What the Same Car Did to Its Own Price
A market median compares different cars each month. This study follows 28,787 individual cards through their own price history: 29.4% changed price at least once, four out of five first moves were cuts, the median first cut was 4.5% or $500, and it came 7.3 days after listing. Cars already on Georgian plates move price more often than import stock.
The problem with every market median, including ours
When a market median falls from $14,000 to $13,500, two completely different things could have happened. Sellers could have cut their prices. Or cheaper cars could have arrived and dearer ones left, with no individual seller changing anything.
A median computed over a monthly cohort cannot tell those apart, because it compares a different set of cars every time. That is a real limitation of the pulse reports and of every asking-price index published anywhere, and the only way past it is to stop comparing cars to each other and start comparing each car to itself.
Each listing's full price history is kept, so every car can be compared with itself and stays in the cohort it joined. This study follows 28,787 cards first published between 1 March and 30 June 2026 through their own histories.
How many times a card changed its price
8,466 of 28,787 cards moved at all; 20,321 never changed price in the window.
Seven in ten cars never move at all
The first finding is how still the market is. Of the 28,787 cards, 8,466 — 29.4% — changed price at least once in the study window. The other seven in ten were advertised at one number from beginning to end.
Among those that did move, movement is not habitual either. 4,894 cards changed price exactly once. 1,848 changed twice, 1,237 three or four times, and only 487 cards moved five times or more. In total the cohort produced 16,738 price steps — about two steps per card that ever moved.
So Georgian asking prices are close to sticky. The negotiating that happens in this market happens off the card, in conversation, and a published price change is the exception rather than the routine.
Seven in ten cars never move at all
The first finding is how still the market is. Of the 28,787 cards, 8,466 — 29.4% — changed price at least once in the study window. The other seven in ten were advertised at one number from beginning to end.
Among those that did move, movement is not habitual either. 4,894 cards changed price exactly once. 1,848 changed twice, 1,237 three or four times, and only 487 cards moved five times or more. In total the cohort produced 16,738 price steps — about two steps per card that ever moved.
So Georgian asking prices are close to sticky. The negotiating that happens in this market happens off the card, in conversation, and a published price change is the exception rather than the routine.
How many times a card changed its price
8,466 of 28,787 cards moved at all; 20,321 never changed price in the window.
Share of cards that changed price, by customs status
Cleared: 4,232 of 12,368. Uncleared: 4,234 of 16,419.
When a price does move, it moves down, early, and by little
Of the 8,466 first moves, 6,778 were cuts and 1,688 were increases — four cuts for every raise. The median first change across all of them is −3.6%; among cuts alone it is −4.5%, or $500 in cash.
The timing is the part worth internalising. The median first change came 7.3 days after the car was listed. Not a month, not a season — one week. Sellers who intend to reprice mostly discover it almost immediately, which suggests the first price is frequently a probe rather than an expectation.
For a buyer this is directly actionable. A car that has been advertised for two weeks at the same price has already passed the moment when most sellers who were going to move did so; that price is more likely to be real than the price on a card posted yesterday. For a seller the reading is uncomfortable but useful: if the first week is quiet, the market has usually already told you, and a $500 adjustment is what four out of five repricing sellers land on.
When a price does move, it moves down, early, and by little
Of the 8,466 first moves, 6,778 were cuts and 1,688 were increases — four cuts for every raise. The median first change across all of them is −3.6%; among cuts alone it is −4.5%, or $500 in cash.
The timing is the part worth internalising. The median first change came 7.3 days after the car was listed. Not a month, not a season — one week. Sellers who intend to reprice mostly discover it almost immediately, which suggests the first price is frequently a probe rather than an expectation.
For a buyer this is directly actionable. A car that has been advertised for two weeks at the same price has already passed the moment when most sellers who were going to move did so; that price is more likely to be real than the price on a card posted yesterday. For a seller the reading is uncomfortable but useful: if the first week is quiet, the market has usually already told you, and a $500 adjustment is what four out of five repricing sellers land on.
Share of cards that changed price, by customs status
Cleared: 4,232 of 12,368. Uncleared: 4,234 of 16,419.
Locally registered cars are the flexible ones
Splitting the cohort by the customs status its cards entered with produces the least intuitive result in the study. Cards that entered already cleared through Georgian customs changed price more often than cards that entered as import-lane stock: 34.2% against 25.8%.
The instinct is the other way round — importers carry financing costs, hold inventory, and should be the ones adjusting. The data says the household selling a car already on Georgian plates is the more responsive party. Two explanations fit and we cannot separate them from listings alone: an individual seller has one car and a reason to sell it, so a quiet week produces a price change; an importer's stock is priced against a landed cost that does not move, and an unsold car is more likely to be held than discounted.
It is worth stating plainly that this measures asking prices. We do not observe what any car actually sold for, and a cut is evidence of a seller's expectation changing, not of a transaction.
Methodology
AutoBridge listing database, using the recorded price history of each listing.
28,787 listings first published between 1 March and 30 June 2026, each followed through its own price history. A price step is counted only when both the price before and the price after are at least $2,000, so a switch to price-on-request is not scored as a total discount. Listings are followed regardless of what later happened to them.
Price paths from 1 March 2026 to 22 July 2026.