Omnichannel is one of those terms that gets repeated so regularly that it can lose its meaning. The underlying idea, though, is that customers don’t shop in a single place, and the brands that win re those that can connect experiences, messaging and measurement across multiple touchpoints. However, what those channels are has changed.
Until recently, an omnichannel plan could be drawn with two big boxes. One for physical stores and one for websites. Today, however, social platforms are becoming storefronts, apps are becoming ecosystems, and large language models (LLMs) are beginning to act as discovery engines, shopping assistants and, crucially, transaction layers in one.
A meaningful shift
Earlier this year, Carrefour became the first European supermarket to enable transactions through ChatGPT. It was an early signal that conversational commerce is moving from concept to capability. It clearly demonstrates a meaningful shift where a customer’s shopping journey can start and finish inside a conversation. Rather than jumping from search results to a retailer site, a shopper can ask for dinner ideas, refine choices based on budget, and then build a basket without leaving the chat interface.
Savvy brands are realising they need to show up in these conversations. Retailers and platforms are already positioning themselves for an advertising layer in conversational experiences. Amazon’s new shopping assistant, Rufus, is an example of how quickly “assistant” can become “inventory,” with plans reportedly moving toward paid placements over time. OpenAI also quietly launched an ads manager for ChatGPT recently as it builds out its ad business. Meanwhile, consumers are increasingly asking LLMs for recommendations; often phrasing queries as “best for me” rather than “best overall.” That shift from keyword search to intent-rich conversation changes what it means to optimise for discovery.
In practice, people are using LLMs in three recurring ways. Firstly, there is exploration, where they ask for broad ideas and options. Then, narrowing where shortlists are produced that fit within given constraints. Finally, reassurance where consumers look for validating a choice prior to purchase. Brands that understand which stage they are influencing can tailor content accordingly.
What will hold the ecosystem together?
As the makeup of omnichannel becomes increasingly digital, superapps are being vaunted as the glue that could potentially hold the whole ecosystem together. But could the LLM become the glue? After all, it is an always-available layer that sits above existing digital channels. Only time will tell. But even if it does, it’s unlikely that brand and retailer apps will vanish any time soon. Apps still deliver unmatched value for loyalty, personalisation, push messaging, saved preferences and post-purchase service. Yet, there is no doubt that social commerce is gathering pace. In fact, some forecasts suggest TikTok Shop could become a top-three global retailer by 2030.
Whilst a modern purchase journey could well include a TikTok Shop, Reddit threads, ChatGPT or a brand app, physical retail is not going to disappear. In fact, we continue to see growth in store footprints, even for digital-native brands. Abercrombie & Fitch is just one good example. Its app detects when a customer walks into a store and switches to in-store mode. About an hour after they leave, it sends a push notification asking for feedback, timed so the visit is still fresh. Participation earns loyalty points, creating a simple loop that connects app, store, and CRM. Shein and TikTok Shop experimenting with pop-up formats, are reminders that whilst discovery may happen in-feed, trust and conversion can still be accelerated in-person.
Connecting the dots
As channels proliferate, one of the hardest problems for brands is how to connect the dots across fragmented journeys. When discovery happens in a chatbot, validation happens in a Reddit thread, and conversion happens in an app, traditional last-click reporting becomes less useful and can actively mislead budget decisions.
LLMs introduce an additional layer of complexity. Most brands can only observe LLM-driven traffic when it lands on owned properties. One solution is to treat AI search as a first-class input, just like SEO and onsite search. Teams can then compile the most common AI-driven phrases and prompts in their category and feed them into content and merchandising workflows so the brand’s answers remain consistent wherever the customer encounters them.
From an attribution perspective, the goal is to make such journeys visible enough to act upon. Start by ensuring analytics can identify traffic coming from LLM, then pair that with incremental measurement approaches so decisions don’t depend solely on click-level signals. Over time, as paid formats emerge in conversational interfaces, brands will need the same discipline they apply to search and social today. Consistent naming, clean landing experiences, and incrementality frameworks that hold up will be key.
What to do now
For marketeers, the immediate goal isn’t to overhaul their omnichannel plan, but to build readiness. New commerce surfaces such as TikTok Shop, conversational interfaces like ChatGPT, and community spaces like Reddit that shape preferences are adding steps to the journey while compressing the time between inspiration and purchase. This is nothing to fear, but everything to embrace. The winners over the next few years will be thosebrands that treat these shifts as a growth opportunity. Showing up with useful answers, creating connected experiences, and building robust analytics frameworks that can keep pace with how customers are making purchase decisions.
