The Store Clerk in Your Pocket

There is a peculiar moment in shopping when a thousand choices become less useful than one good question. A search box can locate a sofa. It cannot ask why the sofa must survive a dog, fit through a narrow stairwell and look tolerable next to an inherited lamp. Operator built its 2015 shopping service for that moment. Send a message, perhaps with a photo, and a person who knew the category would send back a shortlist. If one item felt right, a button marked “I’ll take it” turned advice into an order.

A switchboard for desire

Robin Chan and Garrett Camp developed the idea at Camp’s startup studio Expa. Chan had led Zynga’s Asia business; Camp had co-founded Uber. Their insight was less glamorous than it sounds: much of retail expertise sat idle in physical shops while people at home were trying to interpret product listings alone. Expa called Operator a “request network.” The phrase was ungainly but accurate. A shopper supplied an intention, and the system found someone able to answer it.

At first, that someone was explicitly human. A store associate or trained category specialist saw the request, researched options and replied with photos, descriptions and prices. Shoppers could browse a Discover feed of products and experts, but the heart of the service was the exchange. “Our goal is to help people find the right product,” Camp told Bloomberg. Operator was built for what Chan called high-consideration purchases: the things for which a mistake costs more than a few minutes.

The sequence explains the difference from a conventional marketplace. Amazon and eBay were excellent at selling an item already identified by name. Operator wanted the question before the item: which headphones for a child, which shoes from a photo, which gift for a person you know well but cannot shop for. It did not need to beat search on every purchase. It needed to be useful precisely when search produced too many plausible answers.

The person behind the bot

A human reply makes a product charming and expensive. Operator's answer was to let software handle the repetitive parts. By September 2016, Chan said bots were involved in about half the activity on the service. The app had spread beyond its own iOS interface to Facebook Messenger in the United States, with Android in beta. A chatbot could greet and sort a shopper; a specialist could spend time where taste, context or trust mattered.

That division of labor was practical rather than mystical. The questions that sound easy can be the most costly to answer badly. “Find me a leather sofa” can mean a hundred things. “Find me one that fits my room and my budget” asks for a person who understands the constraints. Operator sold the feeling that someone was attending to those constraints. The bot helped the conversation reach the right desk.

Delivery completed the argument. In late 2015 Operator tested UberRUSH for holiday purchases from San Francisco department stores. Uber later listed Operator among partners using its delivery API. The pairing was neat: Operator could locate the right thing, while another company moved it. Building a courier fleet would have added another business to an already complicated business.

The border was the opportunity

The original American pitch was convenience. The China pitch had a sharper edge: confidence. In 2016 Operator announced a $15 million Series B led by GGV Capital and plans for a Shanghai office under Yolanda Xue, formerly of Wish. It proposed Chinese-language access to American shopping experts and Western goods. For a buyer worrying about authenticity, language and unfamiliar brands, a competent person on the other end of a message could be more than a pleasant extra.

The proposed service would use WeChat and iOS in China. Chan framed it as borderless commerce. The phrase was ambitious, but the mechanics were plain: someone locally understood the buyer; someone in the US understood the products; Operator connected the two and managed the order. A cross-border transaction is a string of small uncertainties. Operator hoped conversation could remove enough of them to justify its place between buyer and seller.

The bill for being helpful

There was no advertised extra fee for the shopper. That made the offer easy to try and the economics harder to read. Operator discussed taking a share of sales, charging retailers and other possible merchant revenue. Experts could earn commissions. The public record does not show a settled mix of these streams. What it does show is a difficult balance: each useful answer consumed human time, while each small purchase offered limited room to pay for it.

The first friction, visible in contemporary reporting, was getting enough people to adopt a new shopping habit. After the US launch, Operator was still looking for viral growth. Its China move was a response to a stronger, more specific problem: authentic American products with guidance in a language buyers already used. That does not prove the move succeeded. It does show how a broad promise became a more focused market thesis.

A reader building something similar can copy the sequence without copying the scale. Start with a category where advice changes the purchase. Let the customer describe a problem in ordinary language. Route it to someone with real category knowledge. Make approval and fulfillment painless. Then measure whether the value of that advice covers the cost of giving it. The model is least convincing for a known product at a known price, where a search result is faster and cheaper.

Operator’s most durable idea was not the chat bubble or the bot. It was the proposition that shopping starts before a customer knows what to type into a search bar. A decade later, software can answer more of those early questions. The standard Operator set remains stubbornly human: did the customer get the right thing?