What to watch during Buen Fin, hour by hour

Buen Fin 2026 runs five days, Friday November 13 through Tuesday November 17, with Monday the 16th a public holiday. Five days is a long time to spend refreshing screens, and refreshing screens is exactly what most of us do: open Seller Central, open the MercadoLibre panel, look at the sales counter, feel something, close it, repeat forty minutes later.

That cycle is not monitoring. It is anxiety with tabs. The problem is that most of the numbers you stare at during the event are not asking you for anything. They go up, they go down, they move for reasons you will not understand live and that only make sense once you see them complete at the close. Looking at them every hour improves no decision and does consume the attention you need for the few things that genuinely require it.

During a peak, the right question is not “how are we doing”. It is “is there anything that, if I do not touch it in the next thirty minutes, will cost me money”. That question has very few possible answers, and they are all the same kind: signals where delay has a cost that grows hour by hour. Everything else, however interesting, is close-of-day reading.

This article separates the two lists. First, the signals worth interrupting whatever you are doing. Then, explicitly, what is better left alone until the evening, because looking at it earlier only produces bad decisions taken on incomplete information.

the signals that genuinely demand live action

A signal earns live review when it meets three conditions. First: the cost of ignoring it grows over time rather than being a fixed amount. Second: there is a concrete action you can execute in minutes, not days. Third: the signal is unambiguous, meaning you do not need two hours of context to know whether it is a problem.

Under that filter, the list shrinks dramatically. Total sales do not qualify: they rise or fall, but there is nothing you can do live with an aggregate number. Conversion rate does not qualify: it needs volume to be reliable and two hours into the event it is still noise. Average order value does not either: it moves with product mix, and mix only makes sense at the end.

Three qualify, and a fourth if you run ads. Here they are.

signal one: sales velocity against remaining stock

This is the only signal that, read properly, actually saves sales. It is not “how much have I sold”, it is “at this pace, how long does what is left last”.

An example with numbers. You enter Friday with 180 units of your main SKU. On a normal day that product sells six. In the first four hours of the event you sold 22. That is 5.5 units per hour. If your useful traffic is fourteen hours a day, the projected pace is 77 units a day, and 180 divided by 77 is 2.3 days of coverage. The event lasts five. In other words: halfway through Sunday you are at zero, with two days of peak left and nothing to sell.

The important part is that you can run that math in hour four of Friday. You do not need to wait for stock to drop to find out: velocity already told you. And in hour four you still have real options. You can reassign units from another warehouse, move inventory between channels, raise the price to slow the pace and stretch coverage, or push an equivalent product instead. In hour thirty, already at zero, you have none.

This reading has to be done per SKU, not per parent product. A parent with 400 units can look healthy while the size carrying 60 percent of demand is already gone. It is the same problem described in sales velocity, only compressed: what you normally observe over weeks, during Buen Fin you observe over hours.

Glossary: sales velocity, the pace at which each SKU moves →

signal two: the price that drifted out of line between channels

The second signal is quieter, which is why it costs more. You scheduled a drop for Friday at midnight on Amazon and on MercadoLibre. One applied and the other did not. Or both applied, but on MercadoLibre an additional discount inherited from an earlier promotion stacked on top and the final price landed below what you authorized.

This does not show up as an error. It shows up as sales. For hours you will see one channel selling beautifully, and the intuitive read is “what a great channel”. The correct read is that you are selling below your floor and every unit that ships subtracts. It is the only Buen Fin problem that looks like success while it is happening.

The action is immediate and always the same: compare published price against approved price, channel by channel, and correct. If the gap is downward and below the floor, you correct it even if it slows sales. The alerts covered in cross-channel price drop alerts exist precisely so you do not depend on somebody remembering to check.

signal three: the broken sync and the overselling behind it

The third signal is technical and it does the most damage in the least time. If inventory updates between your warehouse and the channels stop, the panels keep publishing the last number they received. Over four peak hours, at 5.5 units per hour across several SKUs, that is dozens of units sold that no longer exist.

And overselling is not just a cancellation. It is a customer who bought on the most anticipated weekend of the year and will not get their product, with everything that implies for your MercadoLibre reputation and your Amazon account metrics, exactly when your visibility is highest. The full topic lives in how to prevent overselling between Amazon and MercadoLibre; live, all that matters is catching the silence.

Because that is how you catch it: by silence. A broken integration does not throw an error onto your screen, it simply stops updating. The practical signal is the timestamp of the last sync. If it says three hours ago and you are selling, there is a problem. Reconnecting and, meanwhile, manually lowering published stock on your fastest-moving SKUs is the right move, even if it feels primitive.

the fourth signal: ad spend running away

This one applies only if you are advertising. It is not ACoS, which over short windows is pure noise and belongs to the close. It is budget: if a campaign burns the day’s spend in the first three hours, your ads go dark for the rest of the day, and during Buen Fin that day is the one paying for the year.

The signal is percentage of budget consumed against time of day. If by eleven in the morning you have spent 70 percent of what you allocated to Friday, you have to decide whether to raise the budget or accept sitting out the afternoon. What you cannot do is find out on Saturday.

what can wait until the daily close

Here comes the uncomfortable part, because it is the hardest to honor. All of the following is best looked at once a day, at the close, and not before: total sales, units, average order value, conversion, ACoS, channel mix, year-over-year comparison and margin for the day.

It is not that they do not matter. They matter enormously. It is that none of them has an associated action executable in minutes, and all of them are misleading over short windows. A conversion rate that looks catastrophic in hour two usually normalizes by hour six. An ACoS that looks like hemorrhaging at noon usually settles once attributed conversions land. Watching them live produces the worst kind of decision: the one taken on real but incomplete data.

Margin for the day deserves its own note. It is the single most important number of the event, and it is worth reviewing daily. But not live, because during the day it is incomplete: commissions that settle later are missing, returns that have not happened yet are missing, and closed ad spend is missing. A margin computed mid-afternoon is optimistic almost by definition. What is useful is the closed daily margin, reviewed at night, with one concrete question: has any SKU already slipped below its floor, so it should come out of the event tomorrow.

Everything else, including the analysis of what worked and why, is November 18 work. During the event you only generate data; interpreting it comes after.

Glossary: stockout, when you run out of product to sell →

the rhythm by shift, with a holiday in the middle

With five days and a Monday holiday, a simple scheme beats a detailed one. Three active reviews a day, ten minutes each, in which you look only at the first list: velocity against stock per SKU, published price against approved price, timestamp of the last sync, and budget consumed. One at open, one mid-day, one before traffic winds down.

Between reviews, look at nothing. Let things come to you. Well-configured alerts are what turn a weekend of surveillance into a weekend of exception handling, and that is the difference between reaching Tuesday functional and reaching it wrecked.

Treat the Monday holiday as another weekend day rather than a business day: the traffic pattern looks more like Saturday than like Monday, and your logistics operator may be running reduced staff. If you will need a physical inventory move, plan it before Friday.

what live monitoring looks like in iqseller

The three live signals are, in practice, three views. In iqseller the Inventory module shows real available per SKU alongside stock by warehouse, by Amazon fulfillment center, by Full and by 3PL, and the Forecast module computes coverage against recent velocity, which is the Friday math from the example. Because it sits on the Parent, Model and SKU tree, one size running out does not hide behind the parent total.

The Pricing module compares the price published in each channel against the calendar you approved, and Alerts sends the critical-stock, price-discrepancy and sync-failure notices, which are exactly the three signals on this list. The Profitability module is the one you would use in the nightly review: net margin per SKU with your loaded COGS, commissions from the Amazon settlement and from MercadoLibre orders, FBA and Full fees, shipping, advertising and the VAT and withholding breakdown.

What iqseller does not do is decide for you. It does not raise prices because velocity spiked, and it does not pause campaigns: it shows you the signal with a current number, and you execute the action in the channel, with the criteria you defined before the event. It also does not replenish physical inventory or control your logistics operator’s timelines.

the rule that saves your weekend

If you keep one idea from all of this, let it be the one already developed in data that does not drive a decision is noise, applied to the peak: during Buen Fin, a number with no action executable in minutes does not get watched live. It gets noted and read at the close.

That leaves three things on screen: coverage per SKU, aligned prices and a live sync. If those three are fine at every review, the event is going fine, regardless of what the sales counter says. And if one of them is off, you already know what to do, because you wrote it down before it started.