Discovering Cool Films and What’s Wrong with Movie Recommendation Systems

The biggest problem for an indie film is discovery. That is, having the audience/consumer know that you even exist!

Sure it’s a good move to build an audience early – before shooting – but after the film has been released and the initial buzz dies away it’s very difficult to maintain momentum. After all, if you have engaged an audience before the movie release, part of the thrill for the film fan is being first in the know; maybe influencing the project development in some way. With the movie out, there’s only so much evangelizing the core audience will do before they go off to find new projects.

Although the Internet and digital media has allowed many opportunities to get your work out there, just “getting it out there” doesn’t mean anyone’s searching for it and among those who are, will they find it?

Discovering new media

There are many ways people discover new books, music or films and there’s a good overview of the different systems here. I looked back at the media I’ve bought over the past month and it’s possible to see a combination of discovery methods at work from reviews to technology-assisted finds to friend recommendations. Here’s a quick summary:

• Gray’s Anatomy (book of essays by John Gray) and Nudge (audiobook, reviewed here) – discovered by my listening to the excellent podcast (radio show) Start The Week by Andrew Marr

• Tech Noir (book) – browsing book section of Forbidden Planet, London.

• Altered Carbon (audio book) – downloaded from iTunes. I actively looked for something outside the most popular titles. The synopsis read like something I might like and it had great reviews.

• The Inbetweeners (TV series) – recommended by two friends. It is hilarious. I’d guess that only a friend could have recommended this because you’d have to know about my social references and sense of humor. I’m not sure what movies or TV series in my collection would have given a clue to this… I love Gavin and Stacey, maybe that’s a clue? I doubt it though.

• The Baader Meinhof Complex (film) - good review on the Mark Kermode podcast and the kind of movie I like (political/historical/thriller)

• Let The Right One In (film) – a friend mailed the DVD from Sweden. It was also subsequently and independently recommended to me by a friend.

• Screamers (film) – recommended by Jinni (discussed more below).

• Death Cab for Cutie (music) – recommended by Pandora after I said I liked The Kooks. I’ve since bought all their back catalogue and these are now possibly one of my favourite bands in recent times; so great recommendation

• Matchbook Twenty (music) – I heard a tune in a store and used Shazam to tell me what it was. I bought the album

Looking at the above list, quite a lot of my recommendations come from friends and "curators" or what might be called "trusted sources" - many more than I would have thought actually. So while I'm about to criticize movie recommenders, don't forget the power of the social network (word of mouth) and PR.

So what’s the problem with movie recommendation systems?

Although many rave about Netflix because “there’s so much choice”, most members still continue to choose from the top ten. In fact, at a presentation by NetFlix, their man told the audience that people on all-you-can-eat subscription plans, members will watch what’s new, then the “classics”/cult films and then finally search further afield to more obscure titles. On the subscription plan, there’s no cost to trialing something new but that indie film is still at the mercy of the recommendation system. Netflix says it wants customers to find hidden movies because customers who can’t find anything to watch cancel their subscriptions. In fact, Netflix launched a competition to find a better recommendation algorithm than its current one, Cinematch. There’s an excellent article on the competition here at the New York Times.

The problem (for indie producers) is that most movie recommendation systems work on the basis of trying to match your likes and dislikes with another person (with the same or similar likes) and then assuming that if you both have the same tastes on films it knows you‘ve seen, you may also like films your doppelganger has seen but you haven’t. In the jargon this is called collaborative filtering. MovieLens works this way and Amazon works in a similar way with its “people who bought that also bought this”. This sounds like a reasonable approach but it means someone – anyone? - has to watch and/or rate your indie film for it to be discovered by others. Further, the more people that watch and rate your movie, the better chance of discovery you have. … and there’s the problem. Even with “so much choice” available, online markets are making blockbusters bigger and indie titles weaker. In fact, online recommenders are killing the niche.

Why?:

• because collaborative filters work better with more data… and where’s the most data? With the blockbusters

• even though there’s little or no overhead in selling “the long tail” it’s still in the online store’s interest to sell the most popular title.. the blockbuster

• fashions & fads and opinions on what’s cool and what’s not are being exchanged faster and wider than in the off-line world and this is actually contributing to an apparent homogenisation of taste… again, tending towards those blockbusters.

For an indication of how perilous life is going to become for the indie film business, look as usual to the music business. In an MCPS-PRS (UK music industry representing artists) study of one download site, from 13 million available tracks, 0.4% represented 80% of the revenue! This is corroborated by other studies by Nielson.

That says to me that you shouldn’t rely on technology to get discovered – especially not collaborative filtering. What we need is the film equivalent to Pandora where content is broken down to “genomes” that enables a much finer tuning and isn’t biased towards viewing figures. One such system I’ve been using is Jinni which claims to be just that. I have to say it’s pretty good. But there’s still the problem that adding a film to the database requires a human to classify or code the movie into all its genes. A solution would be to allow producers to add metadata for their films that did this codifying or to automate it – using an AI program to scan through the film and code it (probably a long way off).

If we can’t rely on technology alone, we’re left with friends (social networks) and “curators”. Curators are respected sources/reviewers: you find a film critic that likes the films you like and then trust their judgment. The problem here is that with more and more films to review, getting a well-known or respected film critic to review your film is very tough and usually impossible without a theatrical release. It’s slightly easier in the horror genre I’ve found to get reviewed because many reviewers are fans of the genre and like to find and discover new material. But even with a review, when someone is sat in their armchair ready to watch a movie, will they really go searching for reviews? No. They’ll find the film first and then check the reviews.

So what’s the bottom line for indie producers?

Don’t believe the bullshit about the long tail and how great it is for indies. With all the DVD stores closing or tending towards selling only the top tens, it’s harder than ever to get discovered.

So, if you’re going to spend advertising dollars, do it to get into a top ten. The higher you can get placed, the more likely you are to get recommended – the system is stacked in favour of the best selling titles and you need to be there. Although there are many sites that will create your DVD for you, few people will buy direct from your site - most prefering to buy for Amazon and iTunes. They're trusted and buyers already have their purchasing details stored there (credit card, shipping address etc.). So that’s where you need to be and you need to work on your sales ranking.

Social media (twitter, widgets, youtube etc.) can definitely help but social media can only work for you if there’s something social about your media. Widgets and trailers are only going to be shared and embedded if it enhances the embedder’s/sharer’s credibility – there’s got to be kudos in sharing your widget.

Don’t be a guilty pleasure and don’t be lame. Recognize that owning or recommending your movie says something about the buyer/recommender – and that “something” has to have value in that person’s social network (cool, funny, concerned). A “guilty pleasure” is a film that someone likes but is afraid to come out and say it because they know (or think they know) what popular opinion is! Helen and I love the film Armageddon but I rarely mention it in my social circles ;) You’ve got to be the film people are afraid to recommend.

Get reviewed. I’ve sent out DVD screeners to most people who ask and I’ve been active in finding cool sites that look like the sort of place our audience hangs out. The reviews help with the film’s discovery but they also give a potential buyer clues about whether they’ll like the movie or not.

Make it easy to be recommended. Provide easy URLs for your film so that if someone does want to recommend your film, it’s easy for them.

Comments

[Zen Films]

Check out this research from PRS in the UK about P2P music downloads.

http://www.paidcontent.org/entry/419-music-research-legal-hits-just-as-dominant-on-p2p/