the worse case for a BLOCK was about 13 confirms..this was due to a major bug, rather than normal disagreement consensus
but you asked about a TX confirm. which is more individual and harder to really chart/locate information about
(tx's in opposing blocks normally get accepted in the rule following block asap. so a
TX would in most cases been re-accepted in under 13 confirms usually)
(apart from if there was a high backlog of TX's in mempool and the rule following block creator cherry picked different tx's than the orphaned block and left some tx's lingering)
(apart from in double spend activities where a spender decides not to re-broadcast his tx)
.. yea its more complicated
https://bitcoinmagazine.com/articles/bitcoin-network-shaken-by-blockchain-fork-1363144448/ the 0.8 chain was 13 blocks longer than 0.7. But that was the furthest that the 0.8 chain would get ahead. By then, the two chains were growing roughly in lockstep, and at about 03:30 the tipping point came. The 0.7 chain quickly caught up to being only 10 blocks behind, then 8 blocks, and at 06:19 both chains converged to the same length at block 225454, leading to nearly all remaining miners abandoning the other.
though most tx's were "confirmed"(eventually) in both chains where a tx would seem confirmed on 0.8, then orphaned then confirmed on 0.7 as the merchants switched and as 0.7 caught up.
there was a chance that some tx's dropped by 0.8 were not picked up by 0.7 and needed a user to rebroadcast, where if a user did not rebroadcast they would just keep the coins. and if lucky would have also have already got their goods or service too when it was initially confirmed on 0.8. thus the merchant loses out and the customer can spend their funds later on elsewhere (this is called double spend) or as you using laymans terms call "reversal".
..
anyway
prior to core becoming the de-facto rule setter last summer and anything else being deemed an enemy. there were a chance of upto 4 orphans happening a day due to normal consensus disagreement. most(nearly all) were of 1 confirm blocks that just got dropped within seconds.
https://blockchain.info/charts/n-orphaned-blocks?timespan=2years though now due to cores heavy handedness of litterally banning nodes that do not conform to cores way. the chances of disagrement has dropped
some see it as more confidence to accept tx's with a lower confirm. some see it as core centralised the network that no one can challenge them
where by if core did get a bug, if everyone was also blindly using cores code equally, everyone would have the bug too and so the network would stall or break due to everyone getting affected.
but lets talk about before last summer. and the rationale for 6 confirms
this rationale is that in normal circumstances people should not trust 1 confirm for large amounts as "immutible"
because each day there were up to 4 occurances of 1 confirm orphans(2.8% risk a day).. basically every 36th block a block of 1 confirm may orphan
then the maths increase where once every fortnight a block of 2 confirms may orphan due to normal consensus disagreement
then the maths increase where once every 3 years a block of 3 confirms may orphan due to normal consensus disagreement
and so on
to the extent that at 6 confirms. due to normal consensus disagrement(not a bug) the chance of a 6 confirm orphan happening were centuries
