The SEO, SEM and AI news for the week of 19 October 2026 brings several changes with practical consequences for search visibility, digital advertising and analytics. Google has changed the way sources are displayed in AI Overviews and AI Mode, Microsoft is making it easier to track citations in generative answers, and Google Ads is introducing new controls to supervise queries, ads and budgets. In addition, there are relevant developments in Tag Manager, Google Analytics and Merchant Center.<\/p>\n
What changes this week in SEO, SEM and AI<\/h2>\n
I have selected fifteen developments published between 12 and 18 October, separating those that can already be reviewed in client accounts from tests limited to certain advertisers or countries. This analysis continues the SEO, SEM and AI news from the previous week<\/a>. As every Monday, the aim is not to collect platform announcements, but to understand which ones justify time, budget or strategic changes and which are better simply monitored.<\/p>\n This week brings an important development affecting both content and measurement. Google is making it easier to access the sources used in some generative answers, while Bing is providing a more thematic view of citations. However, these changes need to be assessed without confusing a brand mention with a visit and, even less, with a conversion.<\/p>\n On 15 October, Google announced changes to the way sources are presented<\/a>. These include more links embedded within answers, a highlighted section for certain official websites, news article carousels and access to creator profiles.<\/p>\n According to Google, the objective is to make it easier for users to explore information published outside the search engine itself. Therefore, the change may create new opportunities for a website to gain visibility within a generative answer rather than relying exclusively on the traditional list of search results.<\/p>\n However, there is no guarantee that greater link visibility will ultimately generate more clicks. I would review branded and service-related searches, recording which source appears, in what context and for what type of query. Within a GEO and AI visibility strategy<\/a>, the objective should be to ensure that accurate and relevant information about the company appears, rather than simply accumulating citations without context.<\/p>\n On 13 October, Microsoft announced Topics Portfolio<\/a>, a beta feature that groups queries used to support artificial intelligence answers into broader topics.<\/p>\n The tool shows citations, the approximate size of each topic, the number of cited competitors and queries where additional visibility opportunities may exist. In addition, it can help identify editorial areas where a website already has some authority and others where it is still barely present.<\/p>\n It is an interesting reference for deciding which content should be expanded, updated or developed. However, Microsoft warns that the groups are generated through automated classification systems, that the values are indicative and that not every website will have sufficient data.<\/p>\n For that reason, I would use it as supporting information within a GEO audit rather than as an exact indicator of market share or as an isolated metric for deciding an editorial strategy.<\/p>\n Research published on 15 October identified a possible loss of UTM parameters and referral data<\/a> in links opened from ChatGPT.<\/p>\n Across a sample of around five hundred pages, visits attributed to the assistant declined while traffic classified as Direct increased. OpenAI has not confirmed that this is a permanent change, so the data should be interpreted cautiously.<\/p>\n Before treating a fall in referrals as a genuine loss of visibility, I would compare landing pages, Direct traffic and conversions. If a visit stops passing its origin correctly, analytics software may simply classify it incorrectly.<\/p>\n Consequently, measuring GEO exclusively through sessions attributed to ChatGPT or any other assistant is particularly risky. Generative visibility needs to combine traffic data, mentions, citations, conversions and brand behaviour.<\/p>\n Search advertising continues to move towards increasingly automated systems. Even so, this week’s main developments introduce new ways of understanding the decisions made by the platform. That balance between automation and supervision is particularly relevant for those of us managing client campaigns.<\/p>\n On 13 October, the Search terms and ad combinations<\/a> report began appearing in some accounts.<\/p>\n This new report makes it possible to connect the queries that triggered an ad with the headlines, descriptions and URLs that were used. It also provides some signals related to the origin of the assets and the type of matching involved.<\/p>\n General availability has not yet been confirmed. However, where it is available, it can become a useful tool for identifying queries that concentrate spend and checking whether the message and landing page selected by the system make sense.<\/p>\n When managing Google Ads and digital advertising campaigns<\/a>, a mismatch should not automatically result in blocking a keyword or restricting AI Max. The quality of the conversions should be analysed first to determine whether there is a genuine commercial problem.<\/p>\n On 14 October, Google Ads announced new options for prioritising channels<\/a> across Search, Shopping, Performance Max and Demand Gen.<\/p>\n Advertisers can set a high or low priority and define an indicative maximum percentage of spend for certain secondary channels. As a result, the platform provides slightly more control over how the budget is distributed across different inventories.<\/p>\n However, the limit does not operate as a completely rigid cap. Google explains that it works on a best-effort basis and, in campaigns that are not spending their full available budget, channel priority may have little effect.<\/p>\n Therefore, before using these controls, I would review how investment is currently distributed and the profitability generated by each channel. The new functionality makes testing easier, but it does not replace an analysis of actual performance.<\/p>\n On 12 October, Google announced that Demand Gen Lookalike segments will move by default towards less restrictive audience signals<\/a>.<\/p>\n Until now, a lookalike audience could operate as a relatively clear boundary determining who would see the ads. Under the new approach, the audience can work more as a signal guiding the system while allowing it to expand delivery afterwards.<\/p>\n The difference is important. An audience suggestion may achieve greater reach, although that additional reach does not necessarily mean better customers.<\/p>\n For accounts focused on generating leads from very specific profiles, I would check the source lists, the cost per valid opportunity and the option of maintaining a stricter restriction. On the other hand, campaigns with greater room for exploration may benefit from testing the expansion and then measuring lead quality.<\/p>\n On 16 October, Google extended the maximum period of a beta for automatic cost adjustments<\/a> from three weeks to sixty days.<\/p>\n For eligible campaigns using Target CPA or Target ROAS, the platform may reduce certain costs when performance during the learning phase falls below expectations.<\/p>\n However, not all campaigns meet the requirements and there is no guarantee that any adjustment will be applied. Where it does appear, I would document the final cost and check how it affects the metrics used in reporting.<\/p>\n In addition, I would not use a potential cost reduction to justify poor targeting, weak creative work or incorrect conversion tracking. Campaign quality should not depend on receiving compensation afterwards.<\/p>\n On 15 October, Google Ads API version 26<\/a> was released, adding metrics for Search Lift studies, Performance Max improvements and new technical resources.<\/p>\n Google is also moving towards a more frequent release schedule. The technical changes can be reviewed in the official upgrade documentation<\/a>.<\/p>\n For those developing integrations, internal tools or reporting systems, the next step is to check compatibility and update libraries where necessary.<\/p>\n It is also important to distinguish Search Lift from Conversion Lift. The former attempts to measure increases in searches related to advertising exposure, while Conversion Lift is designed to estimate additional conversions. They are different metrics answering different questions.<\/p>\n Data quality remains a prerequisite for optimising campaigns and evaluating results. This week, Google has published two developments that may simplify certain technical tasks, although neither removes the need to understand what is being measured and who has access to each property.<\/p>\n On 13 October, an optimisation of Google Tag Manager<\/a> was announced that can send data directly to compatible Google destinations without loading additional gtag.js JavaScript for those connections.<\/p>\n The aim is to reduce dependencies and simplify the management of certain tags. In theory, the change may also reduce some of the code that needs to be loaded in certain implementations.<\/p>\n However, I would not assume a speed improvement without measuring it. First, I would review the installed tags, which events they send and whether duplicate or inherited configurations exist that no longer serve a purpose.<\/p>\n In my GA4 audit and Looker Studio dashboard case study<\/a>, I explain precisely why a reliable dashboard depends first on trustworthy sources and verified configurations. An attractive dashboard does not fix an incorrect implementation.<\/p>\n On 12 October, Google Analytics introduced a User Groups Admin API<\/a> that allows groups, members and permissions to be managed programmatically.<\/p>\n The functionality requires administrative privileges and can simplify onboarding, offboarding and access maintenance in organisations working with a large number of properties.<\/p>\n On the other hand, automating permissions does not fix a badly designed access policy. Before using it, I would identify which users are administrators, which external collaborators still retain permissions and which accesses are no longer necessary.<\/p>\n For an agency or consultancy managing several accounts, the main advantage is reducing repetitive work and documenting changes more effectively. The continuity of an analytics implementation should not depend on remembering who configured a particular property two years ago.<\/p>\n Changes in Merchant Center and new AI-assisted shopping experiences reinforce the importance of product and store data. Even so, not all of these developments currently affect Spanish retailers. Therefore, it is important to separate changes that can already be reviewed in real feeds from experiments that simply need to be monitored.<\/p>\n On 14 October, the store_address attribute for certain local inventory feeds<\/a> was announced.<\/p>\n Its purpose is to associate products with the correct physical location offering them, both in ads and in free local product listings.<\/p>\n For example, if a client operates several stores, I would review the submitted addresses and check that they are consistent with the information on the ecommerce site and the company’s business profiles.<\/p>\n This task connects directly with local SEO and Google Business visibility<\/a>. A customer needs to know not only which product is available, but also where they can physically find it.<\/p>\n However, this is not a relevant configuration for every ecommerce business. An online-only retailer does not need to modify its strategy because of this development.<\/p>\n On 16 October, Merchant Center added the warranty<\/a> attribute to communicate certain types of coverage through the feed.<\/p>\n Google notes that this information may be required in markets with specific commercial disclosure requirements, such as Quebec. In addition, it provides a more structured way to communicate information that may influence purchase decisions for certain products.<\/p>\n However, the attribute is not universally mandatory and there is no demonstrated ranking advantage from using it.<\/p>\n I would review which products and countries require additional information and make sure that the declared warranty period matches the actual commercial policy exactly. A complete product listing can help users compare offers, but only when the data is accurate and kept up to date.<\/p>\n On 15 October, Google began the US rollout of Universal Cart across Search and Gemini<\/a>.<\/p>\n The feature allows users to save products from different merchants, monitor price and availability changes and complete certain purchases through Google Pay or continue later on the retailer’s website.<\/p>\n For Spanish ecommerce businesses, this is currently a trend worth monitoring rather than a generally available feature.<\/p>\n Its strategic importance lies elsewhere: part of the purchasing decision can progress outside the retailer’s own website. As a result, consistent catalogues, updated prices, stock information and high-quality product identifiers become increasingly important.<\/p>\n On 16 October, a test of advertising in AI Mode for users with Personal Intelligence<\/a> began.<\/p>\n Context obtained from applications voluntarily connected by users may help personalise the experience. Google states that advertisers do not directly access these applications and that specific safeguards exist for sensitive information.<\/p>\n Even so, the test does not represent a new commercial targeting option available in every account. Its development should be monitored while clearly distinguishing the use of contextual signals from direct access to private information.<\/p>\n In Spain, I would not yet modify a campaign based on an experimental feature without general commercial availability.<\/p>\n On 15 October, Meta introduced Ads Super Burst, First Impression Ads and further expansions of Reels Trending Ads<\/a>.<\/p>\n In addition, the company is testing a system designed to evaluate cumulative attention across several video exposures.<\/p>\n Availability conditions vary. Some features are being rolled out progressively, while others remain limited to testing in the United States.<\/p>\n For small businesses focused on immediate lead generation, I would not replace lead or sales metrics with awareness indicators. On the other hand, for brand campaigns with the appropriate budget and objectives, these formats may be worth evaluating once they are genuinely available.<\/p>\n For SEO and GEO, I would begin by checking which branded and service-related queries trigger generative answers, which sources appear and in what context. I would also use Bing AI Performance when sufficient data is available and compare Direct and referral traffic on pages associated with ChatGPT, avoiding the assumption that a decline in referrals automatically means a fall in demand.<\/p>\n For Google Ads, I would check whether the AI Max combinations report and the new channel priorities are available. Within Demand Gen, I would review lookalike audience behaviour and, where appropriate, automatic cost adjustments. Finally, for analytics, I would verify Tag Manager connections and GA4 permissions; for retailers with physical locations, I would also review local Merchant Center data.<\/p>\n By contrast, Universal Cart, personalised advertising inside AI Mode and Meta’s new awareness formats would remain under observation while their availability is limited. The selection should always reflect the client’s actual needs. There is little value in spending time on a local inventory feature for a business without physical stores, or turning a US-based test into a commercial promise for Spain.<\/p>\n The most interesting aspect of this week is not simply the arrival of new artificial intelligence features. For me, what matters most is that the platforms themselves are recognising, even indirectly, that we need to understand their systems’ decisions more clearly.<\/p>\n Google Ads is beginning to show combinations of queries, ads and destinations. Bing provides information about the topics in which a website is being cited. At the same time, Google is trying to make the sources used in its generative answers more visible.<\/p>\n In my work as a consultant, I prefer a limited but clearly explained data point to a spectacular figure that we do not know how to interpret. The possible loss of ChatGPT referral attribution is a good example. A report may show fewer sessions coming from the assistant without those users necessarily having disappeared.<\/p>\n Something similar happens when an advertising system automatically expands audiences. We may record more conversions and still end up attracting lower-quality commercial opportunities. This is why automation needs context.<\/p>\n Before changing a campaign or a piece of content, we need to know which problem we are trying to solve. We may want to attract more qualified visits, reduce the cost per opportunity, improve brand reputation or make a purchase easier. Each objective requires different data and different decisions.<\/p>\n Without that prior definition, every automated recommendation can look like an opportunity and every movement in a chart can become an excuse to intervene.<\/p>\n The evolution from SEO towards GEO<\/a> requires exactly the same discipline. Appearing in a generative answer, receiving a click and gaining a customer are different phenomena. Therefore, each requires its own form of evaluation.<\/p>\n![\"SEO,](\"https:\/\/www.javiercarmonabenitez.es\/wp-content\/uploads\/2026\/10\/novedades-seo-sem-ia-19-octubre-202.jpeg\" "\\"SEO,")
<\/a><\/p>\nSEO and GEO news: more visibility inside AI-powered search engines<\/h2>\n
1. Google adds more visible links to AI Overviews and AI Mode<\/h3>\n
2. Bing Webmaster Tools introduces Topics Portfolio for AI Performance<\/h3>\n
3. ChatGPT may be hiding part of its referral traffic<\/h3>\n
Google Ads: greater control over automation, budgets and audiences<\/h2>\n
4. AI Max brings searches, ads and landing pages together in one report<\/h3>\n
5. Google Ads introduces channel priorities and new budget controls<\/h3>\n
6. Demand Gen changes the role of lookalike audiences<\/h3>\n
7. Demand Gen extends certain automatic cost adjustments to 60 days<\/h3>\n
8. Google Ads API v26 introduces Search Lift measurement<\/h3>\n
Google Analytics and Tag Manager: simplifying measurement without losing control<\/h2>\n
9. Tag Manager enables more direct measurement connections<\/h3>\n
10. Google Analytics introduces an API for managing user groups<\/h3>\n
Shopping, local SEO and artificial intelligence applied to commerce<\/h2>\n
11. Merchant Center adds physical store addresses to local inventory<\/h3>\n
12. Merchant Center expands product warranty information<\/h3>\n
13. Universal Cart begins rolling out in the United States<\/h3>\n
14. Google tests personalised advertising inside AI Mode<\/h3>\n
15. Meta announces new awareness and measurement formats<\/h3>\n
What I would review in client accounts this week<\/h2>\n
Javier Carmona Ben\u00edtez’s opinion on this week’s SEO, SEM and AI news<\/h2>\n
More automation requires better questions<\/h3>\n