Jump to content

Agafin

Frontloading: Has it really changed in the past five years?

Recommended Posts

I remember reading a lot of comments in the Dory OW thread about how it couldn't match Toy Story 3's multi because movies are supposedly much more frontloaded today than they were just five or six years ago. After less than three weeks, we now know that Dory will come close to, if not match said multiplier despite a bigger opening. This made me curious about whether or not there has really been a general increase in frontloading or if that is just an unfounded perception. After looking at all the Yearly box office results from the past five years (2011 to 2015), I honestly haven't noticed any particular worsening of multipliers whatsoever. In each of those years, a little over half of the movies in the Top 100 (55 to 59) have a multi equal to or greater than 3 (a little less than half if you exclude limited releases or Non-Friday openers). Last year, we also had a movie break the OW record and get a multi of 3.77. The last time a movie broke the OW and had better legs than that was in 1993!

 

After looking even further, I noticed that frontloading definitely had a drastic increase at the turn of the century (in the late 1990s, sophomore holds of 30 to 35% were the norm while the early 2000s shifted that to 45 to 50%) and kept increasing throughout last decade till it reached a peak at around 2009.
Since then nothing has changed. Frontloaded movies with huge preview shares (like Twilight or Potter) generally tumble 67 to 72% in their second weekends and have multis ranging from 2.0 to 2.2, family films still have the best multis (3.5 to 4.5), record openers ($150m+) have a hard time getting multis above 2.6 (but some can still get that and even 3+ if they reach "event satus") etc.

 

So really, why do people keep saying that movies are more frontloaded today than they were 5 or 6 years ago? Are people actually subconsiously referring to the 90s instead? Or am I missing something?

  • Like 4
Link to comment
Share on other sites



I hear some stuff like this from a lot of different topics.  Sometimes it makes sense, sometimes it's silly.

 

For example, discount Tuesday has been a big thing for a few years now but every year I hear, "Discount Tuesday wasn't a thing like it is this year."  And, while yes, discount Tuesday has gradually gotten bigger, it's been pretty big for quite some time.

 

Legs, id agree haven't really worsened from 2009.  Maybe a little factoring in previews instead of midnights. Certain franchises have gotten more frontloaded as they've grown in fanbases, but overall I'd agree with you.

Link to comment
Share on other sites

I think while frontloading overall hasn't necessarily changed, two things have occurred:

 

1. Thursday previews have aided in making weekends a bit more frontloaded, at least when it comes to fan-driven movies.

2. Word-of-mouth spreads faster than ever before, and poorly received films sink faster than they would have done even five years ago. I think had Batman V Superman for example come out five years ago it would have cleared 2x with some margin.

 

Quote

Last year, we also had a movie break the OW record and get a multi of 3.77. The last time a movie broke the OW and had better legs than that was in 1993!

 

Apples to oranges. TFA is the first record breaker since Star Trek: The Motion Picture to be released in December. With the exception of the first Harry Potter film all record breakers since were between May and July.

Edited by cookie
  • Like 1
Link to comment
Share on other sites



Generally speaking, I do think frontloading has gotten more significant in the last five years (and yes, it's been on a steady increasing trend since the turn of the century). I think there's three basic factors: increase in sneak previews (both in total showtimes and for also all movies), wider release patterns, and shorter overall theatrical runs. 

Link to comment
Share on other sites



Wouldn't another factor also be an increase in competition? Not only are supposed tentpoles released week after week, but also now they're showing up at any time of the year. Soon, the slow month of September will look like an animation battleground and October and November will display multiple super-hero movies. What once were months showcasing Oscar dramas will now have intermittent animation, super-heroes, and live-action fairy tails. Multiple bombs are seemingly required to prime the market for break-out successes w/ long legs.

Link to comment
Share on other sites



1 hour ago, Agafin said:

I remember reading a lot of comments in the Dory OW thread about how it couldn't match Toy Story 3's multi because movies are supposedly much more frontloaded today than they were just five or six years ago. After less than three weeks, we now know that Dory will come close to, if not match said multiplier despite a bigger opening. This made me curious about whether or not there has really been a general increase in frontloading or if that is just an unfounded perception. After looking at all the Yearly box office results from the past five years (2011 to 2015), I honestly haven't noticed any particular worsening of multipliers whatsoever. In each of those years, a little over half of the movies in the Top 100 (55 to 59) have a multi equal to or greater than 3 (a little less than half if you exclude limited releases or Non-Friday openers). Last year, we also had a movie break the OW record and get a multi of 3.77. The last time a movie broke the OW and had better legs than that was in 1993!

 

After looking even further, I noticed that frontloading definitely had a drastic increase at the turn of the century (in the late 1990s, sophomore holds of 30 to 35% were the norm while the early 2000s shifted that to 45 to 50%) and kept increasing throughout last decade till it reached a peak at around 2009.
Since then nothing has changed. Frontloaded movies with huge preview shares (like Twilight or Potter) generally tumble 67 to 72% in their second weekends and have multis ranging from 2.0 to 2.2, family films still have the best multis (3.5 to 4.5), record openers ($150m+) have a hard time getting multis above 2.6 (but some can still get that and even 3+ if they reach "event satus") etc.

 

So really, why do people keep saying that movies are more frontloaded today than they were 5 or 6 years ago? Are people actually subconsiously referring to the 90s instead? Or am I missing something?

 

I feel like people just saying "YES it has increased" are missing the bolded part of her(his?) post.

 

Yes people, more blockbusters coming out every month, faster WOM, shorter runs, more theaters per movie opening and increase in previews all should, logically speaking, point to bigger frontloading, but why are you all dismissing her(his?) research?

  • Like 1
Link to comment
Share on other sites



Alright, so I just graphed all the multipliers of films with $50M OW or more against the date. I didn't adjust for inflation, so the graph is biased towards more recent films (which could be a big problem). I'll fix that if I find a table with adjusted OW or when I get a chance to do it manually.

 

There's plenty of other reasons to take the graph with a huge grain of salt, but here it is:

V32gEDS.png

 

Based on the line of best fit, the average yearly decline in the multiplier of films with OW over $50M is 0.046. However, only about 5% of the variance in multipliers can be explained by the date. (Which isn't an unexpected result, in my opinion.)

Edited by Jason
  • Like 1
Link to comment
Share on other sites

I remember when HP3 came out in 2004. Some reports claimed that HP3 would beat Spider-Man's OW record after Friday estimate came out. Then surprisingly, it decreased on Saturday. 

 

I think today's marketing is more geared towards frontloading of BO performance. Movies have to earn money quickly before competitions arrive in a couple of weeks. 

Link to comment
Share on other sites



Yes, and it makes sense, considering franchises run Hollywood these days. Releases are super wide to get big opening weekends and make a lot of money so that they can replace the film with another big weekend earner.

 

Its just another reason original or non-franchise films are having a harder and harder time being hits. They get released one weekend and within only a few short weeks they'll get taken out despite the fact that they need WOM to develop legs and in turn a financially successful run.

 

Another nail in the Hollywood coffin.

Link to comment
Share on other sites

On closer inspection of the graph I did above, the trend is noticeable from around 2000-2010, but isn't noticeable for the past 5 years. So I just did a quick graph of the 2011-2016 films (inclusive), again using unadjusted figures:

8g59qLb.png

 

Based on this data set of films with unadjusted OW of over $50M, there really isn't any correlation at all in the past five years. (The trendline does go upwards slightly, but the R-squared value of 0.009 suggests 1% of the variance can be explained by the date - which is so small it's probably just noise.)

Edited by Jason
  • Like 4
Link to comment
Share on other sites



Guys, an R² of ~0.05 or less can definitely not be used to prove anything concerning a set of data. Like, not even close.

 

Usually, you want an R² around 0.75 (ideally 0.80) to be able to say that the 2 variables are correlated together. With ~0.5, and with good enough arguments, you might be into something.

 

Not saying that front-loading hasn't been increasing, but your data and the way you presented it, do not indicate anything of the sort. Correlation can be misleading if, for example, you assume the variables are linearly correlated, when in fact, they are correlated in a different manner (exponentially, for example)

  • Like 2
Link to comment
Share on other sites



1 hour ago, Daxtreme said:

Guys, an R² of ~0.05 or less can definitely not be used to prove anything concerning a set of data. Like, not even close.

 

Usually, you want an R² around 0.75 (ideally 0.80) to be able to say that the 2 variables are correlated together. With ~0.5, and with good enough arguments, you might be into something.

 

Not saying that front-loading hasn't been increasing, but your data and the way you presented it, do not indicate anything of the sort. Correlation can be misleading if, for example, you assume the variables are linearly correlated, when in fact, they are correlated in a different manner (exponentially, for example)

 

I agree it can't prove anything. In a situation like this though, even if there is a relationship between frontloading and time progression, we couldn't possibly expect an R-squared anywhere near 0.5 because there's obviously far more important factors explaining the variation in multipliers (WOM being the most obvious, although hard to quantify). An R-squared of 0.05 simply means only 5% of the variance is being explained by an imperfect linear model.

I don't expect to be able to prove that that 5% of the variance is statistically significant so I'm probably not going to bother trying. (And I'll be honest, my stats is too rusty for me to recall the best way of going about it, anyway.) But if we set the bar for what we're discussing here to statistically significant, P<0.05, we might not even be able to prove indicators of WOM are correlated with multipliers.

Edit: Now that I think about it, given the size of the sample I think there's a good chance that the R squared of 0.055 in the first graph actually is significant (P<0.05). But hopefully @MikeQ is on the job (see below). Also, I take back what I said about indicators of WOM, there's probably more than enough data for that. But I'm not volunteering to do it!

Edited by Jason
  • Like 1
Link to comment
Share on other sites





17 minutes ago, Jason said:

 

I agree it can't prove anything. In a situation like this though, even if there is a relationship between frontloading and time progression, we couldn't possibly expect an R-squared anywhere near 0.5 because there's obviously far more important factors explaining the variation in multipliers (WOM being the most obvious, although hard to quantify). An R-squared of 0.05 simply means only 5% of the variance is being explained by a crappy and probably incorrect linear model.

I'm quite sure it'd be impossible to prove that that 5% of the variance is statistically significant so I'm not going to bother trying. (And I'll be honest, my stats is too rusty for me to recall the best way of going about it, anyway.) But if we set the bar for what we're discussing here to statistically significant, P<0.05, we might not even be able to prove indicators of WOM are correlated with multipliers.

 

Can't you just test R for significance? I can quickly do it by hand, if you tell me what your N size is. If the R-squared is 0.009, then R must be 0.0948. My guess is a significance test would say this association is not statistically significant over the last 5 years.

 

Peace,

Mike

Edited by MikeQ
  • Like 1
Link to comment
Share on other sites



1 minute ago, MikeQ said:

Can't you just test R for significance? I can quickly do it by hand, if you tell me what your N size is. If the R-squared is 0.009, then R must be 0.0948. My guess is a significance test would say this association is not statistically significant over the last 5 years.

 

Peace,

Mike

 

Mhmm, I just don't remember offhand how to do the signficance test, something Google could probably cure, but I gotta scoot real soon. Anyway, N = 84 for the second graph, and N = 191 for the first.

Link to comment
Share on other sites



17 minutes ago, Jason said:

 

Mhmm, I just don't remember offhand how to do the signficance test, something Google could probably cure, but I gotta scoot real soon. Anyway, N = 84 for the second graph, and N = 191 for the first.

 

Awesome, thanks. The association is significant in the first case, with the 191 multipliers over the last ~20 years that you have in your regression (p<0.002), and the association is, as I expected, not statistically significant in the second case, with the multipliers over the last 5 years.

 

This suggests that multipliers have significantly changed over the last couple decades (at least for films that have opened $50+ million -- and based on your scatterplot, the relationship is that over time, multipliers have gotten smaller/worse, ignoring all other factors). But that when confining the analysis to just the last 5 years, there has not been a significant change in multipliers.

 

For the first model, though R-squared is "only" 0.0555, and therefore only 5.5% of the variance in multipliers can be explained by the date, this isn't bad when considering that there are a host of other factors/variables that we would want to include in our model. Date alone explains a decent enough chunk of the variation, considering that we would imagine that WOM, release date, competition, genre, quality of the film, etc, etc (some factors of which are hard to quantify) would all be important factors to consider as well in determining multiplier.

 

Peace,

Mike

  • Like 3
Link to comment
Share on other sites



We live in the age of instant everything. Cinema viewings will increasingly get more frontloaded as streaming services become even more dominant. The window for cinema releases and home viewing has decreased and will carry on doing so. I would not be surprised if within the next 10 years, cinema viewings are relegated to a 2 month window (or less).

Link to comment
Share on other sites



Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.



  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use and Guidelines. Feel free to read our Privacy Policy as well.