Ecommerce and retail phone support

Retail and ecommerce phone calls are repetitive and relentless. Where's my order. Can I return this. Are you open on Sunday. Do you have this in stock. Each call is simple, but the volume is crushing — and every one that goes unanswered is a customer who now trusts you a little less. Voicemail doesn't cut it; people expect an answer when they call.

Agent Vani covers that line as a fast, friendly support voice. It answers on the first ring and talks with the caller in real time. This is a two-way, two-way call — the caller can interrupt, change the question, and talk over a long answer, and the agent adjusts instantly. It is not a menu tree, and it is not a chatbot reading scripted turns down the phone. It handles the repetitive questions quickly and accurately, and it captures the ones that need a person.

The focus is inbound support. When customers call, the agent covers the line so nothing goes to voicemail during peak, after hours, or on the day your one support person is off. It resolves the common questions and routes the rest, so your team spends time on the genuinely tricky cases instead of reading the returns policy aloud for the fortieth time today.

What the line handles

Most retail and ecommerce calls cluster around a handful of topics. Handled well, the caller gets an answer in seconds:

• Order status. Where an order is, whether it's shipped, expected delivery, and what to do if it's late — captured and routed when it needs a look-up your team must do.
• Returns and exchanges. Your returns window, how to start a return, refund timelines, exchange rules, and what's non-returnable. This is high-volume, factual, and perfect for the agent.
• Store hours and locations. Opening times, holiday hours, which store, parking, and directions.
• Stock and product basics. Whether something's generally available, sizing and variant basics, price, and where to buy — the factual things a knowledgeable assistant would know.
• Shipping and policies. Delivery options, costs, timelines, and the policies customers ask about repeatedly.
• Getting to a person. For a specific account issue, a complaint, or anything the agent can't resolve, it hands off cleanly.

What to put in the knowledge base

The agent answers from the knowledge you give it, so keep it factual and current. For a retail or ecommerce line, include:

• Returns and refunds policy in plain, spoken language: the window, how to start, timelines, exceptions.
• Shipping and delivery: options, costs, timelines, and what happens when something's late.
• Store hours and locations with exceptions written out — holiday hours, seasonal changes.
• Product and stock basics you can state confidently: general availability, sizing guidance, price ranges, variants.
• Common policies: price matching, warranty basics, gift returns, whatever customers ask about.
• Handoff rules: which situations go to a person, and how account-specific look-ups are captured and routed.

Keep fast-changing details — live stock counts for a single item, individual order internals — out of the static knowledge; route those to a person or capture them for follow-up so callers never hear a stale answer stated as fact. The script-writing guide covers how to phrase spoken answers well.

A typical call, start to finish

It's the middle of a sale week and the line is busy. A call comes in and the agent answers with your greeting — the store's name, a friendly tone, in the caller's language.

"Hi, I want to return a jacket I bought last week — how does that work, and are you open today?" Because the call is two-way and realtime, the agent doesn't force the caller through separate scripted turns. It explains the returns process and window directly, confirms today's hours in the same flow, and — if the caller wants to start the return — captures what's needed and routes it into your process.

Mid-answer the caller cuts in: "Actually, can I just exchange it for a bigger size instead?" The agent adapts on the spot, explains the exchange rule, and captures the request. The caller hangs up in under a minute with a clear answer and a next step, without waiting on hold.

For an order-status call that needs a look-up only your team can do, the agent doesn't guess. It captures the order details and the customer's contact information and routes it, so the follow-up is quick and accurate. This is the shape of good inbound support — resolve what's answerable, capture and route the rest. See the inbound support use case.

Escalation and human handoff

The agent should know when to hand off. You configure the rules:

• Account-specific issues requiring a look-up your team must do → captured with details and routed, not guessed.
• Complaints and unhappy customers → routed to a person quickly; a frustrated caller wants a human.
• Anything outside its knowledge → an honest "I don't have that, let me get someone" beats a confident wrong answer.
• Caller wants a person → honoured immediately, no loop.

Handoff can be a live transfer, a captured callback with the details attached, or a message. The agent resolves the repetitive questions and routes the rest cleanly. See the human handoff guide.

Persona, hours, and numbers you control

You configure everything in a free control plane — no charge to set up or tune the agent. You pay only for the minutes it spends on calls. See pricing for how minutes work.

• Persona. Set a voice that matches your brand — upbeat and casual, or crisp and efficient. You choose the greeting, how it names the store, and the default language.
• Hours. Decide when it answers: as overflow during peak when staff can't pick up, after hours so no evening call is lost, or all day. You set what it says outside opening times.
• Knowledge. The policy and product base above, edited whenever returns rules, shipping, hours, or ranges change.
• Numbers. Connect the number customers already call, or a dedicated support line. See product.

Start small — after-hours or overflow on one line — and widen its hours as you trust it.

Failure modes and how to avoid them

• Stale policy. An out-of-date returns window or shipping timeline in the knowledge sends customers wrong information with confidence. Update policy the moment it changes, especially around sales and holidays.
• Guessing on orders. The agent should never invent an order status it can't verify. Configure it to capture and route account-specific look-ups instead of guessing.
• Trapping unhappy callers. Complaints must reach a human fast. Make handoff immediate for anything sensitive.
• Live-stock claims. Don't put fast-changing single-item stock counts in the static knowledge; let those go to a person so callers never hear a stale "yes, in stock."
• Robotic pacing. An agent that can't be interrupted feels like a machine and customers hang up. the two-way, interruptible design prevents this — but test it on real calls first.

Call the line yourself before customers do. Interrupt it, switch a return to an exchange mid-sentence, ask about an order it can't see, and confirm the handoff works. A trial with no card required lets you do this at no cost.

Getting started

Start a trial — no card needed — and configure the agent in the free control plane. Write honest support knowledge: returns and refunds, shipping, hours, product basics, and handoff rules. Keep fast-changing internals out and route them to a person. Point the agent at an overflow or after-hours line, test it hard on real calls, and widen its hours once it consistently resolves the common questions and routes the rest cleanly. You pay only for the minutes it talks, and you stay in control throughout.

The payoff is a line that's always answered by a fast, accurate, friendly voice; the repetitive questions handled in seconds; the tricky cases captured and routed; and a support team that finally gets to focus on the customers who actually need a person.

Why a live, two-way call changes the outcome for retail

Retail callers are usually mid-task and a little impatient. They're standing at a returns desk, holding a package, or deciding whether to buy right now. The difference between a two-way, realtime call and the turn-based experience most people have suffered is the difference between a customer who stays and one who hangs up.

On a turn-based system — press one for this, wait for the beep, hear a scripted line read back — the caller has to bend to the machine's rhythm. They can't interrupt, they can't correct a wrong turn, and they can't ask the follow-up that's actually on their mind without starting over. Retail callers won't tolerate that. They abandon the call and, often, the purchase.

a two-way conversation flips the rhythm back to the caller. They can start explaining a return and, halfway through, decide they'd rather exchange — and the agent simply adapts, because it's listening the whole time, not waiting for its turn. They can talk over a long policy explanation to ask the one thing they care about. They can ramble, backtrack, and think out loud the way people do on the phone, and still get a clean answer. That fluidity is what makes the line feel staffed rather than automated, and it's what keeps the caller on the line long enough to get what they came for.

For a support team, this matters commercially. A caller who gets a fast, natural answer to "can I return this" often converts it into an exchange or a new purchase in the same call. A caller stuck in a menu just leaves. The realtime, interruptible design isn't a cosmetic nicety — it's the mechanism that turns answered calls into retained customers. It also means the handful of calls that do need a person arrive with the caller still willing to talk, rather than already frustrated by a machine.

None of this changes the honest boundaries: the agent still captures account-specific look-ups rather than guessing, still routes complaints to a human, and still answers only from the knowledge you approve. But within those boundaries, the live, two-way feel is what makes the whole thing work — and it's the first thing to test on a real call before you go live. Ask it something, cut it off mid-sentence, change your mind, and see whether it keeps up the way a good assistant would.

FAQ

Can it look up my specific order? It captures order details and contact information and routes account-specific look-ups to your team rather than guessing at a status it can't verify. For the high-volume factual questions — returns process, shipping timelines, hours, policy — it answers directly and instantly.

How does it handle returns questions? It answers from your configured returns and refunds policy — the window, how to start, timelines, and exceptions — in plain spoken language, and can capture a return or exchange request and route it into your process. Keep the policy current and every caller hears the right answer.

Does it feel like a machine? No — it's a two-way, realtime conversation. The caller can interrupt, switch a return to an exchange mid-sentence, or talk over a long answer, and it adjusts. It isn't a press-1 menu or a scripted turn-based bot. Test it on real calls before going live to confirm the pacing suits your customers.

What does it cost, and do I need a card to try it? The control plane where you configure persona, hours, knowledge, and numbers is free. You pay only for the minutes the agent spends on calls — see pricing — and you can start a trial with no card required to set up and test the line first.

What happens with complaints? They hand off to a person quickly. Configure the agent to route complaints and unhappy callers straight to your team, since those want a human. See the human handoff guide.

Can we use our existing support number? Yes. Connect the number customers already call, or a dedicated line, so the experience is seamless. See product for how numbers, persona, and knowledge fit together.

How do we keep answers accurate as policies change? The knowledge base is yours to edit in the free control plane, so when a returns window, shipping timeline, or seasonal hours change, you update it and every caller immediately hears the current answer. Keep fast-changing internals — live single-item stock, individual order details — out of the static knowledge and let the agent capture and route those to a person, so callers never hear a stale answer stated as fact.