Search Marketing Takeaways for Leaders: July 2026
Search is becoming increasingly involved in the decision itself. For many years, the role of search was relatively clear. A customer entered a query, reviewed a set of results and chose where to go next.
That journey is becoming less and less linear. Search platforms now summarise information, compare options, present business profiles, recommend products, insert advertisements and answer follow-up questions. They are increasingly influencing which sources are trusted and, in some cases, helping customers move closer to an action without requiring a conventional website visit.
Clearly this doesn’t mean search is disappearing, but rather that more of the customer’s decision is taking place within the search environment. That creates an important leadership question:
When platforms increasingly answer, recommend, compare and convert, how does a business remain visible, accurately represented and commercially measurable?
Here are the key search marketing takeaways for leaders this month.
1. Search demand is shifting, not collapsing
There is no shortage of headlines suggesting AI is bringing traditional search to an end. New analysis of more than one million high-volume keywords provides a more balanced picture.
The research found that 29% of search demand was declining, but a similar proportion was growing elsewhere. Overall demand remained broadly flat. People had not stopped searching, but the subjects, questions and types of intent attracting demand were changing.
The decline was not evenly distributed. Information-heavy categories faced greater pressure because AI systems can often answer straightforward questions without requiring several website visits. More commercially valuable behaviours, including branded, navigational, local, comparative and transactional searches, appeared more resilient.
Considering this distinction is important. A business may see lower traffic to general guides or informational articles while maintaining, or even increasing, demand around products, services, locations and purchase decisions. Looking only at total organic sessions could make that appear to be an overall decline when it is really a redistribution of demand.
It may also change the role of informational content. Some content will continue to attract traffic. Some will shape AI answers without receiving the same number of visits. Other pages may no longer justify the investment required to create and maintain them.
Leadership takeaway:
Avoid making strategy decisions based on broad claims that search is growing or declining. Review demand within your own market and separate:
Informational searches
Commercial and comparison searches
Branded and non-brand demand
Local searches
Product or service searches
Navigational and transactional intent
The important question is not simply whether search demand remains. It is whether the business is visible where commercially relevant demand is moving.
2. AI search advertising is becoming real, but visibility is splitting into separate systems
Advertising inside AI-led search experiences is moving beyond small-scale experimentation. An SE Ranking study of 50,032 US commercial searches found text advertisements on 29.45% of Google AI Mode queries. Ads were particularly common on expensive commercial searches, appearing on 53.56% of keywords with estimated costs per click of $10 or more.
This suggests Google is already applying its established advertising infrastructure to commercially valuable AI searches. However, the research also found that buying an advertisement did not make the advertiser substantially more likely to be cited within the AI-generated answer.
Only 11.53% of advertising domains appeared among the cited sources for the same searches. At an individual URL level, the overlap was just 1.95%. Around 85% of advertisers did not appear in the conventional organic results for the keywords on which their AI Mode ads were shown.
Paid placement, organic ranking and AI citation are therefore becoming three distinct forms of search visibility. A business may:
Pay to appear beside an AI answer
Rank organically below it
Be cited as evidence within it
Be recommended without receiving a citation
Or be absent from one surface while appearing in another
ChatGPT advertising remains at an earlier stage. OpenAI says users are dismissing advertisements 50% less frequently than when its pilot began, which it sees as an early relevance signal. However, eMarketer estimates that the entire US standalone-chatbot advertising market will reach only $5.41 billion by 2030, far below OpenAI’s reported longer-term advertising ambitions.
Basically, scale, economics and customer response remain uncertain.
Leadership takeaway:
Do not treat search visibility as one combined measure.
Paid placement, organic ranking, AI citation and AI recommendation should be monitored separately. Each has a different role and may produce different outcomes.
Conversational advertising is worth testing where the audience and intent are relevant, but significant budget changes should follow evidence on reach, customer quality, conversion and profitability rather than early engagement signals alone.
3. Google-owned profiles and product panels are becoming the customer’s first page
For many searches, a company website is no longer the first substantial representation of the business that a customer sees.
Research from Profound found that citations to Google.com within AI Mode increased 8.4 times over approximately two months, making it the second-most cited domain in its tracking.
The increase came mainly from Google Business Profiles and Product Knowledge Panels being inserted directly into local and shopping answers. These panels can show opening hours, location, photographs, reviews and product information before a customer reaches the organisation’s own website.
This is especially significant for businesses operating in areas such as:
Hospitality and tourism
Restaurants and leisure
Property and home services
Healthcare
Automotive
Retail
Local professional services
Multi-location organisations
A customer can now form an initial opinion, compare alternatives, read reviews, check availability and sometimes take an action using information hosted by Google.
The same principle applies to ecommerce. Product names, descriptions, imagery, prices, reviews, availability and delivery information increasingly feed search, shopping and AI-led recommendation experiences. Incomplete or inconsistent product data can affect visibility before the website has an opportunity to convert the customer.
These are not technically owned channels. The platform controls the experience, rules and visibility. But the information within them needs to be governed with the same care as the company website.
Leadership takeaway:
Treat Google Business Profile, Merchant Center data, product panels and other platform-hosted profiles as part of the organisation’s core digital estate.
Assign clear ownership for:
Accuracy
Opening hours and locations
Product and service information
Imagery
Reviews and responses
Availability
Offers
Booking or purchasing links
Monitoring unauthorised changes
For many customers, the first page representing the brand may no longer belong to the brand.
4. Search measurement is expanding beyond websites, keywords and traditional results
Search Console has historically focused mainly on how websites perform within Google Search and that boundary is beginning to widen.
Google expanded access to its generative-AI performance reports during June. These reports provide greater visibility into appearances across AI Overviews, AI Mode and generative experiences in Discover, although access remains incremental and the available reporting does not yet provide a complete view of clicks, queries or commercial outcomes.
Google then introduced a new type of Search Console property for social and video platforms.Organisations and creators can connect verified Instagram, TikTok, X and YouTube accounts and see how their posts perform across Google Search and Discover. This includes the search terms generating visibility, impressions, clicks and engagement with individual posts.
This is a meaningful change that acknowledges that searchable content does not only sit on a company website. A video, social post, creator collaboration or short-form clip may be discovered through Google and contribute to a customer journey normally attributed to another channel.
YouTube is also becoming more conversational. Ask YouTube expanded to a larger group of signed-in US desktop users in July. People can ask natural-language questions and receive responses drawing from long-form videos, Shorts and follow-up prompts. YouTube has advised creators to use clear titles, chapters and well-defined sections that answer specific questions.
This further weakens the distinction between search, social and video. A customer may find a social post through Google, question a video through YouTube’s conversational interface and then complete further research through an AI answer, all without following a conventional website-first journey.
Leadership takeaway:
Search reporting should start to cover the organisation’s wider searchable content estate. That may include:
Website visibility
AI answers and citations
Google Business Profiles
Product data
YouTube videos
Social posts
Images
Reviews
Third-party publications
AI-referred traffic
However, visibility and commercial contribution are not the same thing. New platform metrics should support better decisions, not create a larger reporting dashboard with no connection to business outcomes.
5. ChatGPT dominates measurable AI referrals, but referral traffic tells only part of the story
A study analysing 6.77 million sessions referred by large language models found that ChatGPT generated 92.4% of the measurable referral traffic in its dataset. Monthly LLM-referred sessions increased almost tenfold during the period covered by the study, reaching more than 644,000 in May 2026. That concentration makes ChatGPT important for organisations monitoring AI-referred website activity. That being said, the headline needs careful interpretation.
It doesn’t mean ChatGPT accounts for 92.4% of all AI usage, customer influence or AI-led discovery. It means ChatGPT produced that share of externally trackable website referrals within this specific dataset. A customer may use an AI assistant to:
Research a category
Compare providers
Shortlist products
Understand terminology
Validate a decision
Find reviews
Identify questions to ask
None of those activities necessarily generates a click that appears in analytics. The research also demonstrated how volatile referral traffic can be. ChatGPT referrals in the dataset fell by around half during one month before recovering, apparently following changes in model and citation behaviour.
Businesses therefore face two measurement challenges. First, referrals represent only the visible part of AI influence. Second, the traffic that can be measured may change substantially when a platform adjusts its models, interface or preferred sources.
Leadership takeaway:
Track AI-referred sessions, but do not present them as a complete measure of AI search performance. Review:
Referral traffic by platform
Landing pages
Engagement
Conversion rates
Lead or customer quality
Revenue per session
Assisted conversions
Branded search changes
AI mentions and citations
Avoid creating forecasts that assume today’s referral behaviour will remain stable. A platform decision can materially alter traffic without any change to the quality of the organisation’s marketing.
6. Websites need to be understandable to AI agents as well as people
Websites have traditionally been designed for two audiences: people and search-engine crawlers. AI agents are becoming a third.
Unlike a conventional search crawler, an agent may be asked to complete a specific task. It could compare suppliers, check whether a product integrates with another system, understand security standards or establish what a service is likely to cost.
Research involving 100 B2B product websites found that agents could usually retrieve integration and security information from official sources, but had greater difficulty with pricing and feature details.
Pricing and features produced 77% of all third-party citations in the research. Agents frequently left the company’s own website and relied on directories, review platforms or other external sources because the first-party information was unavailable or difficult to extract.
The problem was not limited to organisations hiding prices. Even where a numerical price was publicly shown, agents still cited at least one third-party source in 18% of the tests. Common barriers included:
Information dependent on JavaScript
Interactive calculators
Complex pricing tables
Important details contained in images or screenshots
Information locked inside PDFs
Access restrictions
Ambiguous language
Vague “contact sales” messaging
A page can be visually impressive and persuasive to a human while remaining difficult for an agent to interpret confidently.
This creates a commercial risk - when the official website does not provide an answer that can be easily retrieved and cited, an agent may use an outdated review, an unofficial directory or a competitor comparison instead.
Leadership takeaway:
Machine readability is becoming part of commercial clarity. Review whether an AI agent can reliably establish:
What the business offers
Who it is for
What features or services are included
How pricing works
Which systems or products it integrates with
What evidence supports its claims
What limitations apply
What the customer should do next
Some apparent SEO problems are actually proposition, packaging and communication problems. Greater clarity can help customers, search engines and AI agents simultaneously.
7. AI search is creating new content and reputation risks
Traditional online reputation management often focused on what appeared prominently on the first page of Google. AI answers do not always follow that same hierarchy. An older article may no longer rank highly in conventional results but still be retrieved and cited by an AI system when it considers the source authoritative or relevant.
Search Engine Land highlighted examples in which old negative coverage reappeared in AI-generated answers years after the underlying issue had been resolved. A story does not need to rank first to influence how an AI system describes a company.
That changes the reputation challenge. Publishing newer, more positive content may not be enough to suppress an older source. Organisations need credible, current evidence that helps AI systems understand what happened subsequently and how the business should be represented today.
There is also a risk within brands’ own content. Research by Lily Ray found Google AI Overviews citing self-promotional “best software” listicles created by vendors while excluding the publishing vendor from the recommendation in 69% of the cases studied.
The business had structured and published evidence about its competitors, but Google used that information without providing the intended promotional benefit. This is a useful reminder that businesses do not fully control how search and AI platforms interpret or reuse published material. A comparison page written primarily to influence rankings may become an evidence source for recommending someone else.
Leadership takeaway:
Extend reputation monitoring beyond traditional rankings and review sites. Monitor:
How AI platforms describe the business
Which sources they cite
Whether old or inaccurate stories are resurfacing
How competitors are represented
Whether the organisation’s own comparison content is being interpreted as intended
Comparison content should have a genuine customer purpose and be grounded in fair, supportable evidence.
Where outdated information remains influential, publish credible and current material across authoritative first-party and third-party sources. The aim is not to hide the past, but to make the complete and current position easier to understand.
8. Google’s spam update is a useful prompt to review the content estate
Google released its June 2026 spam update on 24 June. The global update applied across all languages and completed on 26 June.
They didn’t identify one particular tactic, website type or form of content as the sole target. That makes it unhelpful to assume every visibility change had one simple cause or to search for a quick technical adjustment. A better response is to use the update as a review point.
Many organisations have accumulated large content estates over several years. Pages may have been created to target small keyword variations, support old campaigns, meet an historic publishing schedule or respond to SEO practices that are no longer useful.
Generative AI is making content production cheaper, which increases the temptation to create even more. But volume creates cost and risk. Every page needs to be crawled, maintained, reviewed and kept accurate. Weak or repetitive content can make it harder for users and platforms to understand which information is important.
Useful areas to review include:
Scaled or generic articles
Repetitive location or service pages
Pages targeting minimal keyword variations
Lightly rewritten or republished material
Outdated advice
Thin programmatic pages
Manipulative linking
Content receiving no meaningful demand
Several pages competing to answer the same question
Leadership takeaway:
Content strategy should not be measured by how much is published.
Give search and content teams permission to improve, merge, redirect or remove material where that creates a clearer and more useful digital estate.
The objective is not to have the largest possible website. It is to have the strongest set of pages for the customers, questions and decisions that matter.
9. Performance Max is gaining more visibility and potentially more control
Google continues to move paid media towards greater automation, but advertisers are also receiving more information about how that automation is working.
Channel Diagnostics was added to Performance Max in July. The feature provides a central view of missing or disapproved assets that may prevent campaigns from serving across Search, Display, YouTube, Discover, Gmail and Maps. Advertisers can review all channels together or examine individual channels and asset requirements.
An automated campaign may appear active while being unable to serve properly across parts of Google’s inventory because it lacks the correct images, copy, video or approvals.
Google is also testing a new Partners setting within a limited alpha. This may allow participating advertisers to opt in or out of Search Partners and the Google Display Network within Performance Max, providing a level of network control that has not generally been available in the campaign type.
It is important not to overstate that second development. It is an early test, not a control currently available to every advertiser. But it indicates that Google recognises continued demand for greater understanding and influence over where Performance Max budgets are deployed.
The broader lesson remains the same - automation does not remove the need for management. It changes the work from manually controlling every placement towards setting better goals, supplying stronger inputs and checking whether the system is creating commercially useful outcomes.
Leadership takeaway:
Ask paid-media teams and partners to explain:
Where campaigns are eligible to run
Which channels actually receive spend
Whether missing assets restrict delivery
How product, audience and conversion signals are supplied
Which controls are available
How low-quality inventory or outcomes are identified
Whether channel reporting connects to profit, lead quality or revenue
Automation should reduce unnecessary manual work. It should not reduce commercial scrutiny.
10. Budget-constrained bidding targets require action before 17 August
One of the most immediately actionable changes this month relates to Google Ads campaigns using Target CPA or Target ROAS. From 17 August 2026, campaigns that are limited by budget will optimise more consistently towards the target entered in the account.
Google explains that a campaign with a Target CPA of $10 but a recent actual CPA of $5 may begin delivering closer to the stated $10 target after the change. The platform will assume that the entered target represents the advertiser’s real commercial objective.
Previously, some budget-constrained campaigns substantially outperformed their stated targets. That may have allowed old or loosely considered settings to remain in place without an obvious negative effect. After the change, those settings may influence performance more directly.
Google made a Bid Target Adjustment Tool available from 6 July to help advertisers review affected campaigns. The update applies specifically to target-based campaigns that are limited by budget. Campaigns using Target CPA or Target ROAS without a budget constraint are not expected to change in the same way. This is a good example of a small-looking platform update having a potentially material commercial impact. A historic account setting can become a direct instruction to an automated system.
Leadership takeaway:
Review affected campaigns before 17 August.
Confirm that Target CPA and Target ROAS settings reflect:
Current gross margin
Lead quality
Conversion value
Customer lifetime value
Capacity
Growth objectives
Acceptable payback
Recent performance
The answer is not automatically to lower every target or increase every budget.
It is to make sure the number entered into the platform represents a deliberate business decision rather than an inherited or outdated setting.
11. Paid-search performance increasingly depends on first-party business data and the wider customer journey
Google Ads has introduced a beta that allows businesses to supplement website conversion tags with data from backend systems.
Advertisers can connect CRM, ecommerce, order or transaction data to an existing website conversion action. Google can then use transaction identifiers to deduplicate the information and create a more complete measurement signal. This can help recover conversions missed because of browser restrictions, privacy settings, ad blockers or gaps in front-end tracking. More importantly, it brings advertising optimisation closer to actual business outcomes.
A website form submission may be easy to measure but commercially weak. A CRM may reveal that the same lead became a qualified opportunity, sale, repeat customer or high-value account.
Automated bidding becomes more useful when it can optimise towards those deeper outcomes rather than simply the easiest conversion to generate. This connects to a wider argument about paid-search costs. Recent practitioner analysis has argued that cost-per-click inflation does not begin only inside the auction. Weak brand awareness, limited organic visibility, poor landing pages, low conversion rates, weak lead handling and poor customer retention can all reduce the value created from paid-search activity.
A media team cannot solve all of those problems through bids and targeting. A business may pay a competitive market price for a click but still achieve poor economics because:
The proposition is unclear
Competitors are better known
The landing page is weak
The customer journey is difficult
Sales follow-up is slow
Conversion tracking rewards low-value actions
Retention is poor
Customer value is not passed back to the platform
Leadership takeaway:
Paid search is increasingly a business-data and customer-journey issue, not only a media-management issue.
Marketing, ecommerce, sales, CRM and finance teams need shared definitions of:
A conversion
A qualified lead
A valuable customer
Revenue
Margin
Repeat value
An acceptable acquisition cost
Better bidding depends on better signals. Better economics depend on what happens before and after the click as well as inside the advertising platform.
12. Search regulation is moving from fines towards structural change and accountability
Regulation is beginning to influence not only what search platforms are allowed to do, but how their underlying advantages may need to be shared. The European Commission has required Google to provide eligible AI and search competitors with access to anonymised data used to optimise its search services. The requirement is expected to take effect from January 2027.
Google will also need to open 11 Android capabilities to competing AI assistants from July 2027, allowing eligible alternatives to complete functions such as voice-activated searches and bookings. Google has raised privacy and security concerns about the measures. The intention is to make it easier for alternative search engines and AI assistants to compete with Google and Gemini.
A separate development in Germany concerns accountability for AI-generated answers. The country’s media regulator has said Google AI Overviews and services such as Perplexity should be treated as content providers under German media law, rather than simply neutral platforms displaying third-party material. The position follows a court finding that Google could potentially be directly responsible for inaccurate claims generated within an AI Overview. Google plans to appeal.
Both developments are significant. Search platforms have historically benefited from being treated partly as intermediaries between users and published information. AI-generated answers make that distinction harder to maintain because the platform is selecting, interpreting and writing the response.
For businesses, this raises practical questions. If an AI answer contains inaccurate information about a company, who is responsible? How can it be challenged? What evidence is required? And will regulation make it easier for alternative assistants to enter the customer journey?
Leadership takeaway:
Search strategy should account for platform, legal and regulatory change as well as customer behaviour. Monitor:
Which assistants and search platforms matter to customers
How the business is represented in generated answers
Processes for challenging inaccurate information
Dependence on individual platforms
Regional differences in search products
Changes to operating-system defaults and data access
Search competition may increasingly be shaped by regulation and infrastructure, not only the quality of individual products.
Leadership Perspective
Search is no longer simply the route to a website, it is increasingly becoming the layer in which the customer’s need is interpreted, the available evidence is selected and the decision is shaped.
That changes the role of search marketing. Rankings and clicks still matter. Paid-search efficiency still matters. Technical SEO, content quality, product feeds, landing pages and conversion tracking remain important. But they now sit within a broader system.
Businesses need to consider:
How demand is changing by intent
Whether the brand appears in AI answers
Which sources are cited
How platform-hosted profiles represent the business
Whether videos and social content are searchable
Whether AI agents can understand the website
How paid placement interacts with organic and cited visibility
What happens when more decisions occur without a website visit
Whether measurement connects to real commercial outcomes
This does not require abandoning established search practice. In many cases, the fundamentals become more valuable:
Clear propositions
Useful and distinctive content
Accurate product and business data
Strong technical accessibility
Credible proof
Good reviews
Clear pricing and service information
Reliable conversion measurement
Strong landing-page experiences
Appropriate human oversight
The difference is that those foundations now support more surfaces.
A clear service page can help traditional rankings, AI retrieval, paid-search matching, customer conversion and agent-led research.
A well-managed business profile can influence local search, maps, AI Mode and the customer’s first impression.
Accurate CRM and transaction data can improve measurement, bidding and leadership confidence.
Search is becoming the decision layer, but businesses still determine the quality of the information and commercial signals that feed it.
Leader’s Checklist
A few useful questions for leadership and marketing teams this month:
Are we reviewing search demand by intent rather than relying on total traffic?
Do we understand which informational searches are declining and which commercial searches are growing?
Are paid placement, organic ranking and AI citation reported separately?
Have we reviewed whether AI Mode advertising is relevant to our market?
Are Google Business Profiles and product panels accurate, complete and actively managed?
Have we checked whether the new Search Console AI reports are available to us?
Can we connect relevant social and video accounts as Search Console platform properties?
Are we tracking AI-referred traffic without treating it as the whole picture?
Can an AI agent easily understand our proposition, features, pricing and evidence?
Are old or inaccurate stories influencing how AI systems describe the business?
Does our comparison content genuinely help customers, or could it unintentionally promote competitors?
Have we reviewed the quality and continued purpose of our wider content estate?
Are missing assets limiting Performance Max delivery?
Do we understand where Performance Max budget is being deployed?
Have all budget-constrained Target CPA and Target ROAS campaigns been reviewed before 17 August?
Are paid-media platforms receiving meaningful first-party conversion and customer-value data?
Are media performance, landing-page experience, sales follow-up and customer value being considered together?
Do we have a process for challenging inaccurate AI-generated search answers?
Final Thought
The leadership question is no longer only:
Where do we rank?
It is becoming:
Where and how is the customer’s decision being made, and what role does our business play within it?
Search remains one of the most commercially important parts of the customer journey. But the journey is increasingly being interpreted, summarised and shaped before the customer reaches a website.
The strongest businesses will not respond by chasing every new surface or acronym. They will build clearer information, stronger evidence, more reliable data and better-connected customer journeys that help people, platforms and AI systems understand why the business deserves to be considered. That is what visibility increasingly requires.
Useful links
Search Engine Land: What 1 million keywords reveal about AI’s impact on search:
https://searchengineland.com/what-1-million-keywords-reveal-about-ais-impact-on-search-481474
Search Engine Land: Google AI Mode ads reach nearly 30% of queries:
https://searchengineland.com/google-ai-mode-ads-reach-queries-study-482475
Search Engine Land: OpenAI says ChatGPT ad dismissals have dropped 50% as relevance improves:
https://searchengineland.com/openai-says-chatgpt-ad-dismissals-have-dropped-50-as-relevance-improves-480991
Search Engine Land: OpenAI’s ChatGPT ads could miss $100 billion revenue target:
https://searchengineland.com/openai-chatgpt-ads-100-billion-revenue-target-482365
Search Engine Land: Google is AI Mode’s second-most-cited domain:
https://searchengineland.com/ai-mode-cites-google-report-482463
Search Engine Land: Google Search Console AI performance reports rolling out to more users:
https://searchengineland.com/google-search-console-ai-performance-reports-rolling-out-to-more-users-480867
Google Search Central: See how content from social and video platforms performs on Google:
https://developers.google.com/search/blog/2026/07/search-console-social-video-platforms
Search Engine Land: Ask YouTube AI search experience expands to US desktop users:
https://searchengineland.com/ask-youtube-expands-481906
Search Engine Land: ChatGPT commands 92% of AI referral traffic:
https://searchengineland.com/chatgpt-ai-referral-traffic-sessions-data-481630
Search Engine Land: Where AI agents get stuck on your site:
https://searchengineland.com/stuck-ai-agents-482344
Search Engine Land: How AI search gives old negative content new life:
https://searchengineland.com/ai-search-old-negative-content-new-life-482117
Search Engine Land: Google AI Overviews cite self-serving listicles but recommend competitors:
https://searchengineland.com/google-ai-overviews-cite-self-serving-listicles-recommend-competitors-480573
Google Search Status Dashboard: June 2026 spam update:
https://status.search.google.com/incidents/YUX1peHev5a4fkxLDiUQ
Search Engine Land: Google adds Channel Diagnostics to Performance Max:
https://searchengineland.com/google-adds-channel-diagnostics-to-performance-max-481445
Search Engine Land: Google tests Performance Max network controls with new Partners setting:
https://searchengineland.com/google-tests-performance-max-network-controls-with-new-partners-alpha-setting-482469
Google Ads Help: Changes to target-based bid strategies:
https://support.google.com/google-ads/answer/17061251?hl=en
Google Ads Help: Frequently asked questions about changes to target-based bidding:
https://support.google.com/google-ads/answer/17125145?hl=en
Search Engine Land: Google Ads launches beta for supplemental conversion data:
https://searchengineland.com/google-ads-launches-beta-for-supplemental-conversion-data-480606
Search Engine Land: Why CPC inflation starts before the auction:
https://searchengineland.com/cpc-inflation-starts-before-auction-482381
Reuters: Google required to open up to AI and search-engine rivals under EU-mandated changes:
https://www.reuters.com/world/google-required-open-up-ai-search-engine-rivals-under-eu-mandated-changes-2026-07-16/
Reuters: German media regulator says Google’s AI Overviews are subject to German media law:
https://www.reuters.com/legal/government/german-media-regulator-says-googles-ai-overviews-subject-german-media-law-2026-07-14/
