#The Peak Season Support Playbook 2026

What last season's public data tells Shopify merchants about BFCM and holiday customer service, and what to set up before November 27.

Peak season does not break your customer service. It shows you which parts were already weak.

A question that gets a vague answer once a day in October gets that vague answer forty times a day in BFCM week. A return policy that is "mostly clear" turns into a steady flow of tickets in January. A support inbox that one person can clear before lunch becomes a backlog that nobody clears at all.

This playbook is about finding those weak spots now, while you still have four weeks to fix them.

#Contents

• Key findings
• About this playbook
• Part 1: Demand goes up, support capacity does not
• Part 2: What shoppers ask in each phase
• Part 3: What your store content misses
• Part 4: The work, by deadline
• Part 5: Where HeiChat fits
• Appendix: Sources

---

#Key findings

#Finding 1: Demand grew faster than the people available to handle it

Shopify merchants sold $14.6 billion over BFCM weekend 2025, up 27% from 2024. More than 94,900 merchants had their best sales day ever on Shopify.^[1]

Over the same season, NRF expected retailers to hire 265,000 to 365,000 seasonal workers. That was the lowest range in at least 15 years, against 442,000 hired in 2024.^[4]

More orders, fewer extra hands. For a small Shopify team, "hire someone for December" is often not a real option to begin with.

#Finding 2: Questions change with the calendar

The questions shoppers ask move through three phases:

• BFCM week: discounts, codes, stacking rules, free shipping thresholds
• December: delivery dates, order-by deadlines, tracking, gift questions
• After the holiday: returns, exchanges, sizes, refund timing

Retailers expected about 17% of holiday sales to be returned.^[6] Salesforce reported that service conversations handled by AI agents grew 66% from November to December.^[3] The work does not end on Cyber Monday. Its subject changes.

#Finding 3: Shoppers are already using AI to shop and to get help

Adobe measured a 693% year-over-year increase in traffic to US retail sites from generative AI tools during the 2025 season.^[2]

Salesforce reported that AI and agents influenced 20% of global Cyber Week orders. Shoppers used retailers' AI agents for customer service 126% more during the holiday rush than in the two months before.^[3]

The point is not that every store needs AI. The point is that a growing share of shoppers now expect an instant answer, and they will ask the question whether or not you have one ready.

---

#About this playbook

#Source types

The "typical questions" lists in Part 2 are illustrative examples, not counted data. We have marked them that way.

---

#Part 1: Demand goes up, support capacity does not

+27%: Shopify merchants' BFCM weekend sales growth in 2025, year over year.^[1]

#The volume picture

Every large source points the same way: last season set records, and it was concentrated in a short window.

For 2026, NRF forecasts November and December retail sales to grow 4.2% to about $1.013 trillion.^[5] NRF's 2025 forecast came in higher than the actual result (3.1%), so treat this as a direction, not a promise.

#Why sales growth turns into support growth

More orders mean more questions, but not in a straight line. Three things make peak-season questions grow faster than orders:

1. New customers ask more. A returning customer knows your sizing and shipping times. A first-time buyer who found you through a Black Friday ad does not.
2. Promotions create rules, and rules create questions. Every "buy 2, get 15% off, excludes sale items, ends Monday at midnight" offer generates a set of questions about edge cases.
3. Delivery anxiety rises in December. A package that is "in transit" for four days in October is fine. The same package on December 21 becomes a message, then a second message, then a chargeback threat.

Mobile matters here too. With most purchases happening on a phone,^[2] shoppers ask in short, fast messages and expect short, fast answers. They will not search a help center page on a 6-inch screen.

#Why capacity does not keep up

These numbers describe the retail industry as a whole, not Shopify support teams. Still, the pattern is familiar to most small merchants: the founder answers chat, the operations person handles tickets, and everyone is also packing orders. There is no seasonal hire to absorb the extra questions.

#What shoppers already do with AI

Salesforce's numbers come mostly from large retailers on its own platform, so they are not representative of a typical Shopify store. We include them because they show when the service load peaks, not because a small store should expect the same figures.

---

#Part 2: What shoppers ask in each phase

About 17% of holiday sales were expected to be returned in 2025.^[6]

Peak season is not one event. It is three phases, and each one has its own questions. If you prepare only for BFCM week, you will be caught out in December and January.

The question lists below are illustrative. They are the kinds of questions Shopify stores commonly receive in each phase, written in the way shoppers actually type them. They are not counted from a dataset.

#Phase 1: BFCM week (November 23 to December 1)

Main subjects: discounts, codes, stacking, free shipping thresholds, stock.

Black Friday 2025 grew faster than Cyber Monday (9.1% vs. 7.1% YoY), as shoppers moved earlier to catch deals.^[2] NRF's survey found that 43% of holiday shoppers planned to start before November.^[5] Questions about your sale often start the week before the sale itself.

Typical questions:

• "Does the 20% off work on items that are already on sale?"
• "Can I use my welcome code with the Black Friday code?"
• "My code says it's invalid. It's still Friday, right?"
• "How much more do I need to spend to get free shipping?"
• "Is this restocking before Cyber Monday?"
• "Will the price drop more on Monday or should I buy now?"

What these have in common: every one of them has a rule behind it, and you set the rule. The AI or the agent can only answer correctly if the rule is written down somewhere they can find it.

#Phase 2: December (December 2 to 24)

Main subjects: delivery dates, order-by deadlines, tracking, delays, gifts.

USPS published these 2026 recommended order-by dates for delivery by December 25 within the contiguous US:^[7]

These dates count from when USPS accepts the package, not when you print the label. Your customer-facing cut-off date needs to subtract your own processing time.

Typical questions:

• "If I order today, will it arrive before Christmas?"
• "What's the last day to order for Christmas delivery to Canada?"
• "My tracking hasn't updated in 3 days. Is it lost?"
• "Can I send this directly to my sister as a gift without the price on it?"
• "Do you do gift wrapping?"
• "Can I change the shipping address? I just realized it's going to my old place."

The first two questions have a correct answer you can write in advance. The third and sixth need an action: a real order lookup, or a person who can edit the order. Mixing these up is one of the most common peak-season mistakes. We come back to it in Part 3.

#Phase 3: After the holiday (December 26 to January 15)

Main subjects: returns, exchanges, sizes, refunds, gift receipts.

NRF and Happy Returns reported that retailers expected 17% of holiday sales to be returned, with an estimated 19.3% return rate on online sales overall in 2025. NRF also reports that 82% of shoppers return unwanted holiday gifts within a month of receiving them.^[6]

Typical questions:

• "I got this as a gift. Can I exchange it without the order number?"
• "Does the holiday return window apply to things I bought in November?"
• "This runs small. Can I swap for a size up and keep the same price?"
• "I returned it last week. When will I get my refund?"
• "The item I want to exchange for is sold out. What now?"

These questions are slower and more emotional than BFCM questions. A shopper asking about a refund has already paid, and the tone of your answer affects whether they come back.

#How the phases overlap

The phases are not clean. Shoppers still ask about returns in BFCM week ("If I buy now, can I still return it after Christmas?"). Delivery questions start in November for international buyers. Discount questions come back on Boxing Day if you run a sale.

The practical lesson: write the content for all three phases before BFCM week starts, then switch on or adjust each piece as its phase begins. Once BFCM starts, you will not have time to write the January return policy.

---

#Part 3: What your store content misses

Not every unanswered question is a content gap.

Before peak season, many stores review the questions their chatbot or team could not answer, and write a new FAQ for each one. That works for some questions and wastes time on others.

#Three kinds of unanswered questions

When a question goes unanswered, it usually falls into one of three groups:

If you treat all three groups as content gaps, two things go wrong:

1. You write FAQs for questions that need an action. "Where is my order" cannot be answered by a knowledge base article, no matter how well it is written.
2. You spend time on noise, and the real content gaps get less attention.

#Same answer, many phrasings

Shoppers do not use your wording. The same promotion rule gets asked in many ways:

• "Can I use 2 codes?"
• "Does BLACKFRIDAY work with my newsletter discount?"
• "Why did my 10% disappear when I added the other code?"

All three need the same answer: codes do not stack; the higher discount applies.

This has two consequences:

• Keyword-based FAQ bots miss most of these. An FAQ titled "Can I combine discount codes?" will not match "Why did my 10% disappear?"
• When you add a Q&A entry, add several phrasings of the question, not just one. Write the answer once and attach the different ways people ask it.

#A practical sorting routine

Use this before BFCM and then weekly during the season:

1. Export last month's unanswered or poorly answered questions. Include conversations that ended with a shopper asking for a human.
2. Remove the "needs nothing" group: greetings, spam, off-topic messages.
3. Route the "needs an action" group to a process, not to an FAQ. That means order lookup, order editing, or a clear handoff to your team.
4. Group the remaining "needs content" questions by subject. Count how many times each subject appears.
5. Write one answer per subject, starting from the most frequent. Add several phrasings to each.
6. Re-test by asking the original questions again, in the original words.

#The pre-season content checklist

Based on the phases in Part 2, these are the content items most stores need before November 27:

• BFCM promotion rules (what, when, stacking, exclusions, time zone)
• Free shipping thresholds, by currency
• Holiday shipping cut-off dates, by region
• Post-cut-off options (express shipping, gift cards, "arrives after the holiday" notice)
• Gift options (gift wrap, gift notes, sending to another address, gift receipts)
• Holiday return window and which purchase dates it covers
• Gift-recipient exchanges without an order number
• Refund timing
• Size and fit guidance for your top 10 holiday products
• Restock expectations for items likely to sell out
• Support hours over the holidays, and what happens to messages sent outside them

If you can only do five, do the first five. They cover BFCM week and most of December.

---

#Part 4: The work, by deadline

Four weeks is enough time to prepare, if you start with the right things.

#Three setup risks to check first

These are not content problems. They are setup problems, and each one can cause trouble in BFCM week no matter how good your answers are.

Risk 1: Running out of capacity in the busiest week. Most AI support tools have usage limits, whether they count conversations, replies, or tokens. A plan that is fine in October can run out in BFCM week. Find out now what happens when your limit is reached, and whether you can add capacity quickly.

Risk 2: Widgets covering the things shoppers came to click. A chat button, a discount popup, and a free-shipping bar can all compete for the same corner of a phone screen. On mobile, a widget sitting over the "Add to cart" or checkout button costs you orders directly. Check your store on a real phone, not just in a desktop preview.

Risk 3: Old promotion content that keeps answering. If last year's Black Friday rules, or this year's early-access rules, are still in your knowledge base, your AI may quote them confidently. Every temporary piece of content needs an end date.

#The six-line readiness scorecard

Score each line: Ready = 2, Partial = 1, Missing = 0.

How to read your score:

• 10 to 12: You are in good shape. Focus on testing and the weekly review.
• 6 to 9: Fix your missing lines first, then raise your partial lines.
• Below 6: Start with lines 1, 3, and 6 this week.

#The deadline table

#The five-step weekly review

Set aside 30 to 45 minutes once a week, on the same day. One person owns it.

1. Export. Pull last week's unanswered questions, low-confidence answers, and conversations that ended in a request for a human.
2. Set aside. Remove the "needs nothing" group: greetings, spam, off-topic messages.
3. Route. Send "needs an action" questions to the right process: order lookup, order editing, or a ticket for your team.
4. Fix at the source. For "needs content" questions, add or update the knowledge entry. Do not just reply to the shopper and move on.
5. Re-test. Ask the original question again, in the shopper's original words, and confirm the answer is now correct.

Worked example. In the first week of December, the review shows 14 questions about delivery to Canada. Sorting them:

• 9 ask "will it arrive by Christmas" → needs content. You add your Canada cut-off date to the holiday shipping entry, with three phrasings.
• 4 ask about a specific order stuck in customs → needs an action. You route these to your team as tickets.
• 1 is a shopper asking about a competitor's product → needs nothing.

One knowledge entry, written once, handles the nine repeat questions for the rest of December.

#The four-stage timeline

---

#Part 5: Where HeiChat fits

HeiChat is an AI customer support and sales assistant built for Shopify. It learns your products, policies, and knowledge base, and answers shoppers in chat. This table maps the playbook to what HeiChat does today.

#A suggested HeiChat setup order for BFCM

If you are starting from scratch, this order matches the deadline table in Part 4:

1. Install and sync. Check Learned Products and make sure your best sellers are included.
2. Add seasonal knowledge. Promotion rules, shipping cut-offs, gift options, and holiday returns, each with an expiration date.
3. Turn on Knowledge Gap Detection. Let it collect gaps from real conversations for at least two weeks before BFCM.
4. Set up order tracking and handoff. Test both with a real order and a real team member.
5. Check widgets on mobile. Use visibility rules to keep them away from checkout and cart buttons.
6. Test with 20 real questions in AI Model before relying on live traffic.
7. Check capacity. Compare your recent usage with your plan, and decide whether you need Extra Tokens.

---

#Appendix: Sources

1. Shopify, "Shopify merchants generate record-breaking $14.6 billion in Black Friday Cyber Monday sales," December 2025. shopify.com/news/bfcm-data-2025
2. Adobe, "Holiday Shopping Season Drove a Record $257.8 Billion Online with Consumers Embracing Generative AI Tools," January 2026. news.adobe.com
3. Salesforce, "AI and Agents Propel Cyber Week to Record $336.6B in Global Spend," December 2025, and "2025 Holiday Shopping Data." salesforce.com, salesforce.com
4. CNBC, "Retail holiday hiring to hit lowest level in at least 15 years, NRF says," November 2025. cnbc.com
5. NRF, Winter Holiday Data and Trends and 2026 holiday forecast, October 2026. nrf.com
6. NRF and Happy Returns, "Consumers Expected to Return Nearly $850 Billion in Merchandise in 2025," and NRF Winter Holiday FAQs. nrf.com, nrf.com
7. USPS, "2026 Holiday Shipping Deadlines." usps.com