Digital Digest
Marketing Intelligence Weekly
Weekly updates on the digital marketing news you need to know, including data reform, AI, cookie deprecation and platform updates.
AI in digital
Platforms
ChatGPT Atlas is here, OpenAI’s new agentic browser
- OpenAI has launched ChatGPT Atlas, a new web browser built with ChatGPT at its core – designed to be your browser and your AI assistant in one.
- Atlas integrates ChatGPT’s memory and browsing context so it can draw on your past chats & visited pages to help you complete tasks seamlessly (e.g., “Find all the job postings I was looking at last week…”).
- Built‑in “agent mode” lets ChatGPT act on your behalf (opening tabs, digging through docs, ordering groceries), launching now in preview for Plus/Pro/Business users.
- Privacy & control are emphasised: you decide what ChatGPT sees/does, you can toggle site visibility, archive or delete browsing memories, and it doesn’t by default use your browsing content for model training.
- From a marketing lens: this could shift how brands engage with users – the browser itself becomes an AI “assistant channel”, so think about discoverability, how your content might be surfaced in AI‑driven flows, and how user tasks might get delegated to agents rather than traditional navigation.
- Read more
AI in digital
Marketing by numbers
Platforms
LLM traffic for eComm converts worse than Google Search
- A study looking at 12 months of first‐party analytics data from 973 ecommerce sites (Aug 2024–Jul 2025) with combined revenue of approx. US$20 billion showed LLM traffic converted worse than Google Search.
- Traffic driven by ChatGPT (LLM referrals) was about 0.2% of total sessions—roughly 200 × smaller than organic from Google search.
- LLM referral traffic had a lower conversion rate and lower revenue per session compared with organic search and affiliate channels; only paid social converted worse.
- Session depth and average order value from LLM sources were also weaker than traditional search channels; the authors suggest shoppers may still “verify elsewhere” before buying.
- Read more
Cookies
Data privacy
Platforms
Google shuts down its Privacy Sandbox
- Google has retired 10 remaining Privacy Sandbox APIs (including Attribution Reporting, Topics, Protected Audience) for both Chrome and Android.
- The move follows Google dropping plans to phase out third‑party cookies in Chrome, meaning the familiar ad‑tech ecosystem remains intact — at least for now.
- While this offers stability for advertisers in the short term (you can keep relying on existing measurement/targeting tools), it signals that a truly privacy‑first ad framework is still unresolved.
- Google says it will continue privacy improvements (e.g., CHIPS, FedCM, Private State Tokens) but is moving away from the “Privacy Sandbox” branding.
- For marketers: it’s a moment to reassess your dependency on cookie‑based tracking, explore alternative signals and attribution models, and monitor where the ad‑tech ecosystem is heading next.
- Read more
AI in digital
Marketing by numbers
Platforms
81% of AI Mode sessions result in 0 clicks
- Key takeaways from the first ever AI mode study:
- Across the 250 tasks, the median number of clicks per task was 0. Users often didn't click out of AI Mode.
- Looking at the session level, external clicks were still rare. Only 77.6% of sessions had an external visit.
- Most people claimed that they "skimmed quickly" when reading AI Mode results.
- When external visits happened, they mostly occurred in "Shopping" results. 22% of the exits were attributed to a "Shopping Pack"
- When giving product options, AI Mode preferred to send users to marketplaces over brands for generic queries.
- Read more
AI in digital
Platforms
Meta’s Festive Season Guide is here
- Three key elements were highlighted to capitalise on the festive season to lower cost per result:
- Maximise opportunity score. Score is determined by examining an ad account vs other advertisers in the same industry so it does pay to action recommendations.
- Make use of AI advantage + features to optimise and automate parts of your campaign using AI (creative, targeting, placements, budgets).
- Meta also highlighted adding the meta verified badge to your ad accounts to build consumer trust, increased search result visibility and higher traffic through features like links on organic reels. However, we don’t necessarily recommend this due to the additional cost.
- Read more
Platforms
Import cost data directly from Meta and TikTok in GA4
- Google Analytics now offers native integrations to automatically import cost‑data from Meta Platforms (Meta) and TikTok ads, enabling unified cross‑channel spend and ROI tracking.
- The integration supports pulling up to 24 months of historical cost data and then ongoing automatic updates, reducing the need for manual uploads or third‑party connectors.
- Marketers must ensure they remove any overlapping manual cost‑data imports, as the system doesn’t automatically dedupe imported records — duplicates could inflate campaign cost/ROI results.
- With this update, Google Analytics becomes a more complete hub for paid‑media analysis, helping compare spend efficiency across Google, Meta and TikTok campaigns without siloed data.
- Read more
Platforms
Google now lets you “Hide sponsored results”
- Google is replacing individual ad labels with a persistent “Sponsored results” label to clearly group text ads on Search.
- A new “Hide sponsored results” feature lets users collapse ads, focusing only on organic content.
- The updated label design applies to all ad types including Shopping ads, now shown as “Sponsored products”.
- These changes aim to improve transparency and user navigation, and are rolling out globally on desktop and mobile.
- The number and size of ads remain unchanged, with a max of four grouped text ads displayed.
- Read more
AI in digital
Platforms
Ads in AI Overviews expands to more markets
- Google is rolling out AI Overview ads beyond the US to more English-speaking markets by the end of 2025.
- Ads will appear directly within AI-generated summaries, offering brands new ad placements in response to complex search queries.
- This shift marks a major change in search advertising, blending paid results with conversational AI content.
- Marketers will need to adapt strategies to optimise for visibility and intent in this new, more context-driven ad environment.
- Gradual rollout allows for testing and learning, giving advertisers time to refine their approach.
- Read more
AI in digital
Semrush study of 1000 domains shows how link building matters for AI search
- Authority Score is key: Domains need to hit a higher Authority Score threshold before backlinks make a real difference in AI visibility. Weak or incremental backlink growth won’t cut it.
- Link quality > link volume: A strong positive correlation (Pearson: 0.65) was found between link quality and AI mentions. AI systems clearly prioritise a few high-authority backlinks over many weak ones.
- Nofollow ≈ Follow links: Surprisingly, nofollow links perform nearly as well as follow links in boosting AI visibility — especially when from reputable sources like Reddit, Wikipedia, or Quora.
- Image links outperform text links: Image-based links showed stronger correlations (Pearson: 0.415) with AI mentions than traditional text links, especially for higher-authority domains.
- Diverse referring domains matter: The diversity of linking domains strongly influences authority scores. Marketers should prioritise outreach to new, relevant websites instead of stacking links from the same sources.
- Read more
Platforms
Subscription goods now available via Google Shopping Ads
- Google now allows merchants in the U.S. to sell physical goods on a recurring (subscription) basis via Shopping ads, expanding beyond just digital services or one‑off physical goods.
- The update covers specific categories such as apparel & accessories, coffee, healthcare (excluding prescription drugs), home & garden, personal care, pet supplies, prepared foods and toys.
- Merchants must supply new attributes in their product feed: subscription_cost (with sub‑attributes for period, period_length and amount). Only one subscription price per landing page is supported, and discounts on subscription offers are not yet allowed.
- For marketers, this opens up a fresh lever: recurring‑revenue models via Google Shopping, boosting customer lifetime value and loyalty from “auto‑replenish” style buys. Early adopters in suitable categories stand to gain a competitive edge.
- Read more
AI in digital
Platforms
AI-generated trending previews now in Google Discover
- Google has launched AI-powered trending previews in the Discover feed, now live in the US, South Korea, and India.
- These previews summarise trending topics and link directly to publisher or creator content, offering new visibility opportunities.
- A “What’s new” sports feed is also set to roll out soon in the US for mobile users, surfacing fresh sports-related content.
- The updates reflect Google’s ongoing push to integrate AI summaries and personalisation across its platforms.
- Read more
AI in digital
Help desk
Platforms
AI Mode is now in Australia
- Google has officially launched AI Mode in Australia. The update started rolling out last week on 8th of October across the Google Search experience.
- If you’re currently investing in Google Ads, no immediate action is required. Ads are not yet live within AI Mode - Google anticipates introducing ad placements in early 2026. In the meantime, Google recommends advertisers begin leveraging its existing AI-powered ad solutions to prepare for the upcoming integration.
- AI Mode is transforming organic search by prioritising direct, summarised answers over traditional content. With fewer outbound clicks and greater emphasis on visibility within AI-generated search results, success now depends on creating authoritative, intent-driven content that AI recognises as trustworthy and comprehensive.
- At In Marketing We Trust, our team has spent the past six months researching, testing, and adapting our strategies to ensure our clients remain competitive as the search landscape continues to evolve. We’ll continue to monitor AI Mode’s rollout and share updates as more details become available.
- Get a free consultation
AI in digital
Platforms
Built-in apps now in ChatGPT
- ChatGPT now supports built‑in apps that you can invoke conversationally (e.g. “Spotify, make me a playlist”), or have suggested by ChatGPT when contextually relevant.
- At launch the apps include partners like Booking.com, Canva, Coursera, Figma, Spotify, Zillow, Expedia — giving users integrated functionality (maps, slides, listings) without leaving chat.
- Developers can build their own ChatGPT apps using the Apps SDK preview, which is open and built on the Model Context Protocol (MCP).
- All apps must comply with OpenAI’s safety, data minimisation, and privacy policies; users will be prompted explicitly to connect apps and understand what data is shared.
- In coming months, OpenAI will open up submissions, allow monetisation, launch directories, and roll out to Business/Enterprise/Edu tiers.
- Read more
AI in digital
Platforms
AI Mode goes agentic in Search Labs in the US
- Google has added Agentic capabilities to AI Mode in Search Labs, allowing users to make restaurant reservations via natural language prompts.
- Soon, the feature will expand to booking local services and event tickets — a clear push toward task-completion within search itself.
- Available only in the US for now, users must be signed into their personal Google Account with Web & App Activity enabled.
- The AI searches multiple platforms in real-time, returning curated booking options and linking directly to finalise the task.
- For marketers, this hints at a future where search becomes an action engine — making visibility in task-relevant results more important than ever.
- Read more
AI in digital
Platforms
Google replacing meta descriptions with AI-generated content
- Google is experimenting with replacing or supplementing standard meta descriptions in search results with AI‑generated descriptions or AI‑summaries of page snippets.
- These AI snippets are being marked with a small Gemini (Google’s AI branding) icon to signal they were generated by AI.
- In many cases, Google hasn’t always used the meta descriptions you provide anyway — so this shift could give Google more control over how your page is represented in SERPs.
- For SEOs and content marketers: now more than ever, focus on clarity in your headers, intro paragraphs and content structure — since Google’s AI may pull from any part of your page to craft snippets.
- Read more
AI in digital
Platforms
Google DeepMind launches Gemini 2.5 Computer Use model
- Google DeepMind has launched the Gemini 2.5 Computer Use model, a UI-controlling AI agent that mimics human interactions on screens—think automated browsing, form filling, and navigation at scale.
- Marketers can expect AI that actually operates in-browser—clicks buttons, types text, and scrolls through pages—enabling more hands-off automation for tasks like testing landing pages, scraping competitors, or managing web workflows.
- It uses a feedback loop via a “computer_use” API—you give it a screenshot and past actions, it decides the next move—offering flexibility for custom automations without building full software integrations.
- While it's still browser-first (not desktop-native), it's showing early promise for mobile UI support, which could be big for mobile-first marketing stacks and campaign ops.
- Strong safety protocols are baked in, helping dev teams build automations with guardrails, so marketers don’t have to stress about rogue behaviour or compliance risks.
- Read more
AI in digital
Marketing by numbers
Online holiday sales expected to grow 5.3% YoY with mobile and AI trending
- Record spend incoming: Online holiday sales are forecasted to hit $253.4B in the US, up 5.3% YoY, with Cyber Monday peaking at $14.2B.
- Mobile takes the lead: For the first time, mobile is expected to drive over half (56.1%) of online sales, with 70% of retail visits happening on mobile.
- BNPL booms: "Buy Now, Pay Later" usage will grow 9–12%, totalling up to $20.4B, with Cyber Monday breaking $1B in BNPL spend for the first time.
- AI-powered traffic explodes: Visits to retail sites from AI sources like chatbots are projected to surge 515–520%, with Thanksgiving Day seeing 730% growth YoY.
- Hot sellers: Toys, video games, consoles, and electronics are expected to be the most in-demand categories this season.
- Read more
AI in digital
Platforms
Buyers can now shop and purchase directly in ChatGPT
- ChatGPT is introducing Instant Checkout, letting users buy items directly in-chat (starting with U.S. Etsy sellers) without leaving the conversation.
- This is powered by an open‑standard protocol (the Agentic Commerce Protocol) built with Stripe, so merchants and developers can integrate with minimal friction.
- In this first phase, only single‑item purchases are supported; multi‑item carts and broader merchant/region rollout are coming later.
- The vision: shift ChatGPT’s role from just “helping you find something” → “helping you buy something,” making AI an active participant in commerce.
- Read more
AI in digital
Data privacy
Platforms
Meta to target ads based on users’ interactions with AI
- Meta will start using data from users’ interactions with its AI products (chats, voice, images, etc.) to target ads on Facebook and Instagram, with the change going live via a privacy policy update on 16 December 2025.
- The policy change will apply globally except in the EU, UK and South Korea (where privacy laws block this usage).
- Meta claims it will not use topics considered “sensitive” (e.g. political views, health, religion, sexual orientation) for ad targeting.
- There will be no opt‑out option other than not using Meta AI features, according to their announcement.
- For marketers, this adds a powerful new signal (conversational intent) to audience targeting, but risks around privacy, regulation, and brand trust will need to be carefully managed.
- Read more
Platforms
Microsoft launches new Bing Places for Business
- Microsoft has launched a revamped Bing Places for Business, now offering free listing management across Bing Search and Bing Maps.
- A new AI-powered recommendation tool helps businesses prioritise which listing fields to update for better visibility.
- The Google Business Profile import is now faster, with support for bulk edits, dashboards, and real-time status tracking.
- Existing users will be automatically migrated to the new platform, easing the transition.
- Read more
AI in digital
Platforms
Meta announces new AI features ahead of Advertising Week
- Meta launches "Business AI" concierge: A no-code AI agent now helps businesses drive conversions via Meta ads, messaging apps, and even their websites—aimed at streamlining sales and support, especially for SMBs.
- GenAI tools upgraded for creative optimisation: Advantage+ now includes AI music, dubbing for multilingual markets, and persona-based image generation—plus enhanced video visuals for more engaging storytelling.
- New AI-driven shopping experiences: Features like virtual clothing try-ons and dynamic personalised landing pages are being tested to boost shopper confidence and ad performance.
- Boosted creator partnerships via new APIs: Creator discovery on Facebook and Instagram is easier with expanded APIs, allowing brands to search, connect, and convert organic content into optimised partnership ads.
- Meta AI Business Assistant debuts: This 24/7 AI chat tool in Ads Manager gives marketers campaign tips, troubleshooting help, and performance insights—rolling out to select small businesses before broader expansion in 2026.
- Read more
AI in digital
Platforms
OpenAI launches Sora 2 and TikTok-style app
- OpenAI launched Sora 2, a next‑generation text-to-video + audio model, offering more realistic visuals, better physics, and synchronised sound.
- Alongside it, they released a companion app called Sora (invite‑only, iOS currently), which presents AI-generated video clips in a TikTok-style vertical feed.
- A standout feature is Cameos — users (or friends, with consent) can insert their likeness into generated scenes, and can revoke access anytime.
- OpenAI says it's not optimising solely for consumption — the app is built to encourage creation, remixing, and collaborative content.
- Read more
Platforms
New requirements for message assets in Google Ads
- Google will begin enforcing stricter Message asset requirements from 30 October 2025, rolling in full over ~4 weeks.
- Message assets that don’t comply or can’t be verified will be prevented from serving.
- Verification is required — e.g. for message IDs, usernames, and associated phone numbers — to ensure the message asset truly relates to the business.
- Marketers should audit and update existing message assets ahead of the October deadline to avoid ad disruption or disapproval.
- Read more
AI in digital
AI in travel
Platforms
Platforms in travel
Google Search Live launched in US
- Google Search Live has launched in the U.S., now available in the Google app (no Labs opt-in needed), offering interactive voice and camera-based search powered by AI.
- Users can tap the new "Live" icon or access it via Google Lens to have real-time, conversational help using visual input from their camera.
- The feature is designed for on-the-go queries, like travel exploration, troubleshooting tech, or identifying objects in your environment.
- ”Imagine you’re on vacation and getting ready at the hotel before a day of exploration. You can go Live with Search to have a hands-free conversation about the neighborhoods you want to visit, all while you finish applying your sunscreen. And once you’re out and about, you can fire up Live and activate your camera to ask about anything you see.”
- Google positions it as a hands-free, instant knowledge companion — useful for learning new hobbies, helping with school projects, or even choosing a board game.
- Strong play towards mobile-first, visually rich AI interaction, giving brands new reasons to optimise for conversational and visual search experiences.
- Read more
Platforms
Google piloting Brand Profiles in Merchant Center Next
- Google is piloting Brand Profiles in Merchant Center Next, letting retailers surface their brand story, values, and offers directly in Search.
- These profiles support images, videos, business descriptions, and promoted offers — enriching what shows up above just product listings.
- Right now, editing these profiles is locked to “super admins” and the feature is only available to a select group of merchants.
- Google is aiming to expand eligibility over time and notify merchants as they become eligible.
- In Google’s documentation, they note the profile content is auto‑sourced (from your site, Merchant Center data, third‑party & public sources), but some elements (promotions, delivery, returns, social links) can be updated in Merchant Center (in supported countries)
- Read more
AI in digital
Platforms
OpenAI gearing up to turn ChatGPT into an ad platform
- OpenAI is hiring a Growth Paid Marketing Platform Engineer to build internal tools for ad campaign management, real‑time attribution, and ad platform integration — signalling a move to turn ChatGPT into an ad platform.
- The hire is part of a new “ChatGPT Growth” team that will build the backbone of OpenAI’s paid marketing infrastructure from scratch (APIs, data pipelines, experimentation frameworks).
- OpenAI aims to monetise free users via ads by 2026, offering brands direct access to ChatGPT’s ~700 million weekly active users.
- Read more
AI in digital
Data privacy
Platforms
Google introduces VaultGemma, a privacy-first LLM
- Google has released VaultGemma, a 1B‑parameter LLM trained with differential privacy to better protect individual data points during training.
- It uses a mix of open datasets and synthetic data, plus added statistical noise, so that the model is less likely to “memorise” or leak sensitive training data.
- VaultGemma aims to strike a balance: it delivers comparable utility to non‑private models (of older generations) while abiding by strict privacy guarantees.
- Google is open‑sourcing the model, its training tools, documentation, and privacy verification scripts so researchers can experiment and build on it.
- While it’s not intended yet for production use, it provides a privacy‑first testbed and signals where AI development might go for privacy‑sensitive domains like healthcare, finance and government.
- Read more
AI in digital
Data privacy
81% of employees sharing sensitive company data with free AI tools
- 81% of workers using free AI tools admit to feeding in sensitive company data — even though that data could be used to train the AI models.
- A study by HP/Microsoft of Australian SMEs found 58% have already adopted AI, and 88% intend to use it to streamline operations.
- Among tool users: 36% use ChatGPT, 32% use Microsoft Copilot, 35% use Google’s Gemini.
- Top internal uses include reporting, operational tasks, customer interaction, and recruitment support.
- The authors and company leaders flag big risks: data leaks, social engineering, malware exposure — “short‑term gains” may hide severe long‑term consequences.
- Read more
Platforms
Google rolls out seasonal bid adjustments in app campaigns
- Google is rolling out a beta feature allowing seasonal bid adjustments in app campaigns, letting marketers pre‑flag periods of high conversion so Smart Bidding can ramp up proactively.
- It works across all app campaign bidding strategies and is intended for short, high-impact windows (1–7 days), not for routine fluctuations.
- The aim is to avoid Smart Bidding “learning too late” during spikes (e.g. launches, flash sales) and ensure you don’t leave revenue on the table.
- It’s not meant as a replacement for everyday bid management — Smart Bidding will still handle usual conversion shifts.
- Read more
AI in digital
AI in travel
Platforms
Platforms in travel
Google launches Agent Payment Protocol with fully automated payments
- Google has launched Agent Payments Protocol (AP2) — an open protocol for purchases initiated by AI agents. It aims to bridge AI platforms, payment systems and vendors to allow agents to shop and transact on users’ behalf.
- Over 60 merchants and financial institutions are backing AP2. Major names like Mastercard, American Express and PayPal are already involved.
- The protocol requires two layers of approval before a purchase: Intent mandate — user defines what they’re looking for (e.g. “flights to Sydney”) plus rules like price limits or timing. Cart mandate — final purchase approval once a suitable item is found.
- There’s room for fully automated purchases, but only under stricter intent mandates so users’ constraints (budget, timing etc.) are respected.
- Google is releasing the full spec on GitHub and wants the payments & tech community to build around it, possibly through standards bodies.
- Read more
