Reports

Digital Shift Q2 2025: Generative Engine Optimisation and agentic AI advances

Written by Neil Perkin | May 2, 2025, 2:34:31 PM

1. The case for Generative Engine Optimisation

A whistle-stop tour of major generative AI developments and updates included the launch of OpenAI’s autonomous Deep Research tool and Google’s more advanced Gemini 2.5 Pro with multimodal capabilities.

As Google upgrades its AI Overviews and tests a dedicated ‘AI Mode’ for search, studies have painted a gloomy picture of the impact of generative AI summaries on click-through rate, with one study by Seer Interactive and Ziptie.dev pointing to a 70% drop in CTR when an AI Overview was present on the results page.

These and other findings suggesting an increase in ‘zero-click’ searches have led marketers to examine how they can maximise their exposure in AI Overviews in order to gain much-desired visibility. The practice of optimising for these overviews – and other AI-enhanced search engines – has been dubbed ‘Generative Engine Optimisation’.

GEO tactics have considerable overlap with SEO best practices, with recommended measures including:

  • Creating in-depth, well-researched content
  • Implementing Schema markup
  • Citing credible sources
  • Using a conversational, digestible tone
  • And of course, regularly assessing your content’s performance in AI Overviews.

More resources and insight on SEO trends and best practice can be found in Econsultancy’s search marketing hub.

2. Leaps forward in GenAI interoperability

Model Context Protocols (MCPs), which standardise how applications provide context to LLMs, are promising to potentially alleviate the need for rebuilding and retraining models and open up new possibilities and use cases.

For brands, an MCP could contain information such as: brand tone, style guidelines, target audience, and dynamic data inputs like real-time customer data. In a marketing scenario this would potentially enable generative AI actions to be flexibly adapted to specific contexts, campaigns, brands, and customer segments.

Marketers can prepare for this shift with measures such as building context libraries, looking for tools that support MCPs or open APIs, and potentially upskilling team members in context engineering (the practice of structuring information rules to feed into AI systems).

Similarly, Google’s A2A (agent-to-agent) protocol promises to enable agents from different platforms and purposes to interact safely and productively, opening up new possibilities for seamless customer journeys, smarter campaign orchestration, and AI-powered partner ecosystems.

One example of AI agents already being pressed into service to automate rote tasks can be seen at Jellyfish, where AI agents are being used to conduct media-buying tasks, reportedly leading to a 65% reduction in campaign launch times and a 30% improvement in campaign performance, with costs going down by 22%.

Understandably, this begs the question of how junior marketers who used to cut their teeth on these tasks will develop much-needed skills to advance their careers. During the post-presentation discussion, Neil Perkin put forward that “even though AI agents are inevitably going to be able to operationalise and execute a lot of tasks and processes autonomously, you’re still going to need that human … oversight and strategic thinking.

“So I would like to think that even at the most junior levels, there’s still an entry path for junior-level people in the industry … It would be a mistake, of course, to cut off the flow of talent through the industry.”

One way or another, investing in skill development will be critical for organisations to support their teams through this transition.

3. Live commerce is taking hold beyond China

While unquestionably huge in China, live shopping – a blend of ecommerce and livestreaming – has struggled for years to gain meaningful traction in regions like North America and western Europe.

However, TikTok is shifting the tide in these regions, driving a surge in live shopping behaviour and sales, with more than 20% of US digital buyers predicted to make a purchase from a livestream in 2025 (Emarketer).

Dedicated platforms like Whatnot, which has reported that gross merchandise value (GMV) for its livestream sales surpassed $3 billion in 2024, and in-app livestreams conducted by retailers like Zara, have all contributed to the trend.

A ‘sign of the times’ can be seen in a recent partnership between QVC and TikTok that brings QVC’s shopping channel to TikTok in the form of an ‘always-on’ live shopping stream. This forms part of QVC’s plan to revitalise its brand beyond television – and demonstrates the potential of players like TikTok Shop to deliver sales volume.