Black Friday Chatbot Strategy: A Week-by-Week Plan for Peak Season

Peak season compresses a quarter of questions into a few weeks, and most of them arrive at the moment of decision: a shopper with a cart open wants to know whether it will arrive before the 24th, whether two codes stack, and whether it can be returned in January. If nobody answers, they leave. Baymard Institute's running average documented cart-abandonment rate is about 70%, and a sale adds both urgency and doubt to that decision.

Human teams cannot scale to that spike without hiring, and seasonal hires do not know the products. An AI chatbot that answers from your own catalogue and policies covers the repetitive questions at any hour, at any volume, and hands the rest to a person. That is the whole strategy; the rest is the detail of doing it well. The same plan applies to any holiday season chatbot setup, whether your peak is Black Friday and Cyber Monday (BFCM), the pre-Christmas weeks, or a regional sales period.

Two things a chatbot will not do, so plan accordingly: it will not send abandoned-cart emails (Vatdi does not offer cart-recovery automation), and it will not fix a slow site or a stock-out. It removes the "I could not get an answer" reason for leaving, which is the one you control.

Read last year's November and December support inbox and you will find the same dozen questions in different words. Each needs a specific piece of knowledge, and most stores have not written it down.

Date-stamp every seasonal fact inside the text ("free returns until 31 January 2027") so the bot can see when it has expired even if you forget to remove it. For file types and the crawl, see how to train a chatbot on your data.

Coupons are where a chatbot does the most damage if it is careless and the most good if it is precise. Three rules:

1. The bot only offers codes you have entered with their conditions. In Vatdi, coupon offers appear in chat with their conditions when they fit the conversation; the bot answers from what you have entered, but it will repeat a stale code you left in a document. Delete expired codes from every document, not just the coupon list.
2. Write the stacking rule as one unambiguous sentence. "One code per order; sale prices cannot be combined with codes; free-shipping thresholds are calculated after discounts." Ambiguity here produces the chats you least want to read.
3. Decide what the bot says when asked for a code you do not want to give. A truthful "There is no additional code beyond the sale prices shown" is fine. Silence or a guess is not.

Test with the questions shoppers will actually type: "do you have a code", "why did my code not work", "can I use welcome10 with the Black Friday price". Score the answers against the terms document before the sale starts, then again after any code changes mid-sale.

The AI absorbs the repetitive volume; the peak still produces more handovers than a normal month, because more orders mean more address changes, cancellations and delivery problems. Plan the human side deliberately:

• Set agent hours for the peak days, not the normal rota. Extend evening coverage from the Thursday before Black Friday through Cyber Monday, and for the shipping cut-off days. Outside those hours the widget states that a person is not available and takes a message, which is better than an unanswered "connecting you" screen.
• Decide the handover topics in advance. Address changes, cancellations, damaged or missing deliveries, refund exceptions and orders above a value you choose go straight to a person. Everything else the bot handles first.
• Add every colleague who can answer. Vatdi allows unlimited teammates on every plan, so warehouse staff who know stock and shipping can take handovers on the busiest afternoon without a seat charge.
• Use the inbox states. Needs you, Handling, Team handling, Missed and Resolved show at a glance whether someone has picked a chat up; missed handoffs are counted, so you can see at the end of each day where the rota failed.
• Rehearse a handover the week before. One person triggers it from the widget, another accepts it, replies and hands back to the AI. Ten minutes now saves a confused customer later.

For the mechanics see what chatbot human handover is.

Small settings change results during a peak:

• Hide the widget on checkout if it distracts. Some stores find a chat bubble over the payment button costs more than it saves; others want it there for last-second shipping questions. Vatdi lets you hide the widget on chosen pages separately for desktop and mobile, so try checkout-hidden on mobile and visible on desktop and compare.
• Change the welcome and follow-up bubble to the season. "Ask me about Christmas delivery dates" earns more useful conversations than a generic greeting. Change it back in January.
• Set quick replies to the peak questions. Delivery cut-offs, returns window, discount codes, order status. Four buttons remove four rounds of typing.
• Check the widget on a slow mobile connection. The script loads asynchronously and should not hold up the page; confirm it on your theme rather than trusting the claim.
• Turn on visitor rating and thumbs if they are off. You will want the data in the review.

Appearance and placement options are listed on the features page.

The monthly conversation cap is the only volume limit on Vatdi, and it is the one that bites in November. AI replies inside a conversation are never capped, but new conversations are:

Say it plainly: a peak month needs Grow. If you normally sit on Starter, move to Grow before the sale and downgrade in January from the billing page; there is no contract and you can cancel any time. Check the usage forecast report in the dashboard in the second week of October to see how close the current run rate already is to the cap. Knowledge items and synced products also step up by plan (100/200 on Free, 500/2,000 on Starter, unlimited items and 10,000 products on Grow), which matters if you add a gift guide and seasonal FAQs. Details are on pricing.

If you are comparing platforms, note how AI usage is billed elsewhere: Tidio's Lyro is metered by conversation, Intercom's Fin is $0.99 per resolution and Gorgias tiers by ticket count as of September 2026, so a strong sale month raises those bills in proportion. Prices change; check each vendor's pricing page.

The season does not end on Cyber Monday. The first three weeks of January bring the returns and exchanges from everything sold in November and December, and the questions are different:

• "How do I return a gift I do not have the receipt for?" Load the gift-return process explicitly, including what happens without an order number.
• "Can I exchange for a different size?" State whether exchanges are handled as return-plus-reorder or as a swap, and who pays shipping.
• "When will my refund arrive?" State the processing time from receipt of the return and the payment method it goes back to.
• "Is the returns window still open?" Keep the extended window's end date in the policy until it passes, then remove it the same day.

Keep Grow through January if returns volume is high, and keep the handover topics for refund exceptions active; a refused return is a human conversation. Review the quality grades on returns conversations in the second week of January, because this is where policy text tends to be vaguest. Then remove every seasonal document and reset the welcome bubble, so February's bot is not offering December's delivery dates.

Do a ten-minute review every morning of the sale week and a longer one on the Tuesday after Cyber Monday. Reports are included on every plan.

• Shipping cut-off dates per country and service written down with dates, loaded, and tested with "will it arrive before Christmas"
• Order lookup enabled through the store plugin and tested with a real order number
• Catalogue synced with stock and sale prices showing on product cards
• Every live coupon entered with conditions, dates and the stacking rule; expired codes deleted from all documents
• Extended returns window with an exact end date in the policy; gift-return and exchange process written
• Gift wrapping, messages and invoice-price answers loaded
• Handover topics decided: address change, cancellation, damaged or missing delivery, refund exceptions, high-value orders
• Agent hours extended for Thursday to Cyber Monday and cut-off days; every available colleague added as a teammate
• Handover rehearsed end to end, alerts confirmed on the phones and browsers that will be watched
• Widget hidden on checkout where tested to help; welcome bubble and quick replies set to seasonal questions
• Plan moved to Grow; usage forecast checked
• 50-question test set run against the seasonal knowledge, everything under 7 fixed
• Daily ten-minute review scheduled for the sale week
• January: returns knowledge loaded, seasonal documents removed after the window closes, welcome bubble reset