For the purposes of this thread, I'm mostly going to be addressing the original ILM filming model. While Greg Jein did a fine job rebuilding it for Voyager, there are minor inconsistencies that are small enough to ignore. In other words, where the two differ, I"m going with ILM.
Let me also say that I tend to defer to creative/production intent whenever possible. That is, if the creator of a design or other Treknological thing says a certain thing works a certain way, I will assume they know what they are talking about - unless, of course filmed evidence directly contradicts them.
Now, those of you who remember me before my self-imposed exile will recall that I am a big Excelsior fan. I once began writing a tech manual before I lost interest, including creating this cross section:

I like most people went with the assumption that the Excelsior was the 467 meters long that the modelmakers and Mr. Probert had stated them to be. Note the comparison below.

However, on occasion, the actual relative scaling of the ship was dubious:




And if we examine the model herself, well, "By God, that's a big ship." But indeed, how big?

My reconsideration of the ship's size was partly fueled by my reading of an article in Geek Magazine. There, the interviewers discussed with ILM's Bill George (who appears to be the "father" of the Excelsior) the design and creation of the iconic Klingon bird-of-prey. Paraphrasing, George said that the director and producers picked the designs they liked, indicating one in particular as the direction to go, without formalizing the overall model plans. We have previously seen various Excelsior study models, including one that bears a strong resemblance to the finished product.
It is highly likely, in my opinion, that the "designing" and construction of the Excelsior model took place much as Bill described with the bird-of-prey: Nimoy and the rest selected the design they liked a most, and Bill and team took over from there. So, it is entirely likely that the detailing of the ship were not regulated by the 467 meter figure, and more simply just aimed to create a big ship.
Studying the ship in profile, I attempted to make the window rows into decks. (A fool's errand, I know.) I used a combination of methods, looking at several different images and attempting to graft onto a relatively "flat" side image of the ship deck lines. It is somewhat difficult to fully demonstrate the exact method, so bear with me.
Nonetheless, these were my results:


I also examined the secondary hull, as seen in TUC:

I ended up with this:

(Pardon the mess - I used a copy of my original cutaway to create it, hence the floating bulkheads.) I calculated 35 decks, adjusting for slightly taller engineering decks - partly as a conceit to Shane Johnson and partly because the saucer edge and secondary hull lights simply didn't seem to be spaced the same. I was struck by how this height actually seemed to match Doug Drexler's cross section of the Enterprise-B from "Generations."
And, compared to the D for fun:

Doesn't actually seem unreasonable next to the D. And honestly, this configuration seems to solve a lot of problems - such as the teeny tiny bridge dome. I'm not sure if this is the "real" size of the ship in the Trekverse, but I think it could work without being ridiculous. I am fairly convinced that this is about the size the model was built to be.
I'm conflicted, though by the knowledge that this flies in the face of production intent - and also, that there are clearly some instances (particularly on DS9 where the CGI model was used) where the ship was scaled to the official 467 meter size.
What do you think? Is a canonically bigger Excelsior possible? Have I drank the Flavor-Aid?
Let me also say that I tend to defer to creative/production intent whenever possible. That is, if the creator of a design or other Treknological thing says a certain thing works a certain way, I will assume they know what they are talking about - unless, of course filmed evidence directly contradicts them.
Now, those of you who remember me before my self-imposed exile will recall that I am a big Excelsior fan. I once began writing a tech manual before I lost interest, including creating this cross section:

I like most people went with the assumption that the Excelsior was the 467 meters long that the modelmakers and Mr. Probert had stated them to be. Note the comparison below.

However, on occasion, the actual relative scaling of the ship was dubious:




And if we examine the model herself, well, "By God, that's a big ship." But indeed, how big?

My reconsideration of the ship's size was partly fueled by my reading of an article in Geek Magazine. There, the interviewers discussed with ILM's Bill George (who appears to be the "father" of the Excelsior) the design and creation of the iconic Klingon bird-of-prey. Paraphrasing, George said that the director and producers picked the designs they liked, indicating one in particular as the direction to go, without formalizing the overall model plans. We have previously seen various Excelsior study models, including one that bears a strong resemblance to the finished product.
It is highly likely, in my opinion, that the "designing" and construction of the Excelsior model took place much as Bill described with the bird-of-prey: Nimoy and the rest selected the design they liked a most, and Bill and team took over from there. So, it is entirely likely that the detailing of the ship were not regulated by the 467 meter figure, and more simply just aimed to create a big ship.
Studying the ship in profile, I attempted to make the window rows into decks. (A fool's errand, I know.) I used a combination of methods, looking at several different images and attempting to graft onto a relatively "flat" side image of the ship deck lines. It is somewhat difficult to fully demonstrate the exact method, so bear with me.
Nonetheless, these were my results:


I also examined the secondary hull, as seen in TUC:

I ended up with this:

(Pardon the mess - I used a copy of my original cutaway to create it, hence the floating bulkheads.) I calculated 35 decks, adjusting for slightly taller engineering decks - partly as a conceit to Shane Johnson and partly because the saucer edge and secondary hull lights simply didn't seem to be spaced the same. I was struck by how this height actually seemed to match Doug Drexler's cross section of the Enterprise-B from "Generations."
And, compared to the D for fun:

Doesn't actually seem unreasonable next to the D. And honestly, this configuration seems to solve a lot of problems - such as the teeny tiny bridge dome. I'm not sure if this is the "real" size of the ship in the Trekverse, but I think it could work without being ridiculous. I am fairly convinced that this is about the size the model was built to be.
I'm conflicted, though by the knowledge that this flies in the face of production intent - and also, that there are clearly some instances (particularly on DS9 where the CGI model was used) where the ship was scaled to the official 467 meter size.
What do you think? Is a canonically bigger Excelsior possible? Have I drank the Flavor-Aid?