Artificial intelligence is reshaping customer experience by connecting every stage of the omnichannel retail journey – from the first website visit to delivery, returns and post-sale service.
On a recent webinar hosted by Inside Retail premium and features editor Adam Thorn, two technology leaders outlined how AI is enabling what they describe as the “next frontier” of customer experience: One where discovery, purchase and post-sale engagement are orchestrated seamlessly across channels.
Federico Jalil, digital product manager at Freedom Australia, and Kevin Annfield, team lead of customer experience solutions at SAP, shared how data, automation and AI are transforming the way bulky-goods retailers approach omnichannel retailing.
“When you’re dealing with tens of thousands of products, technology becomes essential,” explained Annfield. “The goal is to give business users the flexibility to design and automate the right journeys at the right time, based on their understanding of customers and supported by data.”
Designing for infinite journeys
Freedom Furniture is a classic example of a retailer whose complexity challenge starts with its product mix.
“Freedom has a very wide variety of offerings, from candles at $30 to sofas at $8000, so the purchasing journey will vary depending on what mission the customer is on,” Jalil said on the webinar.
A mattress for the main bedroom is a very different purchase from a spare-room mattress bought ahead of visiting relatives. The former typically involves visiting a store, testing options and returning home to reflect before buying – often online. The latter may be a quick online order.
Rather than attempting to map and optimise every possible journey, Jalil said the retailer has focused on removing friction at every transaction, while leaving room for inspiration.
“It’s very hard to select a number of journeys to optimise,” he said. “So instead of doing that, our strategy is to make sure that wherever you’re purchasing, there is no friction, and there’s room for inspiration.”
The omnichannel foundation – ensuring customers can discover in-store and buy online, or vice versa – was laid before Jalil joined the business. Over the past two years, his focus has been scaling that experience as the product range expanded from about 12,000 online items four years ago to more than 70,000 today through a dropship program.
With that scale, manual curation becomes unworkable.
“We are dealing with volumes that are impossible for a human to manage,” Jalil said. “That’s why the company has adopted this AI-first culture, where we try things in a safe way, implement them, measure and monitor.”
AI beyond the chatbot
For Jalil, AI is often misunderstood as synonymous with generative chat interfaces. The reality is that, much of its impact sits behind the scenes.
Freedom’s search and product recommendation engine analyses multiple data points: Past purchases, browsing behaviour within categories, clickstream data and product attributes. Instead of simply recommending item Y to everyone who viewed item X, the system tailors results to the individual.
“If you’re shopping for a sofa, yes, you might get recommendations on cushions,” Jalil explained. “But the offering of cushions is thousands of cushions. So we are tailoring the cushion recommendation to the actual sofa that you’ve been looking at, and this is all done via AI.”
The results have been measurable. Comparing year-on-year performance before and after implementing the new engine, Freedom recorded months where conversion rates and transactions were up to 50 per cent higher among customers engaging with search.
Search results are now “incredibly tailored” not only to the keyword but to the customer’s intent, Jalil said.
From a commercial perspective, improved recommendations lift average order value. From a customer perspective, they reduce time spent hunting for complementary products. Crucially, the business has avoided over-optimising for speed to checkout.
In contrast to some fashion retailers that place a prominent checkout button directly on the product page, Freedom encourages browsing across ranges, fabrics and related items.
“We don’t want to rush the customer to the checkout,” Jalil said. “We want them to discover, click around, and when they’re ready, they can purchase frictionless.”
The data foundations
Annfield argued that effective AI-led personalisation starts with disciplined data management.
“It actually starts with the product information,” he said. “Having good, solid information about the products that are stored and displayed in the commerce journey is probably the most critical path.”
That information is combined with behavioural data: How long customers spend on list pages, how often they add to cart, and how they move through the site. These signals feed AI-based recommendation tools that surface relevant products at key moments in the journey.
The same principle extends to marketing automation. Using customer engagement platforms, retailers can consolidate data from online and in-store purchases, social engagement and channel preferences to build dynamic segments.
“Sometimes that might be a group of customers of one,” Annfield said.
Automation then delivers the right message through the right channel. A customer who consistently engages with SMS may receive updates that way, while another may respond better to email or in-app notifications. The complexity of those calculations, Annfield noted, exceeds what any human team could realistically execute at scale.
High-consideration purchases and the omnichannel reality
Furniture retail adds another layer of complexity. Customers frequently blend physical and digital touchpoints: Testing a bed in store, then purchasing online; or researching extensively online before visiting a showroom.
For bulky, higher-priced goods, expectation management becomes critical – particularly for made-to-order items with long lead times.
“We have products that we stock and products that are made to order,” Jalil said. “That’s where it is critical for us to stay on top of communication.”
Estimated delivery dates are communicated at checkout and reinforced with updates throughout production and shipping. Integration between order management systems and marketing platforms ensures that when an estimated arrival date changes, a trigger can reset expectations quickly.
“It starts with our integrations to make sure that the data is flowing from one side to the other efficiently and in a timely way,” Jalil said. “As soon as an update comes into our order management system, we can push that update to trigger a message to let the customer know there is a delay.”
At the same time, Freedom is careful not to overwhelm customers with excessive notifications. “It’s a balancing act,” Jalil said. “You don’t want to spam a person every second day.”
Annfield added that these updates rely on deep operational data, including supply chain timelines and variant-level lead times. A leather sofa may require a 12-week wait, compared to six or eight weeks for another material. If that information is not consistently carried through the journey, automated messages risk undermining trust.
Returns in a bulky-goods world
Returns are often cited as a structural challenge for online furniture retail. Unlike apparel, a sofa does not fit in a courier satchel.
Jalil does not view that as a barrier to e-commerce, but as a customer experience design problem.
For smaller homewares and dropship items, returns can be handled in-store or shipped back with relative ease. For bulky furniture, collection must be organised, often requiring coordination with delivery partners and careful communication.
“It is absolutely critical,” Jalil said. “That’s the decision-making point between having a one-time customer and a loyal customer.”
Freedom is investing in clearer self-service options, more transparent return processes and conversational AI tools that allow customers to check order status or initiate returns without waiting for a service agent.
From Annfield’s perspective, the key is connecting online and in-store systems so that an item bought online can be returned in-store with full visibility of the original transaction. Achieving that requires a single source of truth across commerce, point of sale and supply chain systems.
Analytics driving future decisions
Beyond customer-facing improvements, data is informing strategic decisions about assortment and inventory.
Freedom’s expanded dropship program provides real-time insight into demand trends and assortment gaps. Click data, add-to-cart rates and sales performance all feed into product and stock decisions.
The retailer has identified, for example, that increasing in-stock sofa options reduces friction for customers unwilling to wait for made-to-order items. In some cases, stocked sofas can now be delivered within two to three weeks.
“We’re using multiple data touchpoints to make sure the whole operation improves based on this information,” Jalil said.
For Annfield, the next phase is using AI not only to personalise the front end, but to optimise planning and replenishment behind the scenes. By combining commerce data with supply chain and operational data in advanced analytics environments, retailers can improve demand forecasting and restocking decisions.
A continuous evolution
Both speakers emphasised that none of these initiatives are one-off projects.
“It will constantly be a work in progress,” Jalil said of communication journeys and post-purchase optimisation. “There’s always room to continue to optimise.”
In that sense, AI is less a destination than an enabling layer – one that allows retailers to orchestrate thousands of micro-decisions across channels, products and customers in real time.
For bulky-goods retailers in particular, omnichannel is not optional, and online is not inherently ill-suited to large items. With the right data foundations and AI capabilities, even an $8000 sofa can be sold, delivered and supported in a way that feels coherent and considered.
