosity works better when the article can deliver exactly what the headline promises.
Should publishers start making more videos?
Starting with video does not mean every publisher should pivot production. Make a video when demonstration, movement, or a spoken explanation improves the story. A weak clip produced only to chase an experiment consumes resources without adding much for readers.
As a hypothetical example, an SEO team could record a short walkthrough of an interface change. The companion article could provide the observation conditions, destination URLs, comparison table, and limitations. The video shows the interaction; the article lets someone inspect the evidence.
For publishers pursuing Google video features, Google’s documentation describes dedicated watch pages where a single video is the main content. An article with a supplementary clip is a different page type. These are video-search practices, not confirmed requirements for Dive deeper inclusion. Google’s video guidance.
If your existing article already contains the relevant investigation, improve it rather than producing a duplicate. Focus refreshing old content on new evidence, corrected details, and clearer explanations. Preserve the original publication history and explain substantive updates.
Measure exposure, clicks, and business outcomes separately
Start with the correct reports
Google introduced dedicated generative AI performance reports for Search and Discover in June 2026. Its announcement now says the rollout reached websites worldwide on August 31, while the data remains included in overall reporting. These are additional views of visibility, not separate traffic to add to the totals. Official announcement.
The AI Discover help page documents impressions grouped by page, country, and date. Links must enter view; the documented report does not offer a Dive deeper breakdown or click metric. Search Labs experiments are excluded, and the announcement reviewed does not establish whether this test belongs to Labs. AI Discover report documentation.
Use the standard Discover report for its clicks, impressions, and CTR, subject to reporting thresholds. Its page metrics are assigned to canonical URLs, and recent data can be preliminary. Check those details before comparing exports with landing-page analytics. Discover performance documentation.
Do not manufacture a feature-specific CTR
Imagine 500 AI Discover impressions and 40 clicks in the overall Discover report. Dividing 40 by 500 produces 8%, but does not establish Dive deeper CTR. The clicks and impressions have not been shown to describe the same feature or exposure set.
Keep those figures in separate columns with accurate labels. Add observed source appearances and website conversions as different measures. For Scale Xpert, a Discord invitation click should also remain separate from a confirmed community join.
Record the actual source journey
When the test appears in your own feed, capture the initial card, overview, and publisher destination. Record the date, country, device, app version where available, and account conditions without exposing private account information. Personalized observations cannot represent the entire Discover audience.
| Audit field | What to record |
|---|---|
| Initial item | Topic, creator, content format |
| Overview | Summary wording and visible source cards |
| Attribution | Domain, destination URL, associated claim |
| Context | Conditions, timestamp, design variation |
| Limitation | What the capture cannot establish |
You can compare these records in the Scale Xpert publisher discussion community. Include cases where the feature does not appear. A collection containing only striking examples will exaggerate how common the experience is.
What should your team do first?
Choose three existing stories where you have something original to contribute. Audit their evidence, mobile presentation, relevant visuals, and next action. Start by prioritizing work by business relevance, rather than publishing unrelated trending stories to seek exposure.
Next, establish a reporting baseline and log editorial changes. Compare similar periods, acknowledge topic cycles, and retain examples that contradict your initial expectation. Treat early findings as observations to investigate, not causal proof.
Frequently asked questions
Is Dive deeper the same as AI Mode?
The reviewed announcement describes a Discover experiment. It should not be treated as the same interface as AI Mode or AI Overviews in search results.
Does the feature remove publisher links?
The announced example retains related-source links. The concern is how the additional summary affects reader behavior, which still requires measurement.
Can I guarantee a citation by adding schema?
No verified Dive deeper formula was identified. Use appropriate markup for its documented purpose rather than expecting it to force selection.
Must I publish videos to become a supporting source?
The experiment starts with video items, but the example includes related written coverage. No universal video requirement for supporting publishers was established.
Can Search Console isolate Dive deeper performance?
The reviewed documentation does not provide that breakdown. Keep feature observations separate from broader Discover metrics.
Only when each serves a substantial, different need. Otherwise, expand the useful source page and keep its supporting evidence together.
Conclusion
Google Discover Dive Deeper deserves a focused publisher response: make your contribution verifiable, give readers a reason to visit, and measure outcomes without confusing exposure with traffic. The test creates questions about attribution and discovery, but does not answer them for every website.
Build your next update around evidence you can defend. To compare methods and observations with other practitioners, join Scale Xpert on Discord.
