How to reduce customer support e-mails for an online store

Step 1: count what people actually ask

Before changing anything, take the last 100 support e-mails (or the last month, whichever is fewer) and put each one in a category. Keep the categories plain:

The count tells you where to work. If one category is far ahead of the others, start there. In most stores it is order status or shipping, but count your own: a store selling spare parts may be dominated by compatibility questions instead.

Keep the tally. You will repeat it after each step to see what changed, and it is the only honest way to know whether a chatbot later on is actually helping.

Step 2: write the answers where people look

Many "support" e-mails are really a missing sentence on a page. Before automating anything, put the answers where shoppers expect to find them.

Shipping

• Which countries you ship to, and which you don't. "We ship worldwide" invites the question "including Brazil?".
• Cost and delivery time per region, written as numbers, not "fast" or "affordable".
• When orders leave: "orders placed before 2 p.m. ship the same business day" answers a whole class of "when will it ship" questions.
• Who pays import duties, if you ship across borders.
• A link to the shipping page from the cart and from the order confirmation e-mail.

Returns and exchanges

• The window, counted from what (order or delivery).
• The condition required: unused, in original packaging, with tags.
• Who pays return postage, and how the refund is paid and how long it takes.
• What cannot be returned. Say it plainly; a clear "no" saves more e-mail than a vague "contact us".

Product pages

Go back to your tally of product questions and add each answer to the product page it belongs to: sizes and measurements, materials, compatibility, what is in the box, care instructions. If the same question comes up across many products ("which size am I?"), write one guide and link to it from each product.

Contact page

Give one support e-mail address that someone reads, your usual reply time, and the information you need in a message (order number, photos for damage). Asking for the order number up front saves a round of e-mail on every ticket.

Step 3: let shoppers check their own orders

"Where is my order?" is usually the largest single category, and the answer is already in your system. The fixes, from simplest to most complete:

1. Send the tracking link in the shipping confirmation e-mail, and make sure that e-mail is not landing in spam.
2. Explain what the statuses mean. "Fulfilled" or "label created" confuses people; one line on your shipping page saying what each step means prevents a lot of worried e-mail.
3. Offer an order status page that works without creating an account. Many shoppers check out as guests and have no password.
4. Answer order status in chat. On Shopify, our assistant gives a shopper their order status when they type the order number and the e-mail address used for the order. Nothing is shown if the two don't match. See the order tracking chatbot page.

Step 4: an FAQ page, done properly

An FAQ helps only if it answers the questions from your tally, in the shoppers' words, and stays current. Write each answer so it stands on its own, link to the full policy, and delete questions that nobody asks. A long FAQ of questions you invented is just another page to search.

Step 5: put an assistant in front of the inbox

Once your pages hold the answers, a chat assistant can deliver them at the moment the shopper asks, instead of making them search. This is where it pays off, and also where it can backfire if done carelessly. What to require:

• It answers only from your store's own information. An assistant that guesses creates new e-mail ("your bot said I had 60 days to return this").
• It says when it doesn't know and tells the shopper how to reach you. See how to get an AI chatbot that doesn't make things up.
• It hands over the right conversations: refunds, damage, complaints, anyone asking for a person.
• It shows you what it couldn't answer, so you can keep improving the pages from step 2.

With our assistant, a website gets this from one line of code and its pages are read again every day. A Shopify store also gets order status, and, if the owner switches it on, cancelling an unshipped order or changing its address after the shopper types a one-time code e-mailed to the address on the order; the owner is told every time. See AI chatbot for online stores.

Step 6: decide your hand-off rules

Write down which conversations should always reach a person, and make sure they reach you somewhere you will see them quickly. A practical starting list:

• the shopper asks for a person;
• anything involving money: refunds, charges, payment problems;
• damaged, missing or wrong items;
• complaints and anything that sounds upset;
• order changes the assistant is not allowed to make.

Our assistant passes these to you by e-mail, Slack, Discord or Telegram with the conversation, and from the chat apps you can answer in the shopper's chat window while they are still there.

Step 7: a weekly fifteen minutes

Support e-mail creeps back when the store changes and the pages don't. Block fifteen minutes a week:

1. Read the list of questions the assistant could not answer, and add the missing answers to your pages.
2. Skim the hand-offs: was each one necessary, or could a page have answered it?
3. Re-check any page you changed that week (new shipping rates, a sale, a new product line).
4. Once a month, redo the tally from step 1 and compare.

Questions