AITalk: The ecommerce journey in the age of the conversational consumer

Please register or log-in to watch this on-demand episode 

Views on news 

Fashion e-commerce has reached an ‘optimisation ceiling’, where filters, categories and static product pages are no longer enough to guide increasingly-intent-driven shoppers, according to new research from Firney.

 

The UK-based conversational search and hyper-personalisation specialist argues that many fashion websites are still built around rigid navigation structures that fail to reflect how customers actually shop, particularly on mobile where up to 75% of e-commerce traffic now sits and abandonment rates remain high. 

 

While Gen Z tends to start their shopping on TikTok, other generations, less trustful of social media sites and LLMs, still go to Google and brand pages to do so. While with Google, marketers know what the search words are, with LLMs they have no knowledge of what the conversational search in fact was. For brands, to have that conversation on their website could be bring about a step change.  

 

Major shifts in how customers engage with brands 

The e-commerce customer journey has recently become more fragmented. In response to the survey question about what consumers rely on when choosing a product, answers were distributed between product comparisons, reviews and confidence in what to buy – each at roughly 20 per cent.  

 

With AB Tasty’s tools, it’s possible to send personalised messages to would-be customers based on how they behave, leveraging real-time personalisation, where the point is to show the individual that they are being seen by, for example, showing different seasonal clothes based on their location. Overall customer experience including delivery and product return can also play a major role in whether customers remain loyal to the brand.  

 

Product discovery is gradually moving from navigation and filtering to conversations, making the orchestration of various touchpoints rather challenging. Following customer journeys closely over time and maintaining context will likely become a key ingredient of personalisation too. It’s important that data collection happens in line with data protection rules, which is less of a challenge when personalisation relies not on personal but behavioural data – once the customer has accepted cookies.

 

Currently, agentic AI, however, is more extensively used for re-engaging with customers than converting them. Although there is still no recipe for what brands should do to show up in AI-searches, content creation, key keywords and old-fashioned PR activities still matter, while external sources – newspaper articles, Wikipedia or reviews – are becoming increasingly instrumental too. The metrics marketers use to measure the effectiveness of these methods is also a mix of old and new.  

 

The panel’s advice 

• Channels like WhatsApp, previously owned by customer service teams, are shifting to marketing and tech teams with the rise of agentic AI. Whoever is involved, ensure that there is one particular team owning each channel.  
• LLMs look for brands in their answers that have authority on a particular topic – product data and information on third-party websites should also be structured accordingly.  
• Look at where traffic is coming to your site. Experiment with what messages work best at these hot spots.