How to Reduce Support Tickets About Shipping and Returns

"Do you ship to Austria?" "How long do I have to return this?" "Can I pay cash on delivery?" If you run an online shop, you have answered these questions many times, usually by email and usually one at a time. Each reply takes a minute or two. Together they eat the hours you need for the cases that really need a person: a damaged parcel, a refund that went wrong, a large B2B enquiry.

This guide shows how to move the repetitive questions to AskMerra, an AI shopping assistant that answers from your own policies, and how to keep the real cases flowing to your team with all the context they need.

Why the same questions keep coming back

Shoppers ask about shipping and returns at the moment they are deciding. The answer sits behind a footer link, so they send an email, or they leave and plan to "check later".

Most of these questions fall into three groups:

• Shipping: which countries you deliver to, what it costs, the free-shipping threshold, delivery times, carriers and order cut-off times.
• Returns: the return window, what condition items must be in, who pays return shipping, when the refund arrives, whether exchanges are possible.
• Payments: accepted methods, cash on delivery, installments, gift cards.

Each has one correct answer that you have already written down somewhere. The problem is getting it to the shopper at the moment they ask.

Step 1: Put your policies in the knowledge base

In AskMerra, your policies live in Dashboard → Knowledge. You create one document per topic and per language, written in Markdown. The document types follow the questions above: Brand, Shipping, Returns, Payments, FAQ, and Custom for anything else, such as size guides, warranty terms, a loyalty program or store locations.

These documents are included in full in the assistant's context in every conversation. The assistant answers policy questions from them and does not invent policies. If something is not covered, it says so honestly and offers to pass the question to your team.

Write it the way you would answer a customer

Specific beats general. Compare these two shipping notes:

• "We ship quickly across Europe."
• "Orders placed before 14:00 on a weekday leave our warehouse the same day. Delivery to Germany and Austria takes 2 to 4 working days. Shipping is free above €50."

The second one lets the assistant answer "Will it arrive before Friday?" or "How much more do I need to spend for free shipping?" The first one leaves it guessing, and a careful assistant does not guess.

A size budget applies (the Knowledge page shows it), so keep documents factual. If a document is missing in one language, the assistant translates from your default-language version when it answers. It replies in English, Romanian, Italian, French, German or Spanish, whichever the shopper writes in.

Step 2: Turn the top questions into instant FAQ answers

Some questions deserve an exact, approved wording. For those, add FAQ entries: question and answer pairs, per language.

Every message sent to AskMerra goes through a router that tries the cheapest correct answer first:

1. Rules: greetings, thanks, goodbyes and "are you a bot?" are answered instantly and never billed.
2. Your FAQ: if the question closely matches an FAQ entry, your answer is returned word for word, with no AI call.
3. Precomputed and cached answers: common generic questions are answered from earlier good answers, always shown with live prices and stock.
4. The AI model: everything else, with your products and policies in its context.

FAQ matching is semantic, so "Do you deliver to Italy?" and "Is shipping to Italy possible?" can hit the same entry. Write the question the way customers phrase it, not the way your terms and conditions do. Instant answers are matched within the shopper's language, so add entries for every language you serve. In the other languages, the assistant still uses your FAQ as a reference when it answers.

What it costs

AskMerra pricing is based on the AI work your assistant actually does. Greetings and thank-yous are free, and FAQ and cached answers cost almost nothing, because they only need a small search. For comparison, a typical conversation that does need the AI uses about €0.04.

There is a second benefit. If you use up your included usage with pay-as-you-go switched off, or if spend protection pauses AI answers during a traffic spike, FAQ and cached answers keep working. Your most common questions stay covered.

Step 3: Hand real cases to a human, with the full context

Not every question should go to a bot. "Where is my order?", "My parcel arrived damaged", "I want a refund": these need someone who can open your order system and make a decision. AskMerra answers from your catalog and your knowledge base. It is not connected to your orders, and it is not meant to be.

When a shopper asks for a human, or the assistant cannot help, the widget offers a contact form. Here is what happens next:

• The shopper's message and the full chat transcript are emailed to the escalation address you set in Settings.
• The email's reply-to is the shopper's address, so your team answers from its normal inbox by clicking Reply.
• The conversation is marked as escalated in the dashboard, and the email links straight to it.

The transcript is what saves time. Your colleague sees what the shopper asked, what the assistant answered and which products came up. Nobody has to write back "Could you describe the problem again?"

Step 4: Review conversations and improve answers over time

A knowledge base is never finished. The conversations inbox shows every chat, together with the products that were recommended and clicked. You can filter by date, language, rating (thumbs up or down), escalation status, answer type (AI, FAQ, cache, precomputed or rule) and whether a conversation has been reviewed. You can mark conversations as reviewed and export them to CSV.

The analytics page gives the wider view: conversations, satisfaction, escalations, the share of questions answered without AI, and the opening questions shoppers ask most often.

A short weekly review

1. Open analytics and read the most common opening questions. Any policy question on that list that is not yet an FAQ entry is a candidate.
2. Filter the inbox by thumbs down and read those conversations first.
3. If an answer was vague because the policy was vague, fix the knowledge document. If the question has one right answer, add an FAQ entry in each language.
4. Read the escalated conversations. Some will be real order problems. Others will show a gap, such as a missing line about returns on sale items.
5. Mark what you have read as reviewed, so next week starts clean.

When a shopper rates an answer with a thumbs down, that answer stops being reused from the answer cache. Before you publish bigger changes, try them in the Playground. It uses the same AI as real conversations, so it counts as usage, but it never appears as a conversation in your analytics.

What changes for your team

The goal is not to take people out of customer service. It is to stop using them as a lookup service for your shipping table. When the assistant answers repeat questions the moment they are asked, in the shopper's language and at any hour, your team can spend its time where a person makes a real difference.

To see how AskMerra keeps its answers tied to your real data, read accurate AI answers grounded in your catalog. If you sell across borders, see multilingual customer support for e-commerce.

Try it with your own policies

Create your account, then add your shipping, returns and payments documents, write your first FAQ entries, and ask the Playground the questions your inbox gets every week. The knowledge base docs walk you through each step.