by Dana Lee on 22nd Apr 2026 in Video Interviews

In this episode of A Coffee With... Gemma Spence, CEO of VML Marketplaces (WPP Group), joins ExchangeWire COO Lindsay Rowntree for a coffee chat about AI and agentic commerce.

Gemma explains how awareness is shifting towards metrics like share of model and share of mention in AI-generated answers, moving from keyword- to prompt-based strategies. Their conversation delves into advertisers' concerns around agentic commerce, its impact on the consumer decision journey, and how brands can optimise for LLMs and measure outcomes.

Success requires augmenting traditional marketing with a data-first approach: prioritising high-quality structured data and engaging in extensive testing and prompt engineering to decode AI behaviour. While the field is still maturing, early, structured adoption focused on clear use cases is key.

What is Agentic Commerce?

Agentic commerce marks a shift in online shopping where AI-powered agents act as "personalised shoppers", interacting with retailers and brands on behalf of consumers. This creates machine-to-machine interactions that fundamentally alter traditional product discovery and purchasing dynamics.

Gemma highlights 3 'Cs' of advertisers' fears around agentic commerce:

1. Control: How to influence shopping decisions made by an agent
2. Clarity: Limited transparency into how AI models like Rufus or Gemini prioritise and rank products
3. Commoditisation: Fear that brand equity erodes as functional, data-driven factors dominate purchasing decisions

The purchase funnel persists but is augmented. While core stages remain, influence tactics are adapting to AI-driven discovery. With AI behaviour shaped by numerous opaque factors, optimising for agentic commerce is complex and test-heavy. There is no silver bullet; brands must conduct continuous testing and prompt engineering to uncover what works.

Strategies for Success in Agentic Commerce

The first step is ensuring complete, high-quality structured data: consistent product naming, feature-led bullet points, and clear price-pack architecture. If this foundational "digital shelf" data is weak, subsequent optimisation efforts will underperform.

Brands also need a systematic testing methodology to analyse category dynamics, competitor actions, and key levers like price and promotions. Through extensive prompt engineering, brands can see how their products surface versus competitors and identify the signals driving AI recommendations.

Finally, campaigns must both engage human emotion and satisfy machine logic. Brand creative and salience are still essential for end-users, but the immediate "customer" is the AI agent. Brands should fuse brand-building with a functional focus on structured data, authority-building content, and media overlays to ensure visibility within the AI decision path.

The Future of Measurement and Adoption

Although traditional metrics like ROAS and impression share still matter, measurement is evolving to emphasise share of model (a brand’s prevalence within specific AI models like Amazon’s Rufus) and share of mention (frequency of being recommended in AI-generated answers). The aim is to be the recommended choice – lack of mention means lack of consideration.

Adoption is bifurcated. Fast-movers like fashion, beauty, and consumer electronics are leading by actively testing and setting standards. Meanwhile, many CPG brands remain in "watch and see" mode due to complex B2B structures and slower ROI validation.

Regardless, agentic commerce is expected to be a disruptive force across all sectors. According to Gemma, building AI readiness will not be one person's job, but every person's job.

Agentic AIAgentic Commerce