Guided purchase conversations for products that need explaining

Enterprise AIeCommerceHow Mitochondria Works

Mar 22

Baymard Institute's research puts average online cart abandonment at about 70%. For products that need explaining, the loss starts earlier. Specialist food, health and wellness products, technical equipment and premium goods sell when a buyer understands what they are choosing and why it suits them. Most online stores offer a catalogue and a search box. Some add a chat window. Few offer what a capable person in a shop provides: guidance through the decision.

Answering questions is not guiding a decision

eCommerce chat has improved. Language understanding is better, catalogue integration is deeper, and many systems handle visual search and several languages. But most still treat a purchase as information retrieval. The customer asks; the system answers.

Purchase decisions rarely form that way. Buyers move through discovery, education, comparison and reassurance before they convert. Each stage needs a different kind of exchange. A first-time buyer needs explanation. A returning customer wants speed. A health-conscious shopper needs claims that stay within what is supportable. A chat window with one voice and a FAQ behind it flattens all of this.

What communication science says about buying

Petty and Cacioppo's Elaboration Likelihood Model describes two routes that work together in most purchase decisions. On the central route, the buyer weighs the product on its merits: specifications, ingredients, fit with their needs. On the peripheral route, they respond to cues: how credible the guidance feels, how the brand presents itself, how easy the exchange is. Accurate information matters. Whether it leads to a purchase often depends on the surrounding cues.

Cialdini's work on source credibility adds a second point. The same answer is received differently depending on who delivers it. A nutrition question answered with a nutrition specialist's knowledge and register carries more weight than the same facts from a generic shopping assistant.

Ajzen's Theory of Planned Behaviour adds a third. Buyers act when they believe they can complete what they set out to do. Every effortful step lowers that belief. Every step the system removes raises it.

One conversation, several kinds of expertise

Applied to commerce, these findings suggest a clear design. One conversation draws on the expertise each stage needs: a specialist who explains the product, an adviser who relates it to the buyer's situation, and post-sale support. Each has its own knowledge, register and guardrails. A nutrition specialist does not make sales commitments. A product adviser does not overstate compatibility or performance.

The whole journey runs in one thread: discovery, comparison, adding to cart, and post-delivery support. Checkout happens on the store's existing payment system. When the buyer returns, the conversation picks up their history, so a repeat order can start with "the same again, or something new?"

The conversation also runs on channels buyers already use, such as WhatsApp, and in their own language, written natively rather than translated from English.

Knowing when to hand over

What a system will not attempt matters as much as what it will. Product guidance, policy questions and routine support sit within its scope. Complaints that need judgement, questions outside the configured domain and any request for a person go to the brand's team with the full conversation attached. Brands can start with cautious boundaries and widen them as confidence grows. An honest handover builds more trust than a system struggling past its limits.

How to measure a guided conversation

Conversion of guided sessions against unguided ones.
Average order value in guided sessions.
Returns on orders placed through a conversation.
The questions that most often come before a buyer leaves.
Repeat purchase rate among buyers who used the conversation.

How Mitochondria's agentic AI systems support specialist eCommerce

The approach. The configured sales unit (unit 58) is configured to each brand's catalogue, voice and claims policy, and works with the store it already runs, such as Shopify. The design draws on the founding team's decade of research and practice in persuasive communication.

What it enables. Buyers get guided, expert conversations from first question to checkout and after-sales support, at any hour, on the website or WhatsApp. Buyers can specify configurable products in conversation, and the system passes the result to the team as a verified order.

The impact. More of the traffic a brand already pays for becomes orders. Buyers return because each purchase is easier than the last. And every conversation adds to a record of what buyers asked, where they hesitated and what resolved it, which informs merchandising, content and product decisions.