Independent Films by the Numbers

The marketing of Independent Films

Archive for November, 2007

The Lathrios Festival Search Engine is Live

I have just launched version 1 of my Festival Search Engine. It allows someone to search for films or festivals. It is the research tool that I wanted a year ago when we started marketing our latest film.

http://www.lathrios.com

I have some 18 film festivals in the database and will be adding more over the course of the next weeks.

Comments positive or negative are welcome.

1 comment

Best Run Time for Festival Short Films

Deciding upon a run-time for your independent short film is a classic optimization problem. On one hand, you want to make the film very short (~5 min) so it can most easily be slotted into a festival calendar, but it is best to make a film longer (>8 min) to maximize your chances of winning an award. Obviously, these two goals are at odds with one another.

The key is to find a time that is the best blending of timing to be screened and ability for that runtime to win an award.

This is not an easy task. There are a number of ways to do this process given the incomplete data we have, which will give you varying insights. A true understanding of the exact tipping point is impossible with the dataset I have since there are not enough data to provide a statistically valid sample at a very granular level. That being said, some general trends can be discerned.

The 8 minute mark is the tipping point for films having a long enough run time to allow them to win an award. Films shorter than this probably have trouble competing, since one can only do so much story development in 5 minutes. Despite their craft, these films naturally feel less substantial than longer films where more story development can occur.

The 21 minute mark is another tipping point at which the film becomes harder to slot into a festival given the constraints of their schedules.

Within the 8-21 minute span of time, there appears to be a maximum point at around 15 minutes where placement is maximized and wins are solid. My gut tells me that if you have a strong product this optimal run time can be pushed up to around the 18 minute mark without much risk. Indeed, the strength of the 15 minute maximum probably results from being a “round” number for filmmakers and the resulting proliferation of films of this length.

If you have a short film at the far end of the time spectrum (approaching 40 min), the strategy for success is more risky, but given the strength of these films to win awards it can be more rewarding. The key for these films is placement at a top tier festival (Sundance, Tribeca, Toronto, etc…) and have an award win there. If this happens, the buzz around the film will allow you to do well on the festival circuit and not be constrained by the normal restrictions of festival schedules. How can you do that? I leave that up to you and lady luck for the moment.

Comments are off for this post

CHAID Analysis of Festival Films

Last night, I completed a CHAID analysis of my database of festival films exploring what is predictive of a film winning an award or showing at more than one festival in my sample.

CHAID is a common statistical tool of direct marketing optimization and is a fun as stats get. Essentially, it is a massive exploration of a dataset using countless Chi-Square tests. The output from this analysis is commonly referred to a decision tree, which allows the researcher to segment the database using a series of variable predicting another variable. In this case, I used this technique to explore what variable are predictive of winning one or more awards and the showing of a film in more than one festival.

The results are broad, but here are the highlights:

FESTIVAL AWARD WINS

  • The structure of the title seems to have little influence on whether a film would win. There were some loose patterns of adjectives use leading to increased award potential for non-fiction films and the use of a vowels to start a title, but it was not much beyond normal variance created by chance alone. Other parts of speech metrics, number of words in a title, and title length analysis proved insignificant.
  • Self-described country affinity did have some patterning. Films with an affinity with the USA did marginally better than non-USA films, which is not surprising given the number of US film festivals in my sample. Japanese affinity films (n=52) did not garner a single win, although this could be the result of so many Japanese films in my sample being classic films that are not in contention for an award. This being said contemporary Japanese film is under-represented in my sample. Lastly, films with a declared affinity to Israel (made in or dealing with) have a five times greater chance to win an award than films without this affinity. Political correctness prevents me from reading too much into this last insight.
  • Non-fiction films seem to be more likely to win awards than fictional ones. This is perhaps the result of the larger volume of fictional films than non-fictional films at festivals paired with the existence of special documentary awards. This insight is prone to issues resulting from a large number of films in my sample not tagged for this data column.
  • It should come at no surprise that film that show at multiple festivals tend to win more awards. This is a chicken and egg issue, since are they showing at more screening because they won an award or do they increase their chance of winning awards by showing at more festivals. I think it is a mix of both given my experience.
  • As I noted already, short films that are longer than 8 minutes win tend to win awards more than shorter shorts.

MULTIPLE SCREENINGS

  • Sundance seems to be the strongest festival for predicting play at multiple festivals. Some of this result may be timing, since Sundance occurred early in my sample. The general result of this CHAID for festival is that there are two kinds of festivals international feeders (Sundance, Cannes, Toronto, Tribeca, Berlin, etc…) and regional/specialty festival (LA, Newport, Seattle, etc…). Doing well at the feeder festival tends to open up doors for playing at regional/specialty festivals. Again, this should be no surprise to anyone with experience of the festival circuit.
  • Like with award winning behavior, the title structure seem unimportant to success at winning awards
  • National affinity tending to look like alot like the data for winning awards, except that the trends were weaker across the board for these traits.
Comments are off for this post

What is the best running time for winning awards?

I have begun an exploration of what makes for a winning awards film at festivals and once again the run time of the movie seems to be a prime mover. The analysis is not fully complete, but I do know the following:

  • Short movies tend to win more awards the longer they are… if they both play in the same festival, a 35 minute short is more likely to win an award than 5 minute film.
  • There are two run time inflection points for short films in my studies so far — 8 minutes and ~16 minutes. A CHAID study shows that shorts running longer than 8 minutes are twice as likely to win awards than ones less than 8 minutes. This pattern exists at an adjusted P-value of 0.04, which means we have 96% confidence that this pattern is not the result of chance. The inflection point of 16 minutes is based upon an analysis of the chance of winning an award for a given running time mixed with the chance of getting into a festival based on the running time. It seems that the mid-teens to twenty minute mark is the optimal point. I am working on refining this.
  • For features, it seems like the optimal length for winning awards is somewhere around 80 minutes
Comments are off for this post

First Letters of Film Titles

I have often wondered if the first letter of the title matters for film. I do know that films sold in catalogs tend to do better if they begin with first letters of the alphabet… simpily because buyers wil see these titles first when scanning a list of films.

I have completed my first go at the film database and have calculated the first letters for films in my sample. The results are as follows for the top 10 starting letters:

S – 10.1%
M – 7.5%
B – 6.4%
C – 6.3%
L – 6.0%
T – 5.9%
A – 5.8%
D – 5.5%
P – 5.1%
F – 4.9%

I looked at another comparative sample of english language titles to see how these compared – an analyis of Wikipedia titles done in 2004.

http://en.wikipedia.org/wiki/User:Jamesday/report/letter_frequency

The most striking finding using a Chi-Squared analysis is there are many fewer “J” titles in my sample and many more “I” titles. The “I” titles make sense since Wikipedia is not expected to have so many titles starting with 1st person pronouns.

Other trends: Fewer than expected C, K, U, V, and 1 titles; and more than expected S, M, D, F, W, I, O, and U titles in my sample compared to Wikipedia.

I don’t know what this all means yet, but it is fun all the same.

Note: I used the first letter of the second word for all titles starting with “A [word]” and “The [word]” patterns.

Comments are off for this post