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
Google to show more links within AI Mode
- Embedded Link Carousels (Desktop Only for Now): Google has introduced clickable carousels within AI Mode responses on desktop, showcasing multiple contextual source links right inside the answer to encourage deeper exploration. Mobile support is coming soon.
- Smarter Inline Links: The underlying AI models now intelligently determine where inline links should appear directly in the text—especially at moments when users are most likely to want to click and learn more.
- Expanding the Web Guide Experiment: Web Guide, a Gemini-powered feature that groups helpful links by topic, is moving beyond the "Web" tab in Search Labs and will start appearing on the main "All" tab for opted-in users when relevant.
- Read more
Help desk
Platforms
Google August 2025 Spam update
- August 2025 spam update rollout: Google launched its August 2025 spam update on 26 August around midday ET. It's a global update covering all languages and is expected to gradually roll out over several weeks.
- Quick-seen impact: Unlike some updates where effects only appear after a few days, this one generated ranking fluctuations in under 24 hours, with many SEOs noticing shifts immediately.
- What’s targeted — and what’s not: The update penalises sites breaching Google’s spam policies, but notably excludes link spam and reputation‑related abuses.
- Context & comparisons: This is Google’s first spam update since December 2024 and follows closely after the June 2025 core update – marking a more aggressive season for algorithm changes.
- Advice for marketers: Don’t rush changes mid‑rollout. Instead, keep an eye on Search Console trends, audit your content for compliance with spam rules, and double down on quality, E‑E‑A‑T (experience, expertise, authoritativeness, trustworthiness), and user-first SEO tactics.
- Read more
Data privacy
Tranche Two of Australia’s Privacy Act reforms now in consultation
- The Australian Government has kicked off targeted stakeholder consultation for the second tranche of Privacy Act reforms—building on the groundwork laid by the 2024 amendments.
- This follows the release of a 116‑recommendation review report in 2023 and the introduction of Tranche 1 reforms in December 2024.
- The next wave is expected to contain new rules on first and third-party data collection, an updated definition of personal information and the introduction of a ‘fair and reasonable test’.
- While the content of forthcoming reforms remains under development, the consultation signals serious momentum toward completing the overhaul of Australia’s digital‑age privacy framework. Though it remains to be seen whether the second tranche is still on track to be completed this year.
- Read more
Platforms
Google introduces Merchant API and sunsets Content API for Shopping
- Google has launched the Merchant API (August 2025) — a streamlined, simplified API replacing the older Content API for managing product data and Merchant Center accounts.
- Now generally available, it offers programmatic access to data, insights, and enhanced control over how you showcase products and business information across Google.
- This API serves both organic (free listings) and ad-related experiences, bringing everything under one interface for easier management at scale.
- Google will be sunsetting the old Content API for Shopping and it will be no longer in use after August 18, 2026.
- Read more
AI in digital
Marketing by numbers
AI models prefer product descriptions written by AI models
- A peer-reviewed study (PNAS) reveals that large language models (LLMs) consistently favour AI‑written content over human‑written text in head‑to‑head comparisons for product descriptions.
- Preferences quantified: When LLMs themselves generated the AI product descriptions, other models preferred those AI versions significantly more often than human raters did (89% vs 36%).
- “Gate tax” warning: The authors caution that as more commercial platforms use AI to filter or rank content, businesses may feel pressured to use AI tools just to stay visible – creating an unfair advantage for AI‑assisted creators.
- Caveats: The study’s human baseline was small (just 13 research assistants), and the pairwise test method doesn’t measure actual sales or real‑world impact. The cause of the LLM bias is still unclear, and further research is needed.
- Takeaway: With AI systems potentially favouring AI‑written copy, teams should treat this as a testable concern—not a rule. Keep human involvement for tone, veracity and brand voice, and prioritise customer experience over just optimising for AI preferences.
- Read more
Data privacy
Marketing by numbers
Data privacy a major concern for Aussies shopping online
- Data privacy a major e‑commerce friction point: According to the 2025 Digital Commerce Imperative report by Publicis Sapient, 42% of surveyed Australians named data privacy as a top concern when shopping online.
- Data privacy second-most concern for Australians: When shopping online, Australian’s cited data privacy as their second biggest concern (33%), only after customer service issues (39%).
- Australian’s more concerned than other countries: When it comes to data privacy the survey found that Australian’s were the most concerned (42%), before the US (38%), France (35%), UK (30%), and Germany (28%).
- Unfortunatley, we cannot share the report here. Contact us to receive a copy.
AI in digital
Platforms
Upgraded AI Mode rolls out to more countries
- From just a few countries, AI Mode is now rolling out across 180+ countries and territories, though still limited to English.
- It’s not available in Australia yet but it is now across much of APAC, including Hong Kong, India, Japan, Malaysia, New Zealand, Philippines, Singapore, and Vietnam.
- AI Mode is also getting smarter. Now not just summarising, it acts for you: it can find and even book dinner reservations based on your preferences.
- For users who opt into the Labs experiment (especially U.S.-based Google AI Ultra subscribers), AI Mode taps into past activity—like previous searches, Maps history and prior conversations—to serve up tailored results, starting with food and dining but expanding soon.
- Read more
AI in digital
Data privacy
Marketing by numbers
61% of Aussies trust AI‑driven product recommendations
- AI shopping assistants gaining ground: Nearly half (48 %) of Aussies have used AI assistants for product searches while shopping online, climbing to 66 % among those under 45.
- Mainstream momentum and trust: 78 % expect AI shopping tools to go mainstream, and 61 % trust AI‑driven product recommendations—suggesting growing confidence in AI‑based retail.
- Concerns linger, especially on privacy: A whopping 92 % of respondents have concerns about AI in shopping, with 64 % citing privacy and data security as the key issue. Still, 57 % would share their shopping habits for the sake of personalised AI recommendations.
- Read more
Platforms
Language targeting for search campaigns to be automated
- By the end of 2025, Google Ads will eliminate manual language targeting for Search campaigns — instead relying entirely on AI to detect and serve ads in the language users understand.
- This shift only applies to Search campaigns — language targeting remains manual for Display, YouTube, Shopping, and other formats.
- Google’s move is a further step toward automation-first campaign setups, simplifying things for advertisers but reducing direct control. Marketers will need to monitor performance extra carefully to catch any misfires or mismatched language placements.
- Feedback from the PPC community is mixed: some are hopeful this will streamline campaign management, while others worry about AI going rogue — mismatched languages, diluted targeting precision, especially in multilingual or niche markets.
- Read more
AI in digital
Marketing by numbers
Platforms
ChatGPT referrals drop 52%
- Referral traffic from ChatGPT has dropped 52% since 21 July 2025, according to Profound’s analysis of over 1 billion citations and 1 million referral visits.
- At the same time, Reddit citations surged 87% (over 10% of all ChatGPT citations), Wikipedia jumped 62% (nearly 13% citation share), and together with TechRadar, these three now account for 22% of all citations — up 53% in just one month.
- This decline wasn’t triggered by the release of GPT‑5 (launched 7 August 2025). The shift began earlier, suggesting OpenAI manually reweighted its retrieval system to favour “answer-first” sources.
- As a result, branded or conversion-oriented websites—the ones pushing demos, sign-ups, or sales—are receiving fewer citations and thus fewer referral clicks.
- The takeaway? With this new emphasis on answer-first content, brands must pivot to prioritising clear, informative, user‑first content if they want to remain visible within ChatGPT-driven discovery.
- Read more
AI in digital
Marketing by numbers
Platforms
Referral traffic from Google drops 25% from AI Overviews
- Referral traffic from Google Search to premium publishers fell by up to 25%, with a median year‑on‑year decline of around 10 %. News sites dropped nearer 7 %, while non‑news brands dipped around 14 %.
- Broader studies show sharp drop‑offs in clicks and engagement:
-
Users clicked links under AI Overviews only 8 % of the time, versus 15 % without them—nearly halving click‑throughs.
-
Ahrefs found average click‑through rates were 34.5 % lower when AI Overviews appeared.
-
Authoritas reported that top‑ranked sites could lose up to 79 % of their traffic when overshadowed by AI Overviews.
- Google maintains that aggregate click volumes haven’t dropped and criticises these studies as methodologically flawed. Nonetheless, independent data underscores a clear trend of publishers losing visibility.
- Read more
AI in digital
Marketing by numbers
Platforms
The Future of Search: In Marketing We Trust Report
- The way people search and buy online is changing. Younger generations, including Millennials and Gen Z are looking to social media and AI for their product research. We took a deep look into the future of search. SOME key insights include:
- 67% of Gen Z prefer social media to search for local results. New search platforms mean new optimisation approaches are needed.
- -14% YoY traffic drop for Google.com worldwide (26 billion visits). Google is losing its hegemony yet still represents 91% of search traffic.
- 700M weekly active users of ChatGPT as of August 2025. OpenAI and other AI platforms continue to grow.
- 82% of consumers use social media for product research, with YouTube (52%) being the most used, with Facebook closely following (45%), and Instagram (38%).
- Amongst US shoppers, AI tools were preferred when it comes to product comparisons and recommendations. However, for general searches and initial product research, respondents preferred traditional search (Google, Bing).
- Download report
AI in digital
Marketing by numbers
Platforms
Study finds ChatGPT adoption increases use of Google
- A study of 260 billion rows of clickstream data between January 2024 and June 2025 to study the Google Search behavior of users before and after their first interaction with ChatGPT found Google useage increased after ChatGPT.
- Google usage didn’t dip after people adopted ChatGPT. In fact, it slightly increased—so both tools are being used side by side.
- Even long-term ChatGPT users didn’t shift away from Googling—it turns out ChatGPT compliments search habits rather than replacing them.
- For CMOs, that means keep your SEO game strong, add AI‑search tactics, and watch your marketing mix to stay ahead.
- Read more
AI in digital
Platforms
Study of public GPT chats discovers the average prompt is 42 words long
- Users are prompting ChatGPT like collaborators, not search queries – the average prompt was 42 words long, showing people aren’t just searching—they’re engaging AI to complete multi-step tasks.
- Commands rule the day – 75% of the prompts weren't questions but requests for actions, signalling a clear move towards AI as a task-doer rather than just an answer finder.
- Highly specific, localised commercial queries are emerging – prompts often include high-intent product or service searches tied to a city (think “Shanghai high-temp filters” or “Guangzhou mooncake prices”), highlighting a gap in current SEO strategies.
- Persona-based prompts are on the rise – users are asking AI to "act as" experts like food critics, mental health advisors, or recruiters, shaping response tone and style.
- Read more
Platforms
New Conversions with Cart Data Diagnostics tool in Google Ads
- Google Ads has introduced the “Conversions with cart data diagnostics” tool, aimed at helping advertisers pinpoint and fix issues in their product‑level conversion setup—such as missing item details or mismatched IDs—with the goal of ensuring accurate campaign performance.
- The tool assesses conversion tracking health with clear labels—Excellent, Good, Needs attention, or Urgent—and provides actionable alerts to help maintain data integrity before campaign performance suffers.
- Cart-level data (like item ID, price, and quantity) powers richer insights and sharper campaign optimisation—but if this data is incomplete or misaligned, it can lead to underreporting and poor targeting. This diagnostic feature aims to prevent those pitfalls.
- Read more
AI in digital
Platforms
Gemini and Claude can now remember you
- Catching up with ChatGPT, Gemini and Claude have announced new abilities for the AI tools to remember chats.
- Gemini has a new setting that allows it to learn from your past conversations over time. When this setting is on, Gemini remembers key details and preferences you've shared, leading to more natural and relevant conversations.
- Claude can now reference past chats, so you can easily pick up from where you left off. Once enabled for your account you can toggle on in settings.
- Read more about Gemini and/or Read more about Claude
Platforms
Perfomance Max Campaigns get an upgrade
- Campaign-level negative keyword lists are now available for Performance Max campaigns—making it easier to exclude unwanted search terms across multiple campaigns with a single list.
- The limit for search themes per asset group has increased from 25 to 50, helping expand your reach with more relevant search term coverage.
- You can now access search terms reporting for all Performance Max campaigns (including via API), giving more transparency into how search queries are performing.
- Enhanced demographic targeting has been introduced—including more granular age targeting, new gender-based controls (currently in beta), and device targeting to fine-tune which devices your ads reach.
- Read more
AI in digital
Platforms
Google preparing to roll out ads in AI Mode
- Google is preparing a broad rollout of “AI Mode” ads before Q4, pitching brands and agencies on how they differ from traditional search, with placements in Google Search, Google Shopping, and Performance Max campaigns.
- Ads will be text and shopping formats, available in English in the US on mobile and desktop, and are still officially marked as an “experiment.”
- To run ads in AI Mode, AI-powered targeting tools are required, including broad match for Search, AI Max (beta), PMax, and Shopping Ads — with an emphasis on “feed hygiene” to keep product catalogues updated.
- Google says ad targeting will factor in both the user’s search query and the AI Mode’s generated content, offering potentially deeper contextual relevance.
- Contact us for more information.
AI in digital
Platforms
OpenAI launches ChatGPT-5
- GPT‑5 launched on 7 August 2025, replacing all earlier models like GPT‑4, GPT‑4.5, GPT‑4o, and the o-series—it's now the flagship model accessible via ChatGPT and API platforms.
- Unified system with built-in “thinking”: GPT‑5 employs a smart routing system that automatically opts for fast answers or deeper, “reasoning” responses depending on the task—no more manual model selection.
- Significant performance leap: Delivers PhD-level capabilities across writing, coding, health, reasoning, and multimodal tasks (text, code, images, even long‑context coherence up to 256K tokens).
- Mobile and cost‑efficient variants: Users can access scaled-down versions (GPT‑5‑mini, GPT‑5‑nano) that are cheaper and faster—Free, Plus, Pro, Enterprise, Edu tiers have different features or limits.
- Read more
AI in digital
Data privacy
Insights from IAB Australia’s Data State of the Nation Report 2025
- The majority of survey respondents said they were only “somewhat prepared” for tranche 1 Australian Privacy Act changes. Only 28% are “very prepared”.
- There are a range of changes companies are making because of the new or impending further changes to privacy legislation. The most common change respondents are making is increasing the emphasis on the use of first-party data and increasing the use of AI or machine learning solutions.
- Survey respondents from across ad buy and sell sides see new privacy legislation impacting some key aspects of advertiser media plans. Over 80% say the data mix and measurement approaches are aspects where there is at least some impact. 74% say these changes are having at least some impact on personalisation tactics.
- Contextual targeting, data clean rooms and DMPs are the most adopted tools already implemented to manage the changes from new privacy legislation. The implementation of data cleans rooms has increased significantly on last year.
- Only 32% of advertising decisions makers/ influencers surveyed have deployed AI in some use cases or have it fully operational. A further 51% are in exploratory or testing phase.
- Contact us for a copy of the report.
AI in digital
Citation volatility in AI search is massive
- Citation Drift is Massive – Each month, about 40–60% of the domains cited by AI platforms like Google AI Overviews, ChatGPT, Microsoft Copilot, and Perplexity change—even for identical questions from June to July 2025.
- Longer-Term Instability Grows Sharper – Comparing January to July, citation turnover balloons to 70–90%, illustrating that AI search patterns are even more erratic over time.
- Traditional SEO Tracking Won’t Cut It – Unlike standard search (which might see 4–5 core algorithm updates annually), the inherent instability of AI citation patterns renders one-off snapshots or monthly checks nearly meaningless.
- New Monitoring Mindset Required – To navigate this volatility, brands must adopt continuous data collection (daily or weekly), aggregate multiple data points for statistical significance, and emphasise tracking directional trends rather than chasing individual data points.
- Volatility is Built-In, Not a Flaw – The drift isn’t a glitch—it’s how probabilistic models operate. Embracing this volatility lets marketers uncover real patterns rather than fight noise.
- Read more
AI in digital
Platforms
Insights from 50M ChatGPT prompts
- Generative is Now the Dominant Intent: 37.5% of prompts are generative (e.g. “write X”), overtaking informational intent (32.7%). Users want direct outputs, not just answers.
- Collapse of Navigation, Rise of Transactions: Navigational intent drops from 32.2% (traditional search) to 2.1%. Users no longer “go” anywhere.Transactional intent rises from 0.6% to 6.1%, as users start and progress through purchase decisions inside AI chat.
- Emergence of ‘No Intent’ Searches: 12.1% of prompts are conversational, e.g. “thanks” or “make it funnier” — a new behaviour layer with no precedent in SEO.
- SEO Metrics No Longer Apply: Clicks, traffic, and ranks are being bypassed. AI search happens within the interface, meaning landing pages and funnels are skipped entirely.
- New Strategic Focus: Citation over Ranking: Brands must shift from optimising for visibility to optimising for influence. Being cited and recommended by AI is now more valuable than ranking on Google.
- Urgency to Adapt: The shift is accelerating. Traditional SEO funnels are being replaced by AI-mediated discovery and decision-making. Delay means falling behind.
- Bottom Line: Search is the experience now. To stay relevant, brands must optimise for how AI references and recommends them — not how users click through search engines.
- Read more
Data privacy
Platforms
Advertisers targeting EU uers without Consent Mode V2 to lose tracking
- Google has begun strict enforcement of its EU User Consent Policy: as of 21 July 2025, any site or app targeting users in the EU that hasn't adopted Consent Mode v2 will lose access to conversion tracking, remarketing audience building, and both personalised and non‑personalised ads.
- Advertisers not implementing Consent Mode v2 are experiencing declining ad performance, as Google’s machine learning systems lack consent signals to optimise ad delivery.
- Consent Mode v2 enables continued measurement using aggregated modelling of conversions when users withhold consent, but not implementing it means you lose key performance insights.
- Read more
Data privacy
Platforms
Safari’s Advanced Fingerprinting Protection to be default, plus utm tracking disabled
- Safari’s Advanced Fingerprinting Protection will now be enabled by default for all browsing sessions, not just Private mode, making fingerprint-based tracking far less reliable.
- The browser injects benign noise and randomises key fingerprinting APIs (e.g. screen dimensions, Web Audio, canvas, CPU cores) per session and per site to thwart tracking efforts.
- Link Tracking Protection has been expanded: Safari will now strip out common tracking parameters (like utm_source, gclid, fbclid) from URLs and referrer data even during standard browsing.
- These changes significantly limit data brokers and advertisers from building persistent profiles or linking browsing behaviour to campaigns across platforms.
- Read more
AI in digital
Platforms
AI Mode is now live in the UK
- AI Mode is now live in the UK, rolling out from 28–29 July 2025. Look for the “AI Mode” tab on desktop searches and in the Google app (Android & iOS).
- Powered by Gemini 2.5, this mode breaks down complex, multi-part queries using a “query fan‑out” technique and delivers conversational, structured answers instead of traditional link lists.
- Supports multimodal input—text, voice or image—and enables follow-up prompts so users can refine queries without starting over.
- UK regulators are watching closely: the Competition and Markets Authority may designate Google with “strategic market status” by October 2025, which could impact how AI Mode functions especially in relation to publishers.
- Read more
Platforms
TikTok now enables advertisers to track users on websites after clicking an ad
- TikTok has launched a new “Engaged Session” tool, allowing advertisers to track users who spend at least 10‑seconds on a website after clicking an ad—even without a pixel.
- Provides deeper post‑click insights, delivering metrics like Total Engaged Sessions and Cost per Engaged Session that align with Google Analytics and Adobe Analytics standards.
- Integrated within TikTok’s Traffic Objective, and available across all ad placements—including Search Ads—so you can activate it across your campaign mix.
- Read more
AI in digital
eCommerce sites need to test if they’re compatible with AI agents
- Google’s John Mueller suggests ecommerce sites need to be tested for compatibility with agentic AI agents—like ChatGPT—that shop on behalf of customers.
- A recent experiment evaluated the top 50 Swiss ecommerce stores and found most were accessible, but some were blocked by CAPTCHA, Cloudflare Turnstile, maintenance pages or bot‑defense tools.
- Mueller’s advice: “Pro tip: check your ecommerce site to see if it works for shoppers using the common agents…” as increasing numbers of customers may use agentic AI to shop.
- As agentic AI becomes more mainstream, brands risk losing sales if their sites block these agents—so testing now is vital for future‑proofing online performance.
- Read more
AI in digital
Platforms
Microsoft introduces Copilot Mode in Edge
- Copilot‑First Browsing Comes to Edge: Microsoft has launched Copilot Mode, an experimental AI‑driven interface in Edge that replaces the traditional new tab with a single input field merging chat, search and navigation—all with user permission.
- Context‑Aware Web Assistance: With user consent, Copilot can access open tabs to summarise content, compare options (like for shopping or research), and offer tailored suggestions across pages.
- Voice Control & Smart Actions: You can interact with Copilot using voice commands or typed prompts to perform actions like navigating videos, planning bookings, or converting data formats hands‑free.
- Privacy by Design, Opt‑In Trial: Copilot Mode is optional, free for now, and built with Microsoft’s privacy safeguards—data access is transparent and always requires permission. It may later shift to a subscription model.
- Read more
AI in digital
Marketing by numbers
Platforms
Top result in Google CTR has fallen 32%
- Top organic CTRs are dropping fast: Google's top result CTR has fallen 32% (from 28% to 19%) since AI Overviews expanded—showing users are clicking less on traditional results.
- Middle positions gaining ground: Positions 6–10 saw a 30.6% CTR increase, indicating users may be bypassing AI summaries to find more authentic sources further down the page.
- Ecommerce and publishers hit hardest: MailOnline reports CTRs plummeting up to 56% on desktop when AI Overviews appear; product-related widgets are also eating into ecommerce traffic.
- Read more
Platforms
Google launches Branded Searches as new conversion metric in Ads
- Google has launched “Branded Searches” as a new conversion metric, tracking how often users search for a brand on Google or YouTube within 30 days of seeing an ad.
- It’s available for YouTube, Performance Max (across YouTube, Discover and Gmail), and Demand Gen campaigns.
- This metric gives marketers a measurable signal of brand interest and consideration, bridging upper‑funnel ad exposure and consumer action.
- Right now it’s reporting‑only (not used for optimisation), though Google hints that optimisation via bidding strategies may be introduced later.
- Make sure your brand mapping is correctly set up in Ads to accurately attribute searches as branded conversions.
- Read more
