Today’s search news is split between AI discovery and local visibility. Google is giving timely publishers a more prominent place inside AI Mode, while a new analysis shows ChatGPT’s search tooling changed substantially within days.
Local search is just as active: a 1.8 million-profile study offers fresh data on Google Business Profile categories and completeness, and Google is expanding how Local Services Ads turn calls and bookings into paid leads.
In This Edition
- Google Brings Developing-Topic Link Carousels to AI Mode
- ChatGPT Rebuilt Its Search Tool Around a New Query Format
- A 1.8 Million-Profile Study Finds Specific GBP Categories Correlate With Stronger Rankings
- Google Expands Direct Booking and Changes Call Charging for Local Services Ads
- Google Clarifies That Crawlers Download JSON but Do Not Parse It
- Performance Max Is Testing Direct Channel-Prioritization Controls
- Google Search Is Showing Fewer PDFs in Some Queries
- Reports Suggest the August Spam Update May Have Hit Mass-Generated AI SEO Content
Google Brings Developing-Topic Link Carousels to AI Mode
Google made developing-topic link carousels live in AI Mode on August 25, extending a format Google says was already available in AI Overviews. Google Search VP Robby Stein said the carousel is designed to connect users with original coverage and multiple perspectives when a topic is still developing.
The format inserts a horizontal row of article cards inside the AI-generated response. Each card can show an image, headline, source, and publication date, giving timely reporting a more visible role than a standard citation link buried beneath generated text.
Google first announced the developing-topic carousel in May as part of a broader effort to improve connections between AI Search and original reporting. Preferred Sources can also appear prominently in these carousels when the user has selected a publisher.
The August 25 launch matters because it gives publishers another concrete Search surface to watch inside AI Mode rather than treating all AI visibility as a single citation metric.
Why It Matters
For news publishers, industry blogs, and brands covering fast-moving topics, speed still matters, but so does making the article easy for Google to understand as original, timely coverage. Clear headlines, dates, named entities, direct factual openings, and primary reporting all become more important when Google is choosing which links deserve prominent placement inside an AI-generated answer.
Preferred Sources adds another audience layer. A loyal reader who explicitly chooses a publication can increase the likelihood that publication is surfaced for that user, while the developing-topic carousel gives fresh coverage a distinct presentation opportunity.
Sources:
Search Engine Journal
https://www.searchenginejournal.com/google-brings-developing-topic-link-carousels-to-ai-mode/587022/
Robby Stein on X
https://x.com/rmstein/status/1960394353747660930
ChatGPT Rebuilt Its Search Tool Around a New Query Format
Search Engine Journal contributor Suganthan Mohanadasan reported on August 25 that ChatGPT’s visible web-search tool calls changed from JSON to a compact pipe-delimited query format between August 16 and August 20. His analysis is observational and based on browser traffic from one ChatGPT Plus account, so it should not be treated as official documentation of OpenAI’s internal search architecture.
The new format exposes readable search lines that include the search mode, query, freshness window, and sometimes a domain restriction. Mohanadasan also observed that the former search_queries metadata field disappeared from the payload he could inspect.
Across eight tested questions, he found that ChatGPT could vary freshness windows, target specific domains, trigger different search verticals, and sometimes return widgets instead of conventional citation-bearing results.
This builds on earlier reverse-engineering work around ChatGPT retrieval, but it also shows why marketers should be cautious about treating one discovered implementation detail as permanent. The observable search tool changed materially within four days.
Why It Matters
The useful takeaway is not to optimize for one syntax string or reverse-engineered field. The larger pattern is that ChatGPT appears to decompose a user request into multiple searches, sometimes constrain those searches by freshness or domain, and then decide which result types to use.
That reinforces a broader AI-search principle: content has to survive multiple stages, not just rank for one exact keyword. Pages need clear topic signals, explicit entities, useful facts, strong source identity, and content formats that remain valuable when retrieved as one part of a larger answer.
Source:
Search Engine Journal
A 1.8 Million-Profile Study Finds Specific GBP Categories Correlate With Stronger Rankings
Whitespark analyzed 1.8 million Google Business Profiles across 4,209 categories and found that more specific primary categories were associated with stronger local rankings than broad generic categories. The study was published by Whitespark founder Darren Shaw on Search Engine Land on August 25.
Across the full dataset, profiles using specific primary categories appeared in the top 10 about 12.5% of the time versus 9.2% for profiles using generic categories, a roughly 36% higher top-10 presence. The study also found repeated examples where the category most closely matching the search intent outperformed adjacent categories.
Additional categories showed a similar pattern. Profiles with well-matched additional categories ranked 6 to 17 positions better in several category comparisons than profiles using only one primary category. Shaw explicitly cautions that this is correlation, not proof that adding categories alone caused the ranking improvement.
Profile completeness also correlated strongly with visibility. In the study’s completeness index, moving from a score of 0 to 5 was associated with an average ranking improvement of about 19 positions and a top-10 appearance rate that rose from 4% to 13%.
Why It Matters
The study supports a practical local SEO principle: Google Business Profile setup should be specific, complete, and aligned with the actual services a business provides.
But it should not be reduced to a mechanical “add more categories and rank higher” rule. Businesses that maintain complete Profiles are also more likely to invest in reviews, websites, citations, content, and other local SEO work. The strongest takeaway is to choose the most accurate specific primary category, add genuinely relevant secondary categories, complete the Profile, and keep the website and broader local presence strong.
Source:
Search Engine Land / Whitespark
https://searchengineland.com/google-business-profiles-local-seo-success-data-485727
Google Expands Direct Booking and Changes Call Charging for Local Services Ads
Google is making Local Services Ads more transactional in two ways. First, direct booking is expanding from roughly 20 Reserve with Google partners to more than 500. Second, Google says some missed calls and subsequent calls will begin counting as charged LSA leads on October 1, 2026.
For direct booking, advertisers whose Google Business Profile already contains an active supported booking link can have that booking capability enabled automatically in their Local Services Ads. Customers can move from the ad directly into the existing booking flow, and completed bookings are reported as paid LSA leads.
The call-policy change is more operationally sensitive. Google told some advertisers that missed calls during business hours can count as valid paid leads when the caller stays on the line for more than 20 seconds, subject to exceptions. Follow-up calls can also become chargeable when the original interaction did not qualify but a later call meets Google’s lead criteria.
Google says safeguards are being added for spam and routing systems that require users to press a key before reaching the business.
Why It Matters
Local Services Ads are moving closer to an end-to-end lead platform rather than a simple pay-per-call product. Businesses need to evaluate not only ad performance but also call handling, booking-provider setup, missed-call processes, and lead quality.
The direct-booking expansion can reduce friction, but those bookings are paid leads. The missed-call rule raises the stakes further: a business that is difficult to reach during operating hours may pay for more interactions without receiving the same conversion value.
Sources:
Search Engine Land
https://searchengineland.com/google-expands-direct-booking-for-local-services-ads-485807
Search Engine Roundtable
https://www.seroundtable.com/google-lsa-missed-subsequent-calls-41940.html
Google Clarifies That Crawlers Download JSON but Do Not Parse It
Google’s Gary Illyes clarified on August 25 that Google’s crawlers do not parse JSON files during the crawling step. Googlebot downloads the resource, while parsing occurs later in systems such as indexing.
The distinction sounds technical, but it reflects how Google separates fetching from understanding. Crawling retrieves the resource. Parsing and interpretation happen downstream.
Illyes added that Google generally uses shared infrastructure across services, while noting that he had not personally worked on the Merchant Center implementation being discussed.
This does not mean Google cannot use JSON or structured data. It means site owners should be precise about which part of Google’s pipeline is performing the work.
Why It Matters
Technical SEO discussions often collapse crawling, rendering, parsing, indexing, and serving into one vague idea of “Googlebot reading the page.” That can lead to bad diagnoses.
When debugging structured data or machine-readable files, ask whether the problem is retrieval, parsing, indexing, or eligibility. A resource can be fetched successfully and still fail later in the pipeline, the same sequence Googlebot follows for ordinary web pages.
Source:
Search Engine Roundtable
https://www.seroundtable.com/google-crawlers-dont-parse-json-41936.html
Performance Max Is Testing Direct Channel-Prioritization Controls
Google is testing a Performance Max “Channels” setting that lets some advertisers make positive or negative adjustments for Search, YouTube, Display, Discover, Gmail, and Maps. Search Engine Land reports that the feature is in alpha testing and is not broadly available.
The control does not appear to assign a fixed budget percentage to a channel. Instead, a positive adjustment relaxes the CPA the system is willing to accept for that channel, while a negative adjustment tightens it, influencing how aggressively Performance Max pursues conversions there.
That would be a notable change for a campaign type historically built around automated channel allocation. Google already added more channel-level reporting, and this test could create a path from visibility into channel performance toward limited control over channel emphasis.
Attribution remains the complication. A channel that looks inefficient on last-touch conversion data can still influence demand that converts later elsewhere.
Why It Matters
Advertisers have wanted more Performance Max control for years, but more control also creates more ways to make a bad optimization decision.
If this feature expands, marketers should avoid suppressing a channel solely because its directly attributed CPA looks weak. Use assisted-conversion context, customer-journey data, incrementality testing, and channel-level reporting before telling the system where to pull back.
Source:
Search Engine Land
https://searchengineland.com/google-tests-channel-prioritization-controls-for-performance-max-485825
Also Worth Watching
Google Search Is Showing Fewer PDFs in Some Queries
Multiple SEOs reported on August 25 that PDF files appear to be losing Search impressions or disappearing from common queries where the PDF itself previously ranked. Search Engine Roundtable reproduced several examples, including government forms, while other practitioners shared Search Console declines beginning at different points in August.
Google has not confirmed a PDF-ranking or indexing change. Some PDFs still appear when users add filetype:pdf, and at least one prominent government site was also experiencing accessibility issues during the investigation.
For sites that rely heavily on downloadable guides, forms, reports, or manuals, this is worth monitoring in Search Console. It is not yet strong enough evidence to justify converting every PDF into HTML.
Source:
Search Engine Roundtable
https://www.seroundtable.com/google-search-fewer-pdf-files-41946.html
Reports Suggest the August Spam Update May Have Hit Mass-Generated AI SEO Content
Search Engine Journal reported anecdotal evidence on August 25 suggesting that some sites using large-scale automated AI publishing were hit during Google’s completed August spam update. The article also points to a Google research paper on scalable detection of mass-generated AI spam.
Google has not said the August 2026 spam update specifically targeted AI-generated content, and AI-assisted content is not automatically spam. The stronger distinction is between useful content and scaled content produced primarily to manipulate search rankings.
That makes this an important trend to watch, but not a confirmed explanation for the update.
This follows the August spam update first covered as a global rollout in the August 19 Unified SEO Pulse:
Source:
Search Engine Journal
The Unified Take
Today’s stories reinforce how quickly search is splitting into different retrieval, presentation, and conversion layers.
Google’s developing-topic carousel gives publishers a new way to surface inside AI Mode, while ChatGPT’s observable search tooling changed enough in four days to make any static “AI ranking formula” look fragile. Both developments argue for durable content qualities over platform-specific tricks: clear source identity, original information, strong topical labeling, direct answers, and content that still makes sense when retrieved in pieces.
The local-search data tells a similar story from a different direction. Specific Business Profile categories and complete profiles correlate with stronger rankings, but those signals work inside a broader ecosystem that includes the website, reviews, citations, and customer behavior. Local Services Ads are also becoming more integrated with the actual lead journey through booking links and call policies.
On the technical side, Google’s JSON clarification is another reminder that crawling, parsing, indexing, and serving are separate stages. That distinction matters increasingly as marketers optimize for traditional Search, AI answers, structured data, and external AI systems at the same time.
The practical strategy is not to chase every new interface. Build a strong underlying information system: pages that are technically accessible, semantically clear, genuinely useful, and connected to a credible business or publisher entity. Then measure how each search surface chooses to use it.
