TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Facebook Overestimated Key Video Metric for Two Years

283 点作者 tshtf超过 8 年前

27 条评论

coldtea超过 8 年前
&gt;<i>the tech giant vastly overestimated average viewing time for video ads on its platform for two years, according to people familiar with the situation</i><p>They didn&#x27;t &quot;overestimate&quot;. They plainly gave false numbers to drive ad sales.<p>Coldtea&#x27;s law: Never attribute to incompetence what can be explained by profit.
评论 #12565772 未加载
评论 #12565308 未加载
评论 #12565874 未加载
评论 #12563461 未加载
评论 #12563474 未加载
评论 #12565559 未加载
评论 #12567298 未加载
sean_patel超过 8 年前
&quot;How Facebook is Stealing Billions of Views&quot; - Kurzgesagt – In a Nutshell exposed this with detailed analysis and evidence. Watch it here =&gt; <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=t7tA3NNKF0Q" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=t7tA3NNKF0Q</a><p>This is neither a bug nor a &quot;mistake&quot;. It&#x27;s straight up fraud. Fraud because Facebook uses these false #s to prove to their advertizing clients that they are getting a HUGE ROI, when they really aren&#x27;t. Also the investors and analysts bump up their ratings and the stock goes higher and higher on false ad view #s.<p>By counting a view as a legitmate view after the video plays only for 3 seconds, their &quot;algorithm&quot; counted billions of views even when the user has not even seen it, because Auto-play is enabled by default and you have to opt-out &#x2F; disable it and no-one does it. It takes a person roughly 4 seconds to scroll off their feed as they quickly &quot;scan&quot; their friends posts.
评论 #12564717 未加载
评论 #12563723 未加载
评论 #12562871 未加载
codehusker超过 8 年前
On my programmer hand, I can see myself making a similar technical choice. We have auto-playing videos, we shouldn&#x27;t count the views less than 3 seconds because that wasn&#x27;t really an intentional view. But that means we should also exclude them from the overall view count. I don&#x27;t know if FB did that as well.<p>On my shareholder hand, this seems slightly like fraud.<p>On my advertiser foot, this seems slightly like fraud.
评论 #12562125 未加载
评论 #12561743 未加载
评论 #12561653 未加载
评论 #12562151 未加载
emcq超过 8 年前
This seems slightly overblown. I suspect this was a metric useful for engineering that found its way to external usage. That&#x27;s because in many recommendation systems in practice you want to filter out the spurious views, and a simple way to do that is with dwell time.<p>It&#x27;s not uncommon to require at least 50% of pixels in view for 1 second before you have an impression for static images [0, 1]. AOL defines an impression requiring 2 seconds for images [1]. Facebook likely did some analyses to find that 3 seconds was a good cutoff for their site.<p>There are more sophisticated ways to estimate dwell time but they seem uncommon in practice; perhaps due to their difficulty communicating to advertisers what the impression metric actually means.<p>For sites with many bots or inbound marketing you often find users bounce quickly which drives some of this timing. I&#x27;m a bit surprised it needs to be that high for Facebook without many bots or users bouncing quickly. Perhaps this is for mobile users scrolling quickly.<p>[0] <a href="http:&#x2F;&#x2F;advertising.aol.com&#x2F;specs&#x2F;terms&#x2F;aol-viewability-terms" rel="nofollow">http:&#x2F;&#x2F;advertising.aol.com&#x2F;specs&#x2F;terms&#x2F;aol-viewability-terms</a> [1] <a href="http:&#x2F;&#x2F;mediaratingcouncil.org&#x2F;063014%20Viewable%20Ad%20Impression%20Guideline_Final.pdf" rel="nofollow">http:&#x2F;&#x2F;mediaratingcouncil.org&#x2F;063014%20Viewable%20Ad%20Impre...</a>
评论 #12562035 未加载
评论 #12561862 未加载
评论 #12561842 未加载
评论 #12563106 未加载
评论 #12562547 未加载
评论 #12562887 未加载
bluetwo超过 8 年前
&quot;likely overestimated average time spent watching videos by between 60% and 80%&quot;<p>Soooooooo... fraud? They say it didn&#x27;t impact billings&#x2F;revenue, but I have to imagine they will now be in a position to give discounts, if not refunds.
评论 #12561952 未加载
jtchang超过 8 年前
I don&#x27;t think this is overblown at all.<p>I expect to see a class action lawsuit because of this. If you are a big advertiser and you can somehow show damages because of this error there may be a case.<p>There are some adtech companies recently who focus purely on reporting the metrics of your ad purchases. Kind of like acting as a third party. I forgot exactly what they are called.
评论 #12563530 未加载
pducks32超过 8 年前
I don&#x27;t believe for a second that this was unintentional. The amount of press and mind share that Facebook has received for their &quot;rocketing video efforts that rival YouTube&quot; was totally worth pissing off some ad exec. Very few people will see this news compared to &quot;Facebook dominates video&quot; headlines that have been everywhere. This was a smart play on their part.
评论 #12561877 未加载
nostrademons超过 8 年前
My data-weenie hat tells me that this is a silly metric anyway. As an advertiser, you really care about the <i>distribution</i> of viewing times, and in particular what fraction of viewers watch the video all or most of the way through. Arithmetic mean is virtually useless when the data has a power-law or other non-linear distribution, and it&#x27;s highly likely video viewing times exhibits this.
评论 #12561873 未加载
评论 #12561838 未加载
laurihy超过 8 年前
Then again, if a &quot;video view&quot; is defined as &quot;watched over 3 seconds (50% of the video visible in the screen, IIRC)&quot;, then it sort of makes sense that &quot;Average Duration of Video Viewed&quot; doesn&#x27;t include non-views.<p>For sure Facebook should attempt to name their metrics as descriptively as possible, but also advertisers should make sure they understand how different conversions are measured, what&#x27;s included and what&#x27;s not. Another example would be &quot;Clicks&quot; metric, which included all engagement (i.e. likes, shares etc), instead of just &quot;Link clicks&quot;.
评论 #12564315 未加载
评论 #12563719 未加载
mikejb超过 8 年前
I didn&#x27;t fully understand what the bug was, but I found on a German news page[1] with a slightly more detailed description:<p><i>Aufgezeichnet wurde zwar die Gesamtsehdauer aller User, geteilt wurde dieser Wert allerdings nur durch diejenigen User, die das Video länger als drei Sekunden ansahen</i><p>So the average was not calculated correctly: They accumulated the duration of all video views (including of those shorter than 3 seconds), but divided it by the number of &#x27;legitmate&#x27; views - i.e. only those longer than 3 seconds. So you get a pretty big offset if you have many views under 3 seconds (which they probably do, thanks to autoplay).<p>[1] <a href="https:&#x2F;&#x2F;www.heise.de&#x2F;newsticker&#x2F;meldung&#x2F;Facebook-Unklare-Video-Metrik-fuehrt-Werbekunden-aufs-Glatteis-3330188.html" rel="nofollow">https:&#x2F;&#x2F;www.heise.de&#x2F;newsticker&#x2F;meldung&#x2F;Facebook-Unklare-Vid...</a>
cft超过 8 年前
We are using Facebook Audience Network in random order with other ad network for mobile ads. FB ads earnings are about 3-5% of other networks, with exactly the same traffic. Other networks disclose the percentage the revenue split between them and the publishers, but FB does not. What do they overestimate in this case, I wonder?
bonniemuffin超过 8 年前
I don&#x27;t think average watch time is a particularly useful metric, because no matter how they count views, there&#x27;s some kind of arbitrary cutoff. If they count a view as soon as the player finishes loading, they&#x27;ll include a whole bunch of people who didn&#x27;t intend to watch the video, and in fact only &quot;watched&quot; a few milliseconds of it before bouncing. So that will drag the metric down artificially. I bet if they count views as soon as the video starts playing, they could drag down average watch time by improving player load time, because they&#x27;d count more unintentional &quot;views&quot; of very short duration.<p>I hope&#x2F;assume this is part of a suite of metrics that they provide to advertisers, so you can understand it in the context of other things like video completion rate or counts of views that reach X% through the video.
throwanem超过 8 年前
It&#x27;s like watching a wasp land on a nettle. You know somebody&#x27;s going to get stung, but you just don&#x27;t care who.
aaron695超过 8 年前
I don&#x27;t get it, it makes sense?<p>Less than 3 seconds is not a view any more than seeing the picture of the video is a view. It doesn&#x27;t count I&#x27;d say?<p>Or do advertisers have to still pay for a 2 second view?
shostack超过 8 年前
While this is concerning, smart advertisers were looking at the distribution of view length percentages anyway as you can easily pull those columns in and their derived metrics.<p>Likewise, smart advertisers look at lift in conversion metrics when possible, in which case this stat is irrelevant.<p>That said, FB has not exactly helped things by making metric definitions a little obfuscated in general.<p>Personally, I see a bigger concern is them giving 100% view through conversion credit with a 1 day window by default as part of any website conversion action tracking. There are very few cases (like some retargeting situations) where you&#x27;d ever want to give full weighting for a VT, and while there is likely value in VT&#x27;s, I probably wouldn&#x27;t give 100% credit to them by default given the rate at which people scroll on mobile. But advertisers like to see big numbers, agencies like to show big numbers, and so you have platforms like FB aggressively try to push metrics like this and some of their rather loose definitions of &quot;engagement&quot; without great explanation of the nuances or pros and cons. These are largely left up to the advertiser to determine since, to be fair, they are very subjective.<p>Savvy buyers know this and configure their reporting and tracking settings accordingly because FB and PMDs give you those options. They are sometimes just buried.<p>Ultimately, IMHO FB and Google&#x27;s greatest defense towards any of these sorts of claims is better and (more importantly) transparent attribution data and tools. If they can prove their value on the bottom line, other things often don&#x27;t matter to many advertisers. Attribution is a tough nut to crack, but for advertisers spending large sums, it is critical to be successful in these channels.
restlessdesign超过 8 年前
It is unreasonable, even for a developer, to count 3 seconds as a play. Anyone working on a video product would know they were inflating the numbers.
评论 #12562543 未加载
Bedon292超过 8 年前
Ok, what am I not understanding here?<p>If they are only counting a view as over 3 seconds. And then only averaging the those views what is the problem? They tell me 1000 people viewed it (over 3 seconds), and then say the average view time was 10 seconds for those 1000 users. How is this an issue? Why do I care that 9,000 people scrolled past it and weren&#x27;t counted in the viewer numbers? Now if they were reporting this as 10,000 viewers and saying 10 second average, I see the problem. Is that what was happening? Or what am I missing?
评论 #12564476 未加载
smithunsero超过 8 年前
In 2012 there were reports that 80% of traffic from Facebook Ads were automatic bots. So this is not the first time Facebook use tricks to get more money from clients.
评论 #12563205 未加载
评论 #12563748 未加载
sidcool超过 8 年前
This looks like a flat out lie. Startups have to sometimes do that to woo potential investors. Although I like Facebook overall, I do not agree with many of their business practices.
breatheoften超过 8 年前
What was the actual bug? Does this mean that no video had an average view time less than 3 seconds?
评论 #12561764 未加载
bruinbread超过 8 年前
This isn&#x27;t anything new. Several YouTubers have &quot;exposed&quot; how Facebook tracks views. It&#x27;s not accurate, but it&#x27;s the system that has been in place for a while and they&#x27;ve defended it as such.
Arkaad超过 8 年前
I am disappointed by FB. Next we&#x27;ll learn that they do tax evasion.
supermatt超过 8 年前
If face book only count a view as over 3 seconds, and they measure the average view as only those videos over 3 seconds, what is the problem? Surely now its less representative?
randomgyatwork超过 8 年前
When will Facebook &#x27;content creators&#x27;(users) get compensated for the value they create?
yueq超过 8 年前
Will there be a class action on this matter?
tn13超过 8 年前
It might be inaccurate but does not matter much in advertiser perspective. Given the conversions etc. eCPMs conversion to the real expected value.
malloreon超过 8 年前
how much of their video income will they be refunding to advertisers?