Ecommerce customer support during peak season: what to prepare before the peak

Your store does not fail in August. It fails in November. In a normal season customer support is managed with whatever is around: somebody answers between two other things and it works out. In peak season that stops working, and it stops working every year in the same place.

This article is about what can be up and running before the peak. There is nothing here that cannot be prepared in September, and almost all of it is done once and lasts forever.

What happens from November to January

Tickets multiply, and not only because of order volume: the type of question changes. In a normal season most are product queries. In peak season they are «will it arrive before the 24th?», and that question does not accept an answer 48 hours later.

The team starts answering at night. First for a while after dinner, then at the weekend. And since whoever answers is also the one buying stock, negotiating with suppliers or setting up campaigns, what stops is everything else.

In January the second half arrives, which almost nobody budgets for: returns. Back comes the gift that was the wrong size, the one that arrived late and the one that was bought twice. And campaign returns are not closed in January: they are still being closed in March, because each one drags a collection, an inspection and a refund behind it, and because in the middle there is a carrier to chase.

That drag is what turns a six-week peak into a bad quarter. And it is the part that can be prepared best, because it is the most mechanical of the three.

Why hiring in October for December is too late

The maths looks reasonable: the peak starts in late November, so hiring in October leaves plenty of room. It does not, because what takes time is not finding someone: what takes time is that person knowing.

Anyone coming in to handle your store needs to know your catalogue with its variants, your returns policy, your real delivery times per carrier and your tone. That is weeks of real work, not of theoretical training, and it is learned by answering genuine cases with someone reviewing the replies before they go out.

If that curve falls inside the peak, two things happen at once: the person learns at half speed because the queue does not wait, and you review replies in exactly the week when you have least time. The result is someone who knows the job in January, when what you needed was someone who knew it in November.

The version that works is the other way round: by November they already know your operation instead of learning it. In practice that means starting before the summer or just after it, and using the quiet season for what it is good for —learning without pressure and writing down what today lives in somebody's head. We have covered it in more detail on the ecommerce page.

Canned answers: one day's work that shows in December

If you are only going to do one thing off this list, do this one. Take last campaign's tickets and group them by question. There will be fewer groups than you expect, and among them will be almost every email in December: order status, delivery time, address change, size, availability, return, carrier incident, invoice.

For each group write the answer you would give, in your tone, including the bad version: what you say when the order will not arrive by the 24th. That is the one nobody writes and the one most needed, because it is the one improvised at eleven at night and it comes out wrong.

Two warnings worth their cost. One: a canned answer is not a closed script, it is a starting point so nobody rewrites the same thing forty times. Two: if it lives in somebody's head and not in the helpdesk, it does not exist for whoever comes in in November.

The returns policy, in writing and with its edges

Almost every store has a published returns policy. Far fewer have written down what to do when a case falls outside it, and that is what happens in peak season.

The edges worth deciding in September, not in January: what happens to the window when the order was a gift bought in November and opened at Christmas; what happens if the product comes back without its original box; up to what amount a refund can be issued without asking; what is offered when the delay was the carrier's fault and not yours. Every one of those decisions is yours and takes five minutes in September. In January it takes an hour, with the customer waiting.

With that written down, the work of closing returns and chasing carriers can genuinely be delegated, because whoever does it applies your policy and not their own. That is exactly what the returns and incidents role covers, and it is the one hired latest and needed earliest.

Who escalates what

The third thing to prepare beforehand is the shortest to write and the one that saves the most nights: the list of what does not get decided without you.

Write it as a list of cases, not as a principle. For example: a refund above a certain amount, an exception to the policy, a complaint that mentions a third party, a customer announcing they are going to write a review, a business order with an invoice. Anything not on that list gets answered without asking.

What matters is not where you draw the line, it is that there is a line. Without one, one of two things happens: either everything reaches you, and then you have delegated nothing, or nothing reaches you, and you find out in January.

And one more rule: whatever gets escalated is escalated with the case already put together. Who the customer is, what they ordered, what they have been told and what is being proposed. An escalation that is only «look at this ticket» hands the whole job back to you.

The real delivery times, not the published ones

One last short section that saves more arguments than any other. Your site states a delivery time; in peak season the real one is different, and it changes by carrier and by area. If whoever answers does not have that figure, they promise the one on the site and the customer comes back two days later, rightly.

Write down in a table the times you are actually seeing this campaign, per carrier, and the order cut-off date for delivery before Christmas. It takes five minutes a week to update and it is what turns «I think it will arrive» into an answer you can defend.

What to measure to know whether the campaign is going well

Four numbers are enough, and your helpdesk gives you the first three without configuring anything: tickets in per day, first response time, resolution time and escalation rate. The first tells you whether the peak is the one you expected. The second is the one the customer feels. The third is the one that drags into January. The fourth tells you whether the line is drawn in the right place.

Look at them weekly, not per campaign. An average across the whole of December is of no use: by the time you see it, it is over.

Where to start if the campaign is already close

If you are reading this in September, the order is: canned answers, the edges of the returns policy, the escalation list. Three afternoons, and they hold for every campaign after this one.

If what you have is an uncovered role and a campaign coming, what to measure first is one normal week of volume: how many tickets come in, through which channels and how many go unanswered. With that number you can see whether what you need is a dedicated person on customer support all year or a one-off reinforcement. What does not work is deciding it in November with a full queue.