Marketing Intelligence Weekly: AI Ad Expansion, Publisher Controls and Local Ordering Choices
Date: 6th November 2026
TL; DR
Google’s 28 October expansion makes existing campaigns eligible for English-language AI Mode ads in India, without separate placement controls or reporting. Meta’s Andromeda update describes live infrastructure, not a campaign migration.
Google documents separate pickup and delivery switches for restaurant ordering providers. Check access on individual Indian listings. Google’s proposed shared AI-training control is due in 2027, with Google-Extended choices affecting training permission and Gemini grounding eligibility.
Wendy’s is making US breakfast optional locally, not withdrawing it nationwide. Product upgrades remain tests; a menu refresh is expected in 2027.
Pilot Genie Code imports only after checking account and workspace settings. Supported Power BI formats exclude .pbix; direct uploads have a 100 MB limit. Validate calculations, filters and layout.
Google is scaling AI Mode ads in India, Meta has detailed how its ad-selection infrastructure works, and publishers face a coming choice over AI training and Gemini visibility. This issue separates changes worth acting on now from proposals, tests and capabilities that need an account-level check.
Looking for opportunities? Explore the Job Board, discover opportunities that could fit your next career move.
Search, Social & Platform Updates
Google documents pickup and delivery switches for restaurant ordering providers.
Google’s Business Profile Help documents separate controls for showing each third-party ordering provider for pickup and delivery. Under Food ordering, select the provider, then open “Show provider to customers”. Turn off pickup, delivery or both, then save and confirm.
Google says changes typically appear on Search and Maps within a few hours. The preferred-provider setting is a separate choice about which partner to highlight. Availability on Indian listings remains unconfirmed, so check each profile and verify what customers see after saving.
Google plans a shared AI-training control for 2027.
Google announced on October 30 that it plans to support a shared robots.txt AI-training control based on AIPREF work in 2027. The proposal would let websites express training preferences across participating providers. It is not a live, universal opt-out, and Google has not specified an activation month or day. Existing provider-specific controls will continue alongside the proposed shared signal.
Publishers need to consider training rights and Gemini visibility together. If a site refuses training through the shared signal without explicitly specifying Google-Extended, Google plans to interpret that refusal as a Google-Extended disallow. Explicitly allowing Google-Extended can preserve eligibility for Gemini Apps grounding, where web content helps inform responses, but also creates an exception covering Google’s stated training uses. Eligibility is not assured appearance or referral traffic. Retain working controls while deciding which uses the business is willing to permit.
Why it matters
Both developments concern control over how a business appears through Google, but their readiness differs. Restaurant teams can inspect dashboards now; publishers should prepare a rights-and-visibility decision for a proposed 2027 implementation.
What you can do
Audit ordering partners on each restaurant profile. Where “Show provider to customers” is available, configure pickup and delivery separately and verify the result on Search and Maps.
Review Google-Extended settings with content-rights and SEO owners, recording the desired training and Gemini-grounding position before changing robots.txt.
Paid Media & Acquisition
Google scales AI Mode ads in India, with no separate placement switch.
Google’s October 28 announcement includes India among 10 international markets where it is scaling AI Mode ads. Its current eligibility guidance lists India for English-language ads within both AI Mode and AI Overviews, not just ordinary ads positioned above or below an AI answer. Existing Search, Shopping and Performance Max campaigns are automatically eligible. This is market eligibility, not confirmation that every advertiser or query already has access, and it does not establish support for Indian languages beyond English.
The matching rules deserve attention before budgets change. AI Max, broad-match Search, Shopping and Performance Max can match against both queries and the context of AI responses. Exact- and phrase-match Search keywords qualify only where the query expresses direct commercial intent. Advertisers cannot target only these AI placements or separately opt out, although negative keywords and brand exclusions remain available. Activity is included in aggregate campaign metrics and search-terms reporting, rather than separately identified, limiting placement-specific budget analysis. Brazil, Mexico and the UK remain planned Q4 2026 expansions in the supplied announcement.
Meta explains Andromeda’s larger candidate pool, not a campaign migration.
Meta’s October 28 engineering update describes a redesigned Andromeda system already in production. Andromeda handles retrieval: it narrows eligible ads to a shortlist before a separate, more computationally intensive ranking stage evaluates them. Meta says production throughput exceeds one billion ads per second, roughly 100 times its 2024 system. The design brings index matching, scoring and candidate selection onto one GPU and uses specialised models drawing on signals such as sequential ad engagement and semantic features.
These are infrastructure measurements, not new advertiser controls or an instruction to consolidate campaigns. Meta attributes a 2.4% increase in Facebook conversions since 2025 to Andromeda, not specifically to the October publication. It also distinguishes live throughput from faster benchmarks whose additional optimisations were not yet fully integrated into production. For media teams, the useful response is controlled creative testing and reliable conversion measurement, rather than rebuilding an account around an assumed rollout date.
Why it matters
Google’s expansion can change where existing campaigns serve without a separate campaign launch. Meta’s update explains an underlying selection system without requiring a migration. Distinguishing the two helps teams prioritise genuine control and measurement changes.
What you can do
For India-targeted Google campaigns, review search terms, negative keywords and brand exclusions before expanding matching; assess conversion quality using the available aggregate reporting.
On Meta, test meaningfully different creative concepts against a defined business KPI rather than restructuring campaigns solely because of the Andromeda update.
Content, Brand & Communications
Etsy is treating human creativity as part of its product promise.
Its 28 October shareholder letter describes a strategy that combines AI-assisted discovery and seller tools with clearer policies and stronger enforcement intended to preserve authenticity. Its marketing also emphasises the people behind the products, while personalised recommendations combine machine learning with human curation.
That is a more useful distinction than assuming a marketplace must either embrace AI everywhere or reject it entirely. Automation can support administration and discovery while the customer-facing promise remains about who made an item, how it was made and what makes it distinctive.
For brands built around craft, expertise or provenance, those details belong in the buying journey. A production photograph, a clear materials description or an explanation of the maker's contribution can substantiate a claim that otherwise sounds interchangeable with competitors' copy. Etsy's account describes its own strategy; it does not prove that a particular creative technique will increase another business's sales.
Why it matters
When content becomes easier to produce, a brand still needs a specific reason to be chosen. Showing the evidence behind a promise gives customers something concrete to evaluate.
What you can do
Check one important product page for the evidence supporting its main claim. Make the relevant process, materials, expertise or provenance easy to find.
Define which tasks can use AI assistance and who checks the resulting customer-facing material. Keep product descriptions consistent with what buyers will receive.
Measurement & MarTech
Genie Code supports a controlled Power BI or Tableau import pilot.
Current Azure Databricks documentation describes importing supported Power BI and Tableau files into AI/BI through Genie Code. Treat this as an existing capability to check in your workspace, not a new launch or a guarantee of access in every account. Partner-powered AI features must be enabled at both account and workspace levels. Users can request an import or enter /importBI, then attach a file or reference its Unity Catalog volume path. Direct uploads have a 100 MB limit; larger files can be referenced from a volume.
Supported formats include Tableau .twb, .twbx, .tds and .tdsx, plus Power BI .pbit and .tmdl, not arbitrary Power BI files such as .pbix. Tableau workbooks and .pbit templates produce dashboards and metric views, reusable definitions of business measures; .tmdl produces metric views only. Tableau .tds and .tdsx files supply models for workbooks using published data sources. Power BI imports create or reuse Unity Catalog metric views, while Tableau imports create dashboard-local ones unless a matching Unity Catalog metric view exists. Supply a reference screenshot and validate numbers, filters, metric expressions and layout rather than assuming complete migration fidelity. Keep the browser tab open during a single-file import; bulk imports run in the background.
Why it matters
The import workflow offers a way to evaluate migration effort without assuming dashboards transfer unchanged. Marketing reporting depends on consistent KPI definitions, so matching the original numbers matters more than reproducing its appearance alone.
What you can do
Confirm workspace access and required AI settings, then pilot one representative dashboard with a screenshot and a checklist of KPI calculations, filters and layout.
Agree metric-view ownership with administrators and confirm licensing and compute requirements before expanding the migration.
Looking for opportunities? Explore the Job Board, discover opportunities that could fit your next career move.
Comments