Agree totally with Rosie
I mentioned the gro numbers, as you will need that to apply for the copy of the marriage cert.
The numbers referencing a marriage cert will, in no way, lead you to the birth cert for that person!
Research cannot be done by relying on computer "matches" - they can only match the obvious and real life isn't like that. People remarry, use a different name, move area unexpectedly etc.
For an example - ( Ican't remember the exact dates etc without looking it up, so don't go checking it out, but this is the principle)
I had a Martha Parker, with a brother James. Martha was born in, say 1862 in Worthing Sussex. Her brother James married a Martha Price. Martha Price was also born in within a year of 1862 in Worthing. So - of course - all computer programme "matched" the death of Martha Parker with a birth date of 1862 with the birth of Martha Parker in 1862. If I hadn't looked further, it would have been so easy to have accepted this. Of course the actual situation was that this death was for Martha nee Price, and the Martha Parker who had been born in 1862 died as Martha something else as she too had married.
Just a silly example, but shows how you cant accept a "match" on face value, but only take it as a possibility to be followed up and confirmed with research. Same with online trees, as has been said a large proportion are absolute rubbish, but just sometimes something is worth a closer look. But that doesn't mean just accepting it, but looking closer at it and engaging in some research to see if there are documents or evidence to back it up.
So, if you think you have found the marriage in the gro index, and cant find a parish record for it, then you need to get the cert to find out more about it before moving on to the next step. A computer may have "matched" all sorts of births to that marriage - but none of them may be correct! And you certainly cant just pick one randomly with no further evidence to have lead you to it.