Paid Media News (Week 39)
Week 39: campaign changes, commerce workflows and practical reading
The 21–27 September edition separates dated announcements and observations from analysis and implementation documentation. Older references retain their original source dates. Follow each link to the relevant news story, guide or resource.
A digest of the original edition. Updates and availability reflect what was reported at the time.
Demand Gen adds a Business Agent, one-click experiences and Maps placements
Google’s September Demand Gen update adds Business Agent for YouTube Ads: conversational product and brand assistance alongside video ads with product feeds, with a sign-up route. It also introduces one-click image experiences in Shorts and Gmail, plus Maps Promoted pins that use affiliate location assets.
Read this update in detail →Demand Gen’s cost-adjustment beta targets the learning period
Google documents a cost-adjustment beta for eligible Demand Gen campaigns using target CPA or target ROAS. Eligibility and the adjustment calculation matter: this is not a blanket refund for campaigns that miss a target.
Read this update in detail →AI Brief adds languages; unified AI Max reporting is still ahead
Google has announced support for seven additional languages in its closed beta of AI Brief. It also describes a unified view of search terms, creative and landing pages planned for later this year.
Read this update in detail →Shopify’s Google checkout changes where measurement happens
Shopify documents direct Google checkout for eligible US stores selling to US customers, with rollout and product restrictions. For eligible stores, the setting can be enabled by default.
Read this update in detail →AI Max makes branded and unbranded options more explicit
Bia Camargo’s account observation shows clearer options for all, branded and unbranded searches in AI Max. This is a change in the interface she observed, not evidence that a new underlying targeting capability has reached every account.
Read this update in detail →A budget-reduction warning forecasts the potential trade-off
Thomas Eccel reports a forecast warning when reducing a Google Ads budget. Treat it as an estimate to examine alongside marginal returns, lead quality and cash-flow constraints.
Read this update in detail →Local Services Ads are taking up more space in an observed layout
Anthony Higman has documented a larger Local Services Ads presentation. The immediate relevance is visibility: a larger unit changes the search page around local businesses.
Read this update in detail →Merchant Centre’s AI performance tab reaches an observed account
Arpan Banerjee reports access to Merchant Centre’s AI performance tab. Google announced the direction in May; this is a rollout sighting, not a new product launch, in September. Current documentation covers organic AI traffic, with defined market and language availability, but not paid ad performance. Keep those metrics separate from Google Ads results when discussing visibility with clients.
Read this update in detail →AI Mode’s product layouts remain a watchlist item
Reported examples place sponsored and organic products within AI Mode’s shopping experience. Treat the screenshots as observations of particular searches, rather than a new buying option or a promise of placement. Review feed quality and the distinction between paid and organic reporting; the layout alone does not indicate additional campaign controls or a measurable shift in traffic.
Read this update in detail →AI Mode ads can read like the next step in a conversation
Venkat’s search examples show sponsored recommendations framed as conversational follow-ups. The useful creative question is whether an ad makes sense after the user’s preceding query, rather than only beside a conventional keyword result.
Read this update in detail →Google tests darker mobile sitelinks with arrows
Brodie Clark documents a mobile text-ad layout with darker sitelinks, a left-hand rule and arrows beside each link. His examples show a presentation test, not a new asset setting advertisers can switch on.
Read this update in detail →Ask Advisor can propose campaign changes for your approval
Google’s documentation describes an eligible-account beta that can propose actions including campaign pauses, budget and location changes, and appeals. The advertiser approves the action; it is not an unrestricted autonomous account manager.
Read this update in detail →Investment recommendations: distinguish reallocation from extra spend
Google Ads coverage describes Holistic recommendations that may reallocate or add budget, and Growth recommendations that add budget. Read the proposed changes at campaign level before accepting them.
Read this update in detail →Smart Bidding learning: use conversion cycles, not a fixed rule
Google’s current guidance says learning typically takes one to two conversion cycles. It does not support the circulated claim of a universal 50-conversion, three-cycle requirement.
Read this update in detail →AI Max versus DSA: choose an experiment that fits the campaign
Google’s experiment guidance distinguishes campaign configurations. The template cannot be used for a campaign made entirely of Dynamic Search Ads; in a mixed campaign, the treatment leaves DSA ad groups unchanged.
Read this update in detail →More budget does not automatically create more qualified demand
Pauline Jakober’s guide examines why additional PPC spend can fail to produce additional leads. Use it to distinguish between limited budget, limited demand, weak query relevance, and poor conversion quality.
Read this update in detail →Use excluded destinations deliberately in Merchant Centre
Sarah Vlietstra’s guide explains selective destination exclusions in product data. The practical task is to decide where a product should remain eligible, then validate the submitted attributes against Google’s current specification.
Read this update in detail →CSS product pages have rating and review-count attributes
Google’s specification documents rating and review-count attributes for Comparison Shopping Service product pages. Check the CSS-specific scope before adding them to a feed workflow.
Read this update in detail →Local call tracking: check the number actually used
Reporting attributed to Anthony Higman highlights calls using a Business Profile number in local ad experiences. Compare the displayed number, the destination number, and the reporting path in your own setup.
Read this update in detail →Demand Gen on the Display Network: understand CPM billing
Google’s current FAQ confirms that Demand Gen uses CPM billing on the Google Display Network. Keep the billing basis separate from the campaign’s bidding objective when reconciling spend and results.
Read this update in detail →YouTube’s Performance Four is a campaign-planning reference
YouTube for Business’s framework offers a structured way to review performance-oriented video activity. Use it to examine campaign inputs and creative hypotheses alongside the Demand Gen changes in this edition.
Read this update in detail →The Leads beta is an existing workflow to inspect
Emirhan Bayutmuş’s post resurfaces Google Ads’ Leads section. Earlier coverage exists, so this is a reference rather than a confirmed new Week 39 launch. Review the available qualification and conversion workflow against your CRM definitions before relying on it for lead-quality reporting.
Read this update in detail →Message assets connect Search and PMax with messaging apps
Google documents message assets for Search and Performance Max, including WhatsApp and other supported messaging destinations. After an asset is approved, a message-lead conversion goal becomes available; that does not mean it automatically becomes the account’s default optimisation goal.
Read this update in detail →Business Profile suggestions can have a four-day review window
Google documents a four-day review period for certain notified suggestions. That timing does not apply to every profile edit, and it is not a promise of advance notice for every change.
Read this update in detail →Merchant Centre’s impressum guidance has a country-specific scope
Google’s platform guidance distinguishes between hosted impressum availability in Germany and Austria and in Switzerland. Check the requirement relevant to the merchant and destination.
Read this update in detail →Google Ads API v25.2 adds options for tool owners
The v25.2 announcement covers asset-group tracking URL options, benchmark percentiles and Performance Max draft generation. Teams maintaining reporting or campaign-creation tools should review the relevant fields and supported workflows. These are API capabilities, not a promise that identical controls have appeared in every advertiser interface.
Read this update in detail →Google updates its list of advertising self-regulatory codes
Google’s 22 September notice adds references to codes in Argentina, Australia, Chile, Colombia, Paraguay and South Africa. It reiterates that advertisers must comply with applicable local requirements.
Read this update in detail →A Member price label appears in Bing Shopping
Sachin Patel’s screenshot documents a Member price label in Shopping ad results. His explanation of eligibility is an interpretation, not an official product specification.
Read this update in detail →Bing Shopping favicons: a dated account observation
Sachin Patel reported favicons beside retailer names in Bing Shopping ads on 18 September. Later coverage resurfaced the screenshots this week.
Read this update in detail →Microsoft experiments cover Search, Shopping, Audience and PMax
Microsoft’s current help confirms optimisation experiments for four campaign types. Shared and lifetime budgets, and campaigns with DSA enabled, are ineligible.
Read this update in detail →Microsoft Max CPC: distinguish portfolio controls from API changes
Portfolio bid-strategy coverage and Microsoft’s MaxCpc API documentation concern related but different changes. The API specifies 12 January 2027 for the affected new campaigns and existing campaigns not already using MaxCpc; this should not be conflated with reported October changes to conversion-based strategies.
Read this update in detail →Copilot audience generation remains a scoped beta
Microsoft’s help describes a closed beta for eligible Audience campaigns in the US, Canada, UK and Australia. It does not apply to Search or Performance Max, excludes LinkedIn audiences, and generated audiences cannot be shared across accounts.
Read this update in detail →Separate Search and Shopping in Microsoft’s Competition view
Microsoft documents a Search/Shopping filter at account level and within Performance Max. Use it to avoid drawing one competitive conclusion from two different inventory contexts.
Read this update in detail →ChatGPT Ads begins expansion into seven Asian markets
OpenAI’s 23 September announcement starts the rollout to Indonesia, Malaysia, the Philippines, Singapore, Thailand, Vietnam and Taiwan. Eligibility and account access still govern participation.
Read this update in detail →ChatGPT Ads shows a “Limited by budget” campaign status
Anthony Higman’s Ads Manager screenshot shows a “Limited by budget” status and a “Campaign budget reached” message, with a prompt to increase the campaign budget to continue serving. This is an account observation in the beta interface.
Read this update in detail →GA4 hostname filters help define which sites belong in the data
Google documents include filters for approved hostnames. This can help separate legitimate site traffic from unwanted collection, but Measurement Protocol exceptions and empty-hostname behaviour need attention. Build the allowlist from the actual production and supporting domains, then validate the resulting data before relying on it for reporting. Filtering is not a substitute for investigating why unwanted events arrived.
Read this update in detail →Microsoft Clarity groups bot activity by page type
Microsoft Clarity has added Page Classification to its Bot Activity dashboard. It combines content-based classification with URL rules to group activity across page types, such as product, blog, and contact pages.
Read this update in detail →Shopify changes the session baseline, not the order total
Shopify’s session-measurement update can change session-based metrics while leaving orders, sales and customer counts unchanged. Annotate the change in reporting before interpreting a movement in conversion rate as a marketing gain or loss.
Read this update in detail →Search Console separates text-based and multimodal queries
Google’s documentation distinguishes text-based web queries from multimodal image queries. This gives search teams another way to examine how people discover a site.
Read this update in detail →Customer Match adds IP and interaction-time fields, with restrictions
Google’s documentation supports IP addresses and interaction timestamps in Customer Match uploads. A timestamp cannot be provided without an IP address, and IP matching is not supported for end users in the EEA, the UK, and Switzerland.
Read this update in detail →Offline uploads can supplement missing conversion user data
Google’s multi-source guidance describes how to supplement missing user data through offline uploads. Deduplication operates within a conversion action, so event definitions and action mapping still matter. The documented workflow includes a 24-hour recommendation and a 14-day trial before the additional source becomes biddable. Review reconciliation and coverage before interpreting additional recorded conversions as additional business generated.
Read this update in detail →Meta’s Creative Health panel rates diversity across ad sets
Peter Quadrel documents Creative Health ratings of High, Medium, Low, or Unknown for ad-set creative diversity, along with creative-fatigue indicators and imagery from similar businesses. Use the panel to identify where a creative review may be useful, then assess format, message and visual variation against business results.
Read this update in detail →Meta lead-form observations cover corrections and appointments
Gupta Media’s agency-summit coverage reports email-domain correction and appointment booking from the thank-you page of an instant form, including Calendly, HubSpot Meetings and HighLevel. These are reported capabilities whose availability needs to be checked in your account.
Read this update in detail →Meta One: compare the paid-plan features with your actual needs
Jon Loomer’s comparison of Meta One business plans is a reference for evaluating account tiers and their conditions. Separate the features your team would use from the subscription’s broader positioning.
Read this update in detail →Threads ads are loosening their dependency on Instagram identity
Social Media Today reports Meta’s move to allow advertisers to run Threads activity without the previous Instagram identity dependency. Check the identity and permissions offered in your account before changing an established setup.
Read this update in detail →Threads analytics: an older update worth separating from ad results
Earlier coverage describes changes to Threads profile analytics. Treat the material as an organic-account reference and keep its metrics separate from paid campaign reporting.
Read this update in detail →Pinterest’s optimisation score is a recommendations signal
Pinterest describes a campaign optimisation score based on recommended practices. Use the score to inspect the underlying actions, not as a forecast of ROAS or proof of campaign quality.
Read this update in detail →Verified LinkedIn Pages can tackle false employee associations
Oscar Rodriguez’s announcement describes tools for verified Pages to remove people falsely associating themselves with a business. This is relevant brand and identity hygiene.
Read this update in detail →Thought Leader Ads: plan the permission and distribution workflow
LinkedIn’s guide remains a useful reference for promoting employee or expert content. Decide which audience and business outcome the post should serve, secure the relevant author permissions and separate the value of the original content from the paid distribution. This is resurfaced guidance rather than a new advertising launch.
Read this update in detail →Metricool compares Meta and TikTok campaign benchmarks
Metricool’s study covers 628,969 campaigns and compares January–June 2025 with the same period in 2026. Treat the figures as context from its advertiser sample.
Read this update in detail →Short-drama research offers a creative hypothesis
Snap and Ipsos examine interest in short dramas as a content format. The research can inform experiments with recurring characters, episodic storytelling and compact narrative arcs.
Read this update in detail →Creator partnerships need a relationship model as well as a media plan
Victoria McNally’s interview with Jennifer Quigley-Jones explores how creator marketing is developing. Use it to consider briefs, ongoing relationships and performance objectives before treating creators as interchangeable ad assets.
Read this update in detail →Amazon DSP adds Belgium, Poland, Singapore, Egypt and South Africa
Amazon’s 25 September announcement opens local DSP campaign creation in five additional markets. Self-service advertisers and agencies can buy display, video and audio across Amazon properties and the open internet.
Read this update in detail →Amazon’s sponsored-ads tools have three different release schedules
Amazon announced expanded Ads data manager access for sponsored ads, now available globally. Its campaign-intelligence dashboard is scheduled for US advertisers in October, while automatic optimisation is scheduled for the US in January 2027.
Read this update in detail →Amazon SQP can indicate visibility bands, but not replace rank tracking
Alexander Swade’s study analyses six US brands across 44 weeks, covering 61,098 keyword-weeks. Its findings support broad visibility comparisons more than precise week-to-week rank tracking; advertising also complicates the relationship.
Read this update in detail →Reserve Share of Voice is an existing branded-search option
Amazon’s Reserve Share of Voice documentation describes how to reserve eligible branded-search visibility through Sponsored Brands. Official coverage dates to 2025, so Ivan Marynych’s post is a reminder rather than a new launch.
Read this update in detail →Amazon Listing Protection gives eligible sellers room to appeal
Claire O’Donnell’s original announcement describes a grace period for eligible sellers to address a listing issue before removal, as well as three Seller Challenges within a 180-day period. For advertisers, the relevance is continuity: a listing problem can interrupt an otherwise healthy campaign.
Read this update in detail →Seller Assistant adds ongoing workflows and a plugin route
Amazon describes persistent memory and ongoing Seller Assistant workflows, alongside a US-beta plugin for Amazon Quick and Claude. The practical opportunity is to connect analysis with recurring seller tasks.
Read this update in detail →Seller Central starts bringing other sales channels into one workspace
Amazon’s phased US rollout connects listings and orders across channels including eBay, Shopify, TikTok and Walmart. Distinguish the initial connections and fulfilment workflow from forthcoming editing and profitability capabilities.
Read this update in detail →Prime delivery extends to eligible merchant websites through MCF
Amazon describes a Prime delivery option for US Multi-Channel Fulfilment merchants at no additional fee; fulfilment itself remains paid. A separate pricing programme has its own eligibility and terms.
Read this update in detail →Best Buy and Amazon connect retail audiences with Fire TV
Best Buy’s partnership announcement outlines an expanded path to CTV advertising, with Amazon programmatic activation set to begin in February 2027. It is a planning development rather than inventory to assume is available today.
Read this update in detail →McDonald’s tests a media network in company-owned restaurants
Reporting describes a pilot across 450 US company-owned McDonald’s restaurants. That is a defined test, not access to the chain’s entire estate.
Read this update in detail →Google Commerce Media Network links campaigns with marketplaces
Google’s announcement describes marketplace-linked campaigns, with examples focused on South-East Asia. Review the participating marketplace, seller and account eligibility before assuming the same route is available elsewhere.
Read this update in detail →YouTube tests more flexible Shopping-tag placement
Coverage describes a test of dynamic product-tag placement in YouTube Shopping. Treat it as an experiment in how products appear alongside content, not an option every creator or advertiser can already use.
Read this update in detail →Muse and Shopify move towards checkout inside an agent workflow
The Shopify–Muse announcement brings product discovery and checkout into the same agent-led journey. Confirm merchant eligibility, checkout support and the customer handoff before assuming every catalogue is available.
Read this update in detail →Instacart announces shopping support in Muse
Instacart’s original announcement describes grocery shopping in Muse as coming soon. This extends the commerce discussion beyond general merchandise, but it should not be presented as a feature every customer can use now.
Read this update in detail →Amazon’s block on Muse shows the limits of agent access
Todd Bishop reports Amazon blocking Muse shopping access. Keep Amazon’s stated reasons attributed to Amazon and distinguish the access dispute from the separate Shopify and Instacart integrations.
Read this update in detail →The CMA proposes broader search-choice prompts
The UK Competition and Markets Authority proposes annual search-choice prompts and the inclusion of eligible AI assistants. The consultation runs to 9 October.
Read this update in detail →Google’s September spam update begins rolling out
Google began its September spam update on 24 September, applying globally and across all languages, with a rollout expected to take up to two weeks. This concerns organic Search rankings, not Google Ads bidding or ad eligibility.
Read this update in detail →Comscore’s AI report measures usage, not advertising returns
Comscore’s Q2 research describes changes in prompt share across AI services, including gains for Gemini and Claude within its measurement scope. Prompt share should not be confused with ad spend, revenue, unique users or conversion performance.
Read this update in detail →Hotel AI Overviews appear in a reported side-panel test
Reporting credits Lluc Penycate with a hotel-search side-panel observation. It is an organic search-interface test; no Hotel Ads buying change or universal rollout has been established.
Read this update in detail →An AI Mode experiment examines referrals to publishers
Stephanie T. Wang and co-authors report a preregistered experiment involving 1,100 participants.
Read this update in detail →Spot & Tango makes a defined bet on brand investment
Julia Waldow’s interviews describe Spot & Tango committing $3.5 million to brand marketing after previously allocating no budget to it. The useful lesson is the decision framework behind a shift in allocation.
Read this update in detail →AdRoll reports lower display CPMs in its advertiser sample
AdRoll’s Q3 report covers July to 8 September and reports year-on-year CPM declines of 45% for prospecting and 29.1% for retargeting. These are vendor-sample observations, not a market-wide price guarantee.
Read this update in detail →Comscore data moves into datafuelX planning and stewardship
The partnership announcement describes Comscore demographics entering forecasting, pacing and posting workflows through datafuelX. That may reduce friction in working with different TV measurement currencies.
Read this update in detail →Programmatic podcast buying still needs frequency discipline
Anthony Vargas’s reporting examines podcast buying and ad load as more inventory becomes available programmatically. The practical question is how frequency, audience overlap and the listening experience are controlled across supply paths.
Read this update in detail →CTV measurement needs to account for demand creation
Victoria McNally’s reporting explores the space between immediate response and longer-term brand effects in CTV. Separate creative testing, response measurement and attribution assumptions when reviewing a plan.
Read this update in detail →CTV buying amid media consolidation: keep the discussion practical
Sarah Sluis’s podcast, with Allison Schiff and Victoria McNally, discusses buying in a changing television landscape. Use the conversation to examine supply relationships and measurement questions.
Read this update in detail →Copilot adds Home and Code; Autopilot remains a preview
Microsoft’s 25 September announcement introduces Home and Code with gradual Frontier access, while Autopilot is scheduled for a private preview at the end of the month. For marketing teams, assess a defined research, reporting or production workflow before assuming the whole experience is available.
Read this update in detail →GPT-6 Sol and Luna broaden the options for everyday work
OpenAI’s announcement makes Sol and Luna available through ChatGPT Work, Codex, and the API, with access varying by plan; it also distinguishes these surfaces from Chat. For marketers, compare the models on repeatable analysis and production tasks, measuring factual quality and review effort alongside cost.
Read this update in detail →Claude Opus 5.5 changes the cost and capability trade-off
Anthropic announced Opus 5.5 on 22 September, with changes to capability, pricing and effort controls. Marketing teams should compare it on a repeatable task such as analysing an export or preparing a source-backed brief, including review time and retries in the cost.
Read this update in detail →ChatGPT Voice Work extends spoken requests into working tasks
OpenAI’s 23 September release notes describe Voice Work across documents, presentations, spreadsheets, apps and browser work. Access depends on the relevant Voice and Work capabilities, permissions and usage limits.
Read this update in detail →Claude’s personal connectors reach channel conversations
Claude’s 24 September announcement brings personal connectors into channel workflows, initially through a Team beta, with Enterprise to follow. Review what a request can access and what the resulting channel message reveals to other members.
Read this update in detail →Business Profile hours from text: a reported time-saving workflow
Coverage attributes to Ayman Ali an observation of entering business hours through text. It could reduce repetitive profile administration where available.
Read this update in detail →Muse for Mac and connectors: evaluate a concrete workflow
Coverage of Muse’s desktop and connector expansion provides a practical productivity reference beyond the earlier general agent demonstrations. Start with a bounded task such as preparing research or gathering reporting inputs, then inspect access permissions and the resulting work.
Read this update in detail →WhatsApp Business Tools MCP is a workflow reference to investigate
The documented integration route is relevant to teams exploring agent-assisted business messaging. Before implementation, verify the available tools, account permissions and approval requirements against Meta’s current documentation. Access to an integration is not authorisation to send messages, and the available evidence does not establish this as a new Week 39 release.
Read this update in detail →LinkedIn’s Post Proofreader can support a final writing pass
LinkedIn’s documentation describes a proofreading workflow for posts. Use it after the author has settled the argument and checked factual claims, rather than as a substitute for source verification.
Read this update in detail →Separate deterministic tasks, AI judgement and human approval
Anthony Katsur’s opinion piece argues for hybrid advertising workflows. A practical application is to use deterministic steps for calculations and validation, AI for suitable interpretive tasks, and people for consequential approvals.
Read this update in detail →An agent cannot repair an undefined audience by itself
The audience-data discussion highlights a familiar implementation problem: automating decisions does not establish that the input data represents the intended customer. Review provenance, definitions, freshness and permitted use before connecting an agent.
Read this update in detail →Trigify is closing its standalone platform after the team joins HubSpot
Trigify’s 23 September notice says its team is joining HubSpot and the standalone service will retire. It is not a migration of the product into HubSpot.
Read this update in detail →The advertiser takeaway
Choose the updates relevant to your accounts, then use the linked guides and documentation to plan a bounded review.
Explore the original edition
Every update from this edition is linked above. Explore the original newsletter for its full context, contributor credits and reporting links.
Read Paid Media News (Week 39) on LinkedIn