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 rolls out advertising in Australia
- ChatGPT is set to roll out advertising in Australia within weeks, expanding on a US pilot that showed “encouraging early signals.”
- Ads will only be shown to logged-in adult users on Free and ‘Go’ tiers, with paid plans (Plus, Pro, Business, Enterprise) remaining ad-free.
- The move signals a clear monetisation shift for OpenAI, as it looks to scale revenue while keeping the core product accessible to a growing user base.
- Early formats place ads within the chat experience (e.g. alongside responses), opening a new performance channel that blends search-style intent with conversational AI.
- OpenAI maintains that user conversations remain private and that ads won’t influence responses—critical for maintaining trust as the format evolves.
- Read more
Platforms
Google March 2026 broad core update is here
- Google has officially launched its March 2026 broad core update (March 27), with rollout expected to take up to two weeks, meaning ranking volatility will continue into early April.
- This is the first core update of 2026 impacting Search, following a Discover-only update in Feb and a rapid spam update just days earlier.
- As usual, Google hasn’t shared specifics — core updates are broad quality recalibrations, not targeted penalties, focused on surfacing more helpful and reliable content.
- Marketers should expect ranking and traffic fluctuations during rollout and avoid premature analysis until at least a week after completion.
- Any performance drops don’t indicate a penalty — instead, it reflects re-evaluation of content quality vs competitors, reinforcing the need for stronger content fundamentals.
- Read more
AI in digital
Platforms
Bing now shows AI queries mapped to your URLs
- Bing has upgraded its AI Performance dashboard to map “grounding queries” directly to the pages cited in AI answers, bridging a major gap in earlier reporting.
- Marketers can now click a query to see which pages are cited (and vice versa), making it much easier to tie AI visibility back to specific content and prioritise optimisations.
- The mapping is many-to-many, meaning one query can trigger multiple pages and one page can appear across multiple AI retrieval queries—offering richer insight into how AI interprets content.
- This is especially valuable because grounding queries aren’t user searches—they’re the phrases AI generates to build answers, giving a new layer of intent data for content strategy.
- Limitations remain: the dashboard is still in preview, shows sampled citation data only, and doesn’t include click-through or traffic metrics (so visibility ≠ performance… yet).
- Read more
AI in digital
Platforms
Google unveils Google-Agent giving clearer visibility when AI agents interact with your site
- Google has introduced a new “Google-Agent” user agent, rolling out over the next few weeks, designed to power AI agents that browse and take actions on behalf of users.
- Unlike traditional crawlers (e.g. Googlebot), this agent is tied to user-triggered tasks — such as form fills or assisted browsing via tools like Project Mariner.
- It gives marketers and site owners clearer visibility in server logs when AI agents (not humans or bots) are interacting with their sites.
- The move signals a broader shift toward agent-driven web interactions, where Google’s AI actively completes actions rather than just indexing content.
- Google is also testing new verification approaches (e.g. web-bot-auth via agent.bot.goog) to authenticate these AI-driven requests.
- Read more
AI in digital
Platforms
Shopify rolls out Agentic Storefronts enabling brands to sell directly inside AI chats
- Shopify is rolling out agentic commerce to all merchants, letting brands sell directly inside AI chats and assistants via “Agentic Storefronts” — no traditional website visit needed.
- Products, pricing, and inventory are synced in real time across AI surfaces, meaning shoppers can discover and purchase seamlessly wherever they’re interacting (e.g. chat, copilots).
- This signals a shift from search-led to agent-led shopping, where AI doesn’t just recommend products but can handle discovery, decision-making, and checkout.
- Shopify is positioning itself as the infrastructure layer for AI commerce, enabling millions of merchants to plug into AI ecosystems rather than building bespoke integrations.
- For marketers, this means optimising for AI visibility (structured product data, accuracy, availability) becomes critical as agents—not ads or search—drive purchase decisions.
- Read more
AI in digital
Platforms
Anthropic launches Computer Use for Claude Cowork and Code
- Anthropic has launched computer use for Claude Cowork and Claude Code, letting Claude point, click, type and navigate apps on your computer when there isn’t a direct connector available.
- The feature is currently in research preview for Claude Pro and Max subscribers, and Anthropic says it works particularly well with Dispatch, which lets users assign Claude tasks from their phone and pick them up later on desktop.
- Anthropic is leaning hard into safety messaging, saying Claude asks for explicit permission, requests access before opening new apps, and includes safeguards aimed at reducing risks like prompt injection.
- For now it’s supported on macOS only through the desktop app with the feature enabled.
- Read more
Platforms
Shopify reframes SEO as AI visibility with GEO playbook
- Shopify positions Generative Engine Optimisation (GEO) as the next evolution of SEO, as product discovery shifts from search engines to AI tools like ChatGPT and Google AI Overviews.
- GEO focuses on making brands and products appear in AI-generated answers, not just ranked links, meaning visibility now depends on being recommended by AI rather than clicked in search results.
- The shift is already material: AI-driven traffic to Shopify merchants grew 8x and orders 15x year-over-year, with a majority of shoppers now open to using AI in purchase decisions.
- However, the rules remain unclear, with no guaranteed way to “rank” in AI systems, forcing brands to act early despite limited playbooks.
- The implication is a new visibility battleground where brands that adapt first to AI discovery channels capture disproportionate demand, while others risk becoming invisible in high-intent buying moments.
- Read more
AI in digital
ChatGPT citations remain unpredictable as fan-out reshapes search visibility
- An AirOps study finds that ChatGPT retrieves far more content than it cites, with only ~15% of pages ever appearing in final answers, making visibility highly selective.
- This is driven by “query fan-out,” where a single prompt expands into many sub-queries, pulling content from multiple angles rather than a single search result set.
- Google rankings still influence discovery, but overlap is limited, meaning high SERP positions do not guarantee AI citations.
- The implication is a shift from ranking to retrieval and citation optimisation, where structured, extractable content across multiple intents increases the likelihood of selection.
- This reframes SEO into a probabilistic system, where visibility depends on how well content matches the full fan-out of AI queries, not just a single keyword.
- Read more
AI in digital
Vector databases fall short for SEO math as hybrid AI + SQL emerges
- An SEO experiment building a vector database from Google Search Console data shows that while AI-driven analysis improves discovery, it struggles with precise, metric-based queries.
- The system embeds 16 months of search data into a vector database, allowing natural language queries and pattern detection, but relies on semantic similarity rather than exact calculations.
- This creates a clear limitation: vector databases cannot reliably handle thresholds, rankings, or aggregations, where traditional SQL remains objectively more accurate.
- However, vector search outperforms SQL in identifying related topics and content gaps without needing predefined keywords, enabling broader strategic insights.
- The takeaway is a hybrid model: vector databases for discovery and pattern recognition, SQL for precision, highlighting that AI-native data stacks will complement, not replace, traditional analytics.
- Read more
Platforms
Google tightens e-commerce rules as out-of-stock workarounds are banned
- Google updated Merchant Centre rules, requiring retailers to show a visible but disabled “buy” button on out-of-stock product pages to improve consistency and user clarity.
- Previously common tactics, hiding the button or leaving it clickable, are now non-compliant, with stricter alignment required between on-page availability and product feed data.
- The change raises the risk of product disapprovals and ad disruption, forcing retailers to fix page UX and data consistency across systems.
- This signals a broader shift toward stricter control of product accuracy in Google’s commerce ecosystem, where feed integrity and real availability increasingly determine visibility and performance.
- Read more
AI in digital
CATCHES launches AI tool to improve online clothing fit
- CATCHES launched a generative AI sizing tool using physics-based simulation to improve fit accuracy, targeting high return rates in fashion e-commerce.
- The platform models how garments interact with real body shapes, moving beyond size charts and visual try-ons, with AMIRI as its first adopter on NVIDIA infrastructure.
- If effective, it could reduce returns and boost conversions, while shifting competition toward deeper, data-heavy fit simulation.
- Read more
AI in digital
Data privacy
Platforms
Perplexity fails first test in agentic ecommerce as judge rules in Amazon’s favour
- A US federal judge granted Amazon a preliminary injunction blocking Perplexity’s “Comet” AI shopping agent from accessing Amazon’s site and making purchases on behalf of users. The court said Amazon showed strong evidence the tool accessed password-protected accounts without Amazon’s authorisation.
- The ruling requires Perplexity to stop using its AI agent on Amazon and destroy any Amazon customer data collected through the tool, though the company has a short window to appeal the decision.
- At the centre of the case is a key question for AI commerce: do AI agents inherit a user’s permissions online? The judge rejected Perplexity’s argument that its bot simply automates actions users request.
- Amazon argues the bot disguised itself as a normal browser and scraped data, posing potential security and fraud risks, while Perplexity claims the lawsuit is about protecting Amazon’s ad-driven shopping ecosystem.
- The case is an early legal test for “agentic commerce”—AI tools that browse, compare products and complete purchases—which could disrupt retail search and the billions retailers earn from sponsored listings and ads.
- Read more
AI in digital
Data privacy
Platforms
OpenAI updates privacy policy in relation to ad targeting
- OpenAI updated its privacy policy to support the rollout of ads in ChatGPT, clarifying how ad personalisation, measurement, and data use will work as the platform moves deeper into advertising.
- Advertisers won’t get access to user conversations or chat history. Instead, they’ll only receive aggregated metrics (e.g., views and clicks) while ads are targeted using limited signals and user settings.
- Ads will primarily appear for Free and “Go” tier users, while paid tiers such as Plus, Pro and Enterprise will remain ad-free for now.
- OpenAI may receive conversion data from advertisers (e.g., if a user clicks an ad and later purchases), helping measure ad effectiveness — a standard practice in digital advertising.
- Read more
AI in digital
AI in travel
Platforms
Platforms in travel
Google launches Ask Maps, a Gemini conversational feature for itinerary planning
- Google Maps has launched “Ask Maps,” a Gemini-powered conversational feature that lets users ask complex, real-world questions directly in the app (e.g. finding a phone charging spot without long café queues). It replaces traditional keyword search with natural language queries.
- The tool delivers personalised recommendations by analysing data from Google Maps’ massive database of places, reviews and community contributions, tailoring results to user preferences and past behaviour.
- Ask Maps helps users move from discovery to action, suggesting locations, building custom itineraries, and even enabling actions like restaurant bookings directly within the conversation.
- The feature launches first on mobile (Android and iOS) in the US and India, with plans to expand to desktop and other regions over time.
- Read more
AI in digital
Platforms
Google Ads to automatically add AI voice over on your PMax campaigns
- Google Ads is rolling out AI-generated voice-overs for Performance Max video ads, automatically adding narration to videos that don’t already include spoken audio. The system pulls text from your existing headlines and descriptions to create the voice track.
- The feature is opt-out, not opt-in. Advertisers will be automatically enrolled unless they disable “video enhancement control” before 20 March 2026 in their campaign settings.
- When enabled, Google generates a new video asset with the AI voice layered over the original video, which then competes in auctions alongside the original creative.
- Only silent or music-only videos are affected. Ads that already contain narration or spoken audio won’t be modified by the feature.
- Read more
AI in digital
Majority of Australians can’t tell if an Ad is AI or not
- New B&T/Ideally research suggests most Australians still can’t reliably spot AI-made ads: only 12 per cent said they were very confident they could tell the difference, based on a nationally representative survey of 400 people.
- For marketers, the key takeaway is that disclosure matters more than detection. While 67 per cent said it would matter at least somewhat if an ad was AI-created, two in five said discovering that wouldn’t change their perception of the brand.
- Trust impact looks fairly limited overall, but not evenly spread. Women, older Australians and lower-income groups were more likely to say they’d trust a brand less if they found out the ad was made with AI, while younger and higher-income audiences were generally more relaxed or even positive.
- The bigger risk for brands isn’t the creative itself, but governance and data use. OneTrust’s Michael Schanker argues the real danger is around privacy, training data and agency-client transparency — particularly in more risk-sensitive categories where brands will want clear disclosure on tools, trade-offs and potential exposure.
- Read more
Marketing by numbers
Australia’s finance sector spent big on ads in 2025
- Australia’s finance category went hard on media in 2025, with ad spend up 20% year-on-year to $756 million, signalling a much more aggressive fight for customer attention across banking, credit, savings and investing.
- The big banks are leading the charge, with Westpac the top advertiser, followed by Commonwealth Bank, ANZ and NAB. That points to a highly competitive market where share growth is being driven as much by brand visibility as by product offers.
- Rose Lopreiato, Nielsen Ad Intel’s Australia Commercial Lead, said: "Financial brands are operating in a category where competitive activity and consumer pressure are rising at the same time. Ad Intel shows where advertising investment is increasing and which brands are driving that pressure, while CMV helps explain the consumer mindset behind it, from switching intent and cost-of-living stress through to demand for credit, savings and investment products. Together, those insights help marketers understand not just where the market is moving, but why.”
- Read more
AI in digital
Platforms
Update on ChatGPT ads
- OpenAI is rolling out ads in ChatGPT — the company has started introducing ads to users of the free and Go subscription tiers in the U.S. market as part of a broader shift toward monetisation.
- COO Brad Lightcap says this will be “an iterative process” — emphasising that the company wants to get it right, particularly around user trust and privacy.
- Lightcap noted that if ads are done right, they “can be additive to a product experience”, but he also acknowledged it will take time and asked people to “give us a few months” to evaluate how this goes.
- He didn’t discuss plans for expanding ads beyond the U.S. at this stage.
- Read more
AI in digital
Platforms
Google has patented AI-generated landing pages
- A recently granted patent describes a system where Google could generate a custom AI page for a search query rather than sending users directly to a traditional website URL.
- The system would use the user’s query (and potentially context) to build a dynamic page tailored to intent — for example, pre-filtered product results instead of a generic category page.
- Importantly, this is patent speculation, not an announced or implemented feature in Google Search right now.
- Even if not deployed soon, the patent hints at how Google could blend AI content and search results more deeply going forward.
- Read more
AI in digital
Platforms
Microsoft rolls out mitigations to thwart AI recommendation poisoning
- Microsoft has moved to stop prompt injection attacks that were being used to bias AI outputs by tricking models into favouring specific companies or responses when users clicked “Summarize with AI” buttons.
- These attacks — dubbed “AI Recommendation Poisoning” — embed hidden commands in URLs to make the AI assistant remember or prefer certain companies, subtly skewing future replies on everything from health to finance.
- Microsoft says the technique worked not just against Copilot but also ChatGPT, Claude, Perplexity, Grok and others, showing the broad nature of the threat.
- The company is rolling out ongoing mitigations and protections to detect and block these injections, as part of a larger push to protect AI models from manipulation.
- Read more
AI in digital
Platforms
Perplexity announces “Computer” AI workspace
- Perplexity launches “Computer”, a new AI-powered digital worker that goes far beyond simple chatbot responses, able to carry out complete workflows, from researching to coding and managing projects autonomously.
- It orchestrates multiple specialised AI models (around 19), routing tasks to the best model for the job, meaning research, design, app builds or analysis happen in one place rather than across many tools.
- End-to-end execution is the big differentiator, with Computer breaking large goals into sub-tasks and running them in the background over long periods (hours to months).
- Available now to paid subscribers, it’s positioned as part of Perplexity’s strategy to move from search answers into full project output, potentially streamlining workflows for developers and teams.
- Early buzz and experiments highlight real-world potential, from building apps to complex analytics, but also spark debate about limitations vs traditional tools.
- Read more
Marketing by numbers
Platforms
Gen Z’s love of TikTok is cooling, with preferred search providers changing
- Gen Z’s stated preference for TikTok over Google has halved — only about 4% of Gen Z now say they’re more likely to rely on TikTok than Google for search, down from 8% previously.
- TikTok use for search still widespread, with 49% of U.S. consumers saying they’ve used TikTok as a search tool, but that doesn’t mean they prefer it over traditional search engines.
- ChatGPT is emerging as a broader challenger, with 14% of all consumers saying they’d rather use ChatGPT than Google for search — double the share for TikTok.
- Google remains dominant when users are asked which platforms are most helpful for search, far ahead of TikTok and other alternatives.
- Business investment in TikTok marketing is cooling, with fewer small business owners planning to increase spend on affiliate marketing and challenges persisting in converting engagement into sales.
- Read more
Platforms
New no-code MMM planning tool from Google
- Scenario Planner lets marketers create and test budget scenarios and forecast ROI without needing data science skills or code. It uses Google’s Meridian Marketing Mix Model under the hood to turn complex analytics into real results you can act on.
- Unlike traditional MMM that mostly looks backward, Scenario Planner lets teams play with “what-if” budget allocations, simulate future spend outcomes and stress-test strategies before committing dollars.
- Insights are visualised clearly so marketers can quickly see projected ROI and efficiencies across channels, making it easier to optimise spend and justify decisions.
- Historically, MMM outputs were hard to operationalise without analytics expertise; Scenario Planner aims to close that gap so more organisations can make data-driven budget decisions.
- Read more
AI in digital
Platforms
Google is redesigning how links show up in AI search summaries
- In both AI Overviews and AI Mode, groups of source links will now display in a hover-activated pop-up on desktop with site names, favicons and short descriptions.
- Link icons will also be more prominent across both desktop and mobile AI responses to encourage users to click through to original content.
- Google says this new UI makes it “more engaging” and helps users access content across the web more easily, though it’s part of ongoing experimentation.
- The update is part of industry pressure around AI summaries reducing traditional link clicks — past data showed a notable drop in organic CTR when AI Overviews were present.
- Google has been iterating on link visibility over the past year, signalling it sees citation display as a design challenge to keep refining.
- Read more
AI in digital
Platforms
New AI feature in Google Search Console now live for everyone
- Google has officially rolled out the AI-powered configuration tool in Search Console’s Performance report, letting you ask natural-language questions and get instantly configured reports instead of doing manual setup.
- Instead of choosing filters, metrics, comparisons, and date ranges yourself, you can tell the AI what you want to see — it applies the right settings for you.
- The feature can handle metrics like clicks, impressions, average CTR and average position plus filters by query, page, country, device, search appearance and date.
- It currently only works in the Search results Performance report (not Discover or News), and because it’s AI, double-check the filters it applies before making decisions.
- This rollout reflects a trend toward more conversational, AI-assisted SEO analysis — reducing technical friction and letting teams focus more on insights.
- Read more
AI in digital
Platforms
ChatGPT has English bias, searching in English when you don’t
- ChatGPT often leans on English sources even for non-English queries: Peec AI’s analysis of over 10 million ChatGPT searches shows nearly half of the background research queries (fan-outs) are conducted in English, even when the original question was asked in German, Spanish, Polish, etc.
- Local brands can become invisible: Because those English background searches tend to surface global and English-language sources, dominant local companies in non-English markets often don’t appear in ChatGPT’s answers, even for relevant local queries.
- Majority of non-English prompts still include English fan-out steps: Across languages examined, most non-English prompt sessions included at least one English background search, meaning this isn’t an isolated quirk but a widespread pattern.
- Implications for AI visibility strategies: Marketers targeting non-English audiences should consider how AI systems’ reliance on English sources affects brand visibility and potentially adapt AI visibility optimisation — not just traditional SEO — to ensure local relevance is reflected.
- Read more
AI in digital
Platforms
Majority of ChatGPT citations come from the first 30% of content
- ChatGPT heavily favours the top of the page. 44.2% of citations come from the first 30% of content, meaning intros carry disproportionate weight — get your key insights and positioning up front.
- Burying key points reduces citation chances. Only 31.1% of citations come from the middle 40% of text, making deeply embedded product features ~2.5x less likely to be referenced.
- Conclusions still matter. 24.7% of citations come from the final third, especially summary-style sections just before the footer — but actual footers are largely ignored.
- The pattern is statistically rock solid. Analysis of 18,012 citations (validated across batches from 1.2M total) shows a consistent “ski ramp” distribution favouring top-loaded content.
- Within paragraphs, depth beats position. 53% of citations come from the middle of paragraphs — ChatGPT looks for the sentence with the highest information value, not just the first line.
- Read more
AI in digital
Platforms
Anthropic announces Claude Sonnet 4.6
- Claude Sonnet 4.6 is Anthropic’s most capable Sonnet-tier model yet, with smarter coding, enhanced computer use, stronger long-context reasoning, better agent planning, and improved output quality across design and knowledge work.
- The model introduces a 1 million-token context window (beta), letting it ingest entire codebases, complex documents or lengthy research in one go — a big deal for deep execution and long-term planning.
- Benchmarks show Sonnet 4.6 approaching or even matching flagship Opus 4.6 on many real-world tasks (coding, office workflows, agentic use) while staying at Sonnet-level pricing — a cost-effective choice for heavy users.
- It’s now the default model across Claude.ai and Claude Cowork for both free and paid plans, and accessible via API and cloud partners — meaning broad availability without extra pricing.
- Anthropic reports stronger resistance to prompt injection attacks and overall safer behaviour compared to prior Sonnet versions.
- Read more
Platforms
Cloudflare Launches Markdown for Agents Feature
- Websites can now serve content in markdown to AI agents.
- This avoids manual HTML parsing by AI systems.
- It reduces token usage and improves processing efficiency.
- The conversion happens automatically at the network level.
- It reflects a shift towards building for AI consumption, not just traditional search.
- Read more
AI in digital
The first AI Platform to Offer An Official Report
- Bing Webmaster Tools has introduced an AI Performance report in public preview, showing how often your site is cited in AI generated answers.
- The dashboard tracks citations, pages referenced and the queries that trigger AI mentions.
- This signals a shift towards measuring visibility within AI answers, not just traditional search rankings.
- Read more
No results found.
