If you are a TikTok Shop Indonesia seller whose best seller ran out of stock at 11am on the first peak day, with 40 orders already paid for and a warehouse that will not restock for five days, the peak day does not reward a better plan. It rewards the right order of operations, executed at the right hour. The instinct is to zero the stock and refund everyone, and that is the most expensive of the four options in front of you. Zeroing the stock takes the listing down, which the ranking system records as a change of availability and treats as a fresh start. Cancelling the orders records a fulfilment failure that stays on your store for the rest of the campaign. Delaying costs a voucher and a message. Restocking from another store costs a transfer. Only one of those four is cheap, and it is not the one the dashboard makes easiest.
Work the numbers on a Rp200,000 SKU that has 40 oversold orders on day one. Goods cost 45%, or Rp90,000 a unit, so the forty units are Rp3,600,000 of stock you do not have. Cancel them and the direct loss looks small: you refund Rp8,000,000 of collected revenue that was never yours, and you keep nothing. But the cancellation rate is what gets recorded. Forty cancellations out of roughly 800 orders that day is 5%, against a 3% fulfilment ceiling, and crossing it costs you visibility rather than money on the invoice. Your store does Rp25,000,000 a day at a 28% margin, which is Rp7,000,000 of gross profit a day. A 25% exposure cut over the remaining seven days of the window is Rp1,750,000 of gross profit a day, or Rp12,250,000 across the window. Delaying instead costs a Rp10,000 apology voucher per order, Rp400,000 in total, and it does not count as a cancellation at all. Same forty orders: Rp400,000 to delay, Rp12,250,000 of gross profit to cancel. The ratio is thirty to one, and the only reason sellers cancel is that the button is easier to find.
On the peak day, the sequence matters more than the decision
On the peak day you are not choosing a strategy, you are choosing a sequence, because the four things you can do interfere with each other. Cut the traffic first and the stock lasts long enough to serve the orders you already have. Zero the stock first and the listing goes down while the ads are still running, so you pay for clicks that land on a dead page. Cancel first and the metric is already recorded, and no later fix takes it back. Reprice first and the ranking system starts relearning the listing on the exact day you need it to be stable. Each correction costs less than the one before it only if it happens in the right order.
Write the day as six hours, not as a workday. Hour zero is when you notice the run rate: units sold per hour multiplied by hours left in the traffic peak tells you whether the stock survives until the evening. Hour one is traffic control. Hour two is the inventory decision. Hour three is order triage. Hour four is the creator side. Hour six is the reset, where you decide what tomorrow's peak looks like. Everything after hour six is tomorrow's problem, and tomorrow's peak is set by what you did in the first six hours of today's.
First, stop the traffic — not the listing
Cut the traffic in three layers, and do it before you touch a single inventory field. Paid first: pause the ad groups carrying that SKU, because they are the ones buying clicks you cannot serve. Creator second: open the campaign list and pause the running collaborations on that SKU, since a creator video published this morning will keep sending people to a product you cannot ship. Organic last, and you cannot pause it, so edit the listing instead: take the promotion badge off, change the first image to a variant that is in stock, and write the restock date into the description. Cutting traffic costs nothing and buys you the four hours you need to decide the rest.
One warning about pausing the campaign: pause it, do not delete it. A deleted campaign loses its accumulated history and starts from zero when you restart, while a paused one resumes with the same learning. The same rule applies to the ad group and to the creator collaboration. On a peak day you are pausing things for six hours, not retiring them for the season, so every stop should be reversible in one click.
The six-hour sheet is the only thing you need to have open on the peak day: one row per SKU, five columns — units sold in the last hour, units left, hours of stock remaining at that rate, orders already paid but not shipped, and the owner of the next action. Get the peak-day six-hour dispatch sheet from a DAMI specialist and copy the columns, because the sheet is not the valuable part; the discipline of filling it at the same three moments every day is.
Second, hold half the stock instead of zeroing it
Do not set the stock to zero. Zero means unavailable, and unavailable means the listing leaves the category page, the ranking system records an availability change, and every click already in flight lands on a sold-out page. Set it to half of what is actually arriving, so the listing stays live and you keep taking orders you can serve. The half you hold back is not for selling faster, it is the buffer that lets you delay the remaining forty instead of cancelling them, because a delay you can honour in three days is a message and a delay you cannot honour is a cancellation waiting for its date.
If you run more than one store, check the other stores before you send a single delay message. The same SKU sitting in a second store is inventory you already own, and a transfer between your own stores costs a courier fee rather than a supplier lead time. On a peak day that difference is measured in hours: a courier transfer lands tomorrow, a supplier restock lands in five days, and the campaign has seven days left. Most sellers who cancel forty orders on the peak day are cancelling stock that is physically in their own warehouse in another city.
Third, sort the orders by value and delay before you cancel
Triage the orders by value, not by order number. Start with the highest-priced SKUs, because a delay message on a Rp450,000 order protects a margin large enough to cover the voucher twice over, while a delay on a Rp90,000 order protects very little and annoys the buyer just as much. Send the message as a choice rather than a notice: hold the order with a Rp10,000 voucher credited today, or refund immediately with no penalty to us. Roughly two-thirds take the delay when the date is specific, and the ones who do not take the refund were going to cancel anyway, so you lose nothing you had.
Every delay message needs three things or it will produce a second message: the exact ship date, not a range; the voucher amount and when it lands; and a reply channel a human answers. Vague dates are what turn a delay into a cancellation two days later, because the buyer has no way to tell a promise from a hope. Write the date you can keep even if the courier is late, then keep it.
Fourth, prices move last, after the window closes
Price moves last, and on most peak days it does not move at all. Two reasons. The first is mechanical: a price change is a signal the ranking system responds to by re-testing the listing, and re-testing takes one to two days during which your conversion data is not comparable to yesterday's — which means you cannot tell whether a bad day was the price or the traffic. The second is contractual: the discount you registered for the campaign sets a floor, and a price raised or cut mid-window can be measured against the 30-day low you declared when you signed up. If you must move something on the peak day, move the tail: the long tail SKUs carry almost no ranking equity, so repricing them costs you nothing you cannot rebuild tomorrow.
The exception worth knowing: if the SKU is already out of stock, its price is not doing any work, and repricing it now costs nothing because there is no conversion to disturb. That is the one case where a peak-day price change is free — the listing is paused anyway. Use it to reset a price that was drifting too low before the window, so that when stock lands you start from a number you chose rather than a number you inherited.
How DAMI covers the six hours
DAMI brings an 8-million-plus creator database, competitor and competitor-store creator search, automated outreach tasks, AI-assisted outreach, AI video creation, sample management, campaign management, social media management, store analytics, and multi-store team dashboards into one place. The reason this goes wrong at scale is that the three lists you need — what is selling, what is being promoted, and what has been paid but not shipped — live in three different places, and by the time you have read all three the stock situation has changed. Pulling them onto one screen is the part that can be systematised.
Pull the last six hours by SKU and by channel
Open store analytics and split the last six hours by SKU, then by traffic channel, so the same three numbers appear next to each other: units sold, units left, and the channel that is still buying. The output is a three-row list of the SKUs whose stock finishes before the peak does — not a dashboard, three rows you can act on.
Pause outbound work on exactly those three SKUs
In campaign management, pause the running collaborations on those SKUs and stop sampling for them, while leaving every other SKU untouched. The output is a stop list with three names on it, which is the difference between losing one day of one product and losing the campaign's creator coverage.
Check the other stores before sending a single delay message
Use the multi-store view to see whether the same SKU is available in another store you own, and raise a transfer rather than a purchase order. The output is either a transfer note with an arrival date, or a confirmed decision that there is nothing to transfer — and only in the second case do you write the delay message.
Questions sellers ask on the peak day
What is the difference between setting stock to zero and taking the listing down?
Less than sellers assume, and more than they want. Both make the product unsellable, but zero is reversible in a minute and a takedown has to go through review again. Zero also keeps the listing's reviews, its position in your own store, and the links your creator videos already point at. Take the listing down and those links become dead ends.
How long can a delay be before buyers refuse it?
Seven days works in most categories; beyond seven the refusal rate climbs sharply because the purchase stops feeling like a peak-day deal. If your restock is more than a week out, do not offer a delay — offer a partial shipment of what you have plus a voucher on the remainder, and keep the order alive that way.
A creator video published this morning is still pushing the sold-out SKU. What now?
Do not ask for the video to be taken down; a removed video wastes the spend and the creator's slot. Ask for a pinned comment with the restock date and a link to the in-stock variant, and pause the paid amplification behind it. The organic views keep arriving, but they land on a page that answers the question instead of a sold-out page.
Do platform vouchers keep applying while the SKU is out of stock?
The voucher only applies to orders that complete, so an out-of-stock SKU does not consume voucher budget — but the campaign budget you registered is committed for the window regardless. That is why the SKUs you keep live should be the ones with the deepest margin, not the ones with the most traffic.
My boss says forty cancellations out of eight hundred orders is nothing. Why spend money on delays?
Because the forty orders cost Rp3,600,000 in goods and the cancellations cost Rp12,250,000 in gross profit, and only one of those numbers appears on the day's report. Forty cancellations put the cancellation rate at 5% against a 3% ceiling, and the exposure cut that follows runs for the remaining seven days of the window. The vouchers are Rp400,000. Show the two numbers side by side and the argument ends.
What to do in the next fifteen minutes
The expensive mistake on a peak day is not a wrong call. It is a right call made six hours late, after the traffic is already paid for and the orders are already on the books. DAMI brings an 8-million-plus creator database, competitor and competitor-store creator search, automated outreach tasks, AI-assisted outreach, AI video creation, sample management, campaign management, social media management, store analytics, and multi-store team dashboards into one place.
Open yesterday's order export and sort it by SKU, not by time. For the top ten SKUs, divide units sold in the last six hours by six to get an hourly rate, then divide the units you actually have by that rate. The answer is hours of stock. Any SKU under six hours goes on a list, and that list is tomorrow's first task — before the ads, before the creators, before the meeting.
Start with DAMI while the numbers are still small enough to read.
