I've experienced this several times on the South-Western at Kingston.
I've experienced this far too often - before I retired, barely a week went by without a train being non-stopped from Teddington to Waterloo or vice versa, thereby skipping the busiest stations on the route (Kingston and Norbiton) to ensure a service to the Shepperton branch (whose combined usage stats (2,202,478 for the six stations in pre-COVID days, 2019/20) were less than Norbiton's (2,283,210) and less than half of Kingston's (4650286). SWR and SWT alike treated Shepperton as an important desination and Kingston as a wayside halt. Running empty trains past crowded platforms seems a poor use of resources.
A favourite trick of "Control" was to terminate one train in the bay at Kingston, dumping all the London-bound passengers to catch the next train - and then run the next one through without stopping. The excuse that this was "to reduce congestion" did not go down well on what would inevitably become an extremely congested platform 3 (which has only very basic facilities because the main facilities are on platforms 1 and 2, on the down side, an arrangement dating from when the station was a terminus and not updated in the 150 years since it became a through station).
And far too often the run-through would not be routed on to the fast lines at New Malden, so simply joined the queue of stoppers on the slow lines and made up no time at all. Whether this was because "Control" forgot to tell the signallers of their intentions or some other reason I don't know.
To add insult to injury, the approach of a run-through wouild be anno0unced as "please stand back, the approaching train is not scehduled to call at this station", which everyone knew was not true - it WAS scheduled to call, but "Control" had other priorities than getting their clients to work on time and in comfort.
The station staff at Kingston were rarely of any help - seemingly having less information than that available to any passenger with a smartphone, rarely visible on the platform 3 unless there was a train present, and inclined to give out misinformation - on a recent occasion we were told "No trains to Waterloo" when in fact there was a train five minutes away (albeit going the long way round via Richmond)
I was given the excuse that a late arrival at Waterloo would cxause delays to spread to whichever line the train was scehduled to operate on next, which again shows that the staff knew less about the scheduling than could be easily determined from a smartphone (or indeed the pages of Modern Railways) Although some diagrams do cycle through several branches (what comes in from Chessington goes out to Guildford, then comes back and goes to Hampton Court, etc), the diagramming of both the Shepperton and Kingston Loop services are, and have been for many years, self contained.
Doesn't always work. Last autumn, I was travelling from Leeds to Skipton in the morning. The incoming service was substantially late and so in a vain attempt to recover the diagram, Northern decided to run it non-stop to Skipton. This was clearly announced on platform and train. Great idea, which, if it had worked, would have put the set in Skipton in time for an almost on-time deaprture back to Leeds.
The train got as far as Shipley where, "cleverly", the preceding Bradford > Skipton stopper was NOT delayed by all of about 2 minutes to allow the (now) "express" to proceed at line speed; rather it was dispatched on time. And so the (now) "fast" train followed it stop/start arriving in Skipton something like 15 minutes after it should have departed for its return.
Not very joined-up thinking IMO.
As I said in my earlier post, this lack of joined up thinking seems to be all too common. If a train is to run empty its headcode is chnaged from a Class 2 to a Class 5. Would it be possible, if a train is to "skip stop", to make it a class 1 to inform/remind the signalling staff? After all, the whole point of skip stopping is to make up time.
A recent failure of joined up thinking, resulting in both trains being delayed further, was a late decision to skip stop a delayed train, resulting in most of the passengers having to transfer to the next train (which was already waiting at the adjacent platform). However, one of those passengers was in a wheelchair, resulting in a further delay whilst staff and a ramp were found so that the passnger could leave the first train (and a different ramp to allow him to board the other one, as 455s and 707s need different ramps). This delayed the first train by a further seven minutes, completely wiping out any saving that might have been made by the skip-stopping. (The obvious solution would have been to skip all the stations except the one the wheelchair user was going to)