Google says it can predict movie box office with 94% accuracy What makes a successful movie? Is it a good plot? Strong characters? Popular leading actors? Impressive special effects? Not necessarily, says Google – and moreover, the online giant says it's figured out how to predict a film's box office performance with up to 94 per cent accuracy. According to a Google whitepaper published …
I think you are absolutely right. Marketing did it's job and now I know about a movie. Next I am doing a search for reviews. I read them then I decide if I want to see it or not. I think that this explains the big gaps in the graph shown in the article.
By the time the searches are rolling in it's too late. The movie will either be good or not, it's not as if the studios can take that info and tweak the movie last minute or something (nor would I think they have any idea what to tweak even if they did have the ability). It doesn't help with making new movies either since they already have the metric that shows what makes a successful movie which is the sales.
I suppose it may help if you are putting bets on stocks or something. But don't see how it helps the studios make more blockbuster movies.
It matters because success != good. Success for a studio is making gobs of money. It doesn't matter if the movie is any good, only that enough punters are willing to give it a try.
What Google have shown is that if you can control the buzz effectively then you can make money even if your movie is a stinker. It tells studio's where to put their post-production efforts.
I think it is aimed at people who would pay for this info such as cinemas who would like to gauge demand and therefore assign resources, or mad directors / actors who have the money to pay google to work out their ego impatiently.
Then again it just sounds like Google boosting their own ego, next they might work out that the search levels for garden chairs went up around the May bank holidays, crazy I know.
Not true. A lot of marketing dollars and strategies are being decided then. You can use the advance info to decide which of the movies in your stable for the Christmas or Summer period is worth the extra advertising dollars. Its great news for Google as it will pitch this along with its AdWords business to help persuade movie studios that they should advertise more on Google.
What of course this doesn't address whatsoever is whether the movies are worth seeing. When I see the returns for franchises such as Transformers of the Caribbean or The Fast and the Curious I can predict with 94% accuracy that these will suck. TV has come on nicely, but feature films .....
By the time the searches are rolling in it's too late. The movie will either be good or not, it's not as if the studios can take that info and tweak the movie last minute or something (nor would I think they have any idea what to tweak even if they did have the ability).
Others have already pointed out the fallacy in the "too late" argument, since marketing for the film can still be underway. There's also the small matter of rolling out to more theaters, delayed international releases, etc.
Beyond that, with digital projection and distribution, studios certainly could make changes to a film after it opens to general distribution. Some films already have been re-edited between an initial limited distribution and wider roll-out; it's probably just a matter of time until studios start regularly re-editing films after the opening weekend. The distribution infrastructure may not be up to it yet, but there's no technical obstacle, just economic ones - and if the studios think it's worth doing, they'll try it.
A/B testing is another possibility. Release two versions of a film on opening weekend, then expand into more theaters using the more-popular version (or distribute both versions to particular targets based on projected interest).
Yes, I was noticing the same thing. It doesn't seem predictive, but more reactive.
I also noticed that the peaks don't correlate, meaning some high search peaks are lower than the box office peaks.
Me thinks this is just Google drumming up business from the notoriously clueless film studios to take their marketing money. If you watch Soderbergh lecture on the "State of Cinema", they have NO idea why or why not various movies are successful or tank...(and what's worse, the studios don't even seem interested in any self-analysis or A/B testing -- given the "big data" that is available).
http://www.deadline.com/2013/04/steven-soderbergh-state-of-cinema-address/
Google is good at vacuuming that ad money from the clueless & insecure.
If I am typical the measure is indeed reactive rather than predictive. I typically do three or four searches to check the showings (times and locations) before going to see a film, having already decided what to see and when. In making that decision I may indeed look at the trailers, but rarely suficiently in advance for anyone to make a meaningful "decison" based on collating the search patterns of "millions like me". If movie-makers and distributers were to want infomration of predictive value it would be more sensible to give me a discount (or other benefit) in returning for responding to a survey on what I like to watch and where, when and how I like to do so. That said, using such data to find search spikes which do not correlate with spikes in box office takings (or vice versa) and the looking at the causes might indeed be an intersting exercise. The value of using search engine analyses should not be under-estimated - even though this example has been over-hyped.
Have to agree about the reasons my friends and I search for movies. I'd say 90% of the time we've already decided we want to see the movie, at a local cinema and whether it will be afternoon (90% because its cheaper) or evening (10% if we want to avoid the screaming kiddies or that's the only time slot in which it is showing). We're just checking to confirm 1) what time we need to be there, 2) tickets are still available. It's actually fairly rare that we have a leaner and want to check reviews or see trailers to decide if we are going to see it.
Me thinks that the Chocolate Factory is trying to reverse the trend. It wants those nice paid ads to be before the release, not after. That way it can get the bug bux.
Predicting new movie releases success is a very iffy process. It might be a little bit easier with sequels, but you never know.
In any event, predicting pent-up demand for a product has always been difficult, and will continue to be. If Goooogle could predict where the demand is going to be, their crystal ball is better than mine. Unfortunately crystal balls just don't work, so predictions aren't there!
Good luck anyway.
None of my family, friends, neighbours ever type web address into the URL box anymore. Instead all text happily goes into the search box which surprise surprise often defaults to the Big G. I know its easier to surf this way, and spelling is corrected, but people are just mindlessly feeding the G monster... I try to suggest they navigate first to the specific site i.e. IMDB.com before continuing the search, but its hopeless.... I think its especially bad when they punch in embarrassing medical issues etc. I have to remind them they're still logged into Gmail in the other window. But they don't care. What can I do? Tell them they're kissing their privacy goodbye in a thousand cuts....
Correlation between searches for information about an upcoming film and people going to the cinema to see it?
Colour me surprised - not.
Success and what I am interested in is if people wanted to see it when they walk out of the cinema.
Hollywood learning how to better sell krud with the aid of google is going to result in more fail.
I forgot to mention.
I have zero intention of ever visiting a cinema again.
I have a nice big plasma, nice sound system with a volume control, a pause button, comfy chairs, friends not strangers. Don't have to travel, pay parking or risk the car getting nicked or keyed. It's cheap and I can even have a fag. I really want to forgo all that because Hollywood thinks I ought to pay extra to see something once a few months earlier than otherwise?
Cinemas like the top whatever music charts are just an obsolete relic of the media industry. I don't understand why they haven't curled up and died.
I'm with you.
There are a few that have been condemned as stinkers which I actually like[1]. Likewise there are several that have been widely acclaimed, yet which I reckon are turds.
[1] e.g. I really liked Disney's "John Carter", despite the universal howls of derision at that from all quarters.
And that's a problem which will become more important. People act solely on correlations they have found. Correlations were the reason for austerity politics. The paper praising it saw a correlation between failing states and a high deficit... and also turned out to be based on bad data and have an error in the spreadsheet.
This example may still make a bit more sense, however the causality is different. If a movie is well known it is both more often seen and more often searched for.
March/April= spring break (get the damn kids out of the house); May= school's out but summer school's not yet started; June/July= give me aircon!; November= Thanksgiving; December= winter holiday break. Not too surprising that people with access to the 'Net would use it to do more research into what movie they were gonna see tonight anyway (agree with Don Jefe, though, that thirteen sources seems a bit overkill), right? Move it along, there's nothing to see.
'No word of mouth impact either'...
That's actually a big problem in lots of data analysis, there are massive forces that weigh on any trend that can't really be measured only estimated, but if it is estimated it blows up the rest of the analysis. I understand that you can't analyze what you can't measure, but decisions are made based on the available measurements, which leads to some very poor assumptions.
My guess is that 80-90% of the time those estimates are accurate. That's enough for general use predictions on the bottom line. You'll get outliers, some of which I expect would be obvious if they had someone in the pre-screening room who actually enjoyed watching movies, but they typically won't adversely effect your bottom line.
I'd actually guess the more problematic part for the bottom line is timing your releases. If you put your #1 expected revenue maker up against an opponent's #1 money maker, you're probably both going to lose money. So you have to try to time it so your #1 is against their #2 or lower movie. Admittedly, this is a problem of their own making because if the movie isn't raking in money on week 1, distribution is gone by week 3.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
