Prices are information

Demand is not "how much people want"; it is a schedule: at each price, how many units would be bought. It slopes down because at a lower price more people find the thing worth it, and the people who already found it worth it buy more. Supply is the same schedule for sellers: at each price, how many units would be offered. It slopes up because a higher price makes it worth making the harder, dearer units.

Hold the price above the crossing and sellers offer more than buyers take: a pile of unsold stock, and the seller who cuts first sells first. Hold it below and buyers want more than is offered: a queue, a waiting list, an out-of-stock page, and the seller who raises first loses nothing. The crossing is the one price with no pile on either side. That is all "equilibrium" means, and it is why a price drifts there without anybody steering.

Surplus is what the picture adds that a spreadsheet does not. Every buyer who would have paid more than P* and did not have to is better off by the difference, and every seller who would have accepted less is likewise. The triangle between the curves and the price line is the total gain from the trade existing at all. It is the quantity that a badly set price destroys, and it is invisible in revenue.

A price on any real shelf is somebody's decision, made at some point in the past, with whatever they knew then. It sits near the crossing because the piles are costly, not because it was computed. Three things keep it off the crossing for long stretches. Menu costs: changing a price takes effort and can annoy customers, so sellers wait until the pile is large. Uncertainty: the seller does not know the demand curve; they know last month's sales at last month's price. Rules: a price held by a contract, a regulator or a platform cannot drift at all, and the pile becomes permanent, which is what a shortage under a price ceiling is.

The data habit this gives you: when a price has not moved for a year, you have learned that the seller was comfortable, and nothing about the curve. Only a price that moved tells you how buyers respond, and that is the first filter the build-along applies to a real marketplace.

Say a price out loud as a sentence and it reads: "Given everything everyone knows, the last unit of this was worth about this much to somebody, and cost about this much to somebody else." That sentence is short enough to act on and long enough to be useful. A buyer who sees it can decide without the backstory; a seller who sees it can decide whether to make one more.

The same compression is what a price hides. It does not say why. When a data team is asked to "explain the price increase" they are being asked to decompress the message, and most of the decompression is not in any table they hold: the mine, the factory, the fashion are in four other people's heads. The honest analysis says which part of the movement the data can account for and which part it cannot, and that habit, saying what the numbers can and cannot carry, is one the deep-dive track applies to a real elasticity in its second session.

One caution that stops a common overreach: a price is a message only when it moved. A price that has sat at R$ 69.90 for a year tells a stranger that the seller was content. It carries no news, and it says nothing at all about how buyers would respond to R$ 79.90. That distinction is the whole of Part 3.

Hold a price below the crossing by rule and the message stops arriving. Buyers keep buying as if the thing were plentiful, sellers stop offering the dearer units, and the gap between the two becomes a queue: rationing by waiting, by connection, by luck. Hold a price above the crossing and the message inverts: sellers offer units nobody will take, and the surplus has to be stored, destroyed or bought by whoever set the floor. In both cases the price was frozen precisely because the message it would have sent was unwelcome, and the cost of not sending it lands on somebody anyway.

None of this says a ceiling or a floor is always wrong. It says that a held price does not make the scarcity go away; it changes who bears it and removes the signal that would have told others to help. Whether that trade is worth making is a question about values, not about curves, and the curves only tell you what the trade costs.

A movement along the curve has a signature: the price changed and the units changed the other way. A shift has a different one: units changed and the price did not, or the price changed and units went the same way, which no demand curve can produce on its own. So the first test is embarrassingly simple: look at the price column before the units column.

The second test is harder and is where most monthly reports fail. A price cut in January, on a marketplace that was also growing 20 percent a year, moves the point along the curve and shifts the curve in the same month. The two are tangled and one product cannot untangle them. What can is comparing many products with themselves over many months, so that the growth every product shared can be separated from the price change only one of them made. That separation is the whole of deep-dive session 2, and its answer on the real marketplace is that "the curve moved" accounts for almost all of the naive number and "along the curve" is what is left.

The 45 and 121 in the figure are real. They are one product's units in December 2017 and January 2018, the month its price went from R$ 75 to R$ 69.90. The figure draws both stories on those two numbers because, from those two numbers alone, both are true.

Demand shifts when something other than the price changes how much people want at every price: income, the price of a substitute or a complement, a fashion, a review, the season, a rival going out of stock. Supply shifts when something other than the price changes what sellers will offer at every price: the cost of an input, a new technology, a tax, a new competitor entering. A marketplace growing shifts both: more buyers at every price and more sellers at every price, which is why the total units on a growing platform tell you almost nothing about any one product's price.

The list is not for memorising. It is for asking, before any "the price did it" claim: which of these also changed that month, and did it change for every product or just this one?