“微博机器人”来了

    微博界对美国联合航空公司（United Airlines）并非全然是友善的。2011年，在美国九家主要航空公司中，美联航在社交媒体上的口碑是最差的，Twitter上关于美联航的负面微博差不多有5.5万条。祸不单行，今年春天美联航与大陆航空（Continental）的订票系统合并后故障不断，许多打电话订不到票的顾客转而在Twitter上发牢骚，短短几小时内，他们就发出了几百条愤怒声讨美联航的微博。

    面临来自社交网络的巨大压力时，无论是一家航空公司也好，或是任何其他公司也罢，他们应该如何应对？如今社交媒体已不仅仅是让企业与顾客直接联系的一种新鲜的方式，它往往也会消费大量的营销和客服资源。社交网站舆论的数量如此庞大，足以令人抓狂。戴尔公司（Dell）的社交媒体倾听控制中心（Social Media Listening Command Center）就是一支专门负责倾听社交媒体声音的团队。它由70余名专家组成，每天大约要追踪25,000多条与戴尔公司有关的微博。

    每天光是要阅读这么多信息就是一件不切实际的事，更不用说逐条回复了。因此越来越多的公司正在采取一种对策，应付日益繁重的工作量。对于某些从事这一领域的人来说，这种对策不啻于对社交网络的一种亵渎——可以说，这些公司正在抽离社交媒体中的“社交”部分。

    这些公司正在通过一些自动化工具来捕捉和分析社交网络上的谈话，优先筛选出那些值得关注的消息。比如某些自动规划应用和内容优化应用可以通过追踪大多数粉丝以及热门话题来评估企业该在什么时候发布关于哪些内容的微博，有些公司甚至在Twitter和Facebook上部署了虚拟代理人和先进的“微博机器人”，自动回复消费者的咨询和评论。

    咨询公司Altimeter集团的社交媒体分析师耶利米•欧阳指出：“这个领域仍处在初期阶段，但是正在成长。它的优点很明显：各大品牌必须能够扩大对社交网络的研究范围。有些企业使用社交网络的顾客人数正在增长，这些企业也不想忽视自己的顾客。”

    不过，这种做法的缺点也很明显。社交媒体之所以如此重要，正是由于社交媒体依托于真实的人际互动。社交网络出现之前，消费者和企业的社区经理往往都是通过传统的“联系表格”和“自动求助栏”等工具进行互动。Facebook和Twitter出现之后，至少在一开始的时候，它们令消费者和企业可以在一个公共论坛中进行沟通，而且往往可以获得几乎是即时的沟通结果。不过耶利米•欧阳近日在科技博客Mashable的一篇文章中警告称，一旦这个沟通渠道失去了真实性，它们可能会褪变成“另一个过度品牌化的企业媒体”，失去最初的吸引力。

    因此，摆在我们面前的问题变成了：有没有办法能在不失去真人沟通的前提下实现“微博自动化”？

    或许有。有一个领域的自动化技术已经取得了显著进展，这个领域就是信息收集。所谓“自动监听技术”已经诞生好几年了。它可以在社交网络上自动监听提及某一特定内容的微博或某些特定关键词，甚至可以追踪发出消息的地理位置。广告公司经常采用这种技术在社交网络上收集提及客户的贴子或微博，然后根据重要性对它们进行分类。这种新一代的信息分类技术使企业得以过滤社交媒体上的杂音，只回复最紧急的消息。

    自动规划技术同样也已日臻成熟。许多企业都具备合理规划微博和网贴并及时更新的能力，人们也并不是不知道它所带来的好处——据营销平台Hubspot公司报道，合理规划了微博等社交媒体的企业，销售机会可以增长到原来的三倍。随着新型智能规划应用的出现，企业在规划微博的过程中可以减少很多臆测的成分。这些自动规划应用采用了先进的电脑算法，可以评估粉丝们大概会在什么时候上线，甚至可以分析他们在聊些什么，然后再即时发送微博或消息，以便实现最佳的消费者接受效果。（其中一家使用了微博自动规划技术的社交媒体系统就是我的HootSuite公司。）耶利米•欧阳在科技博客TechCrunch的一篇文章中写道，这种“精确狙击”的方法可以“根据大家讨论的话题来选择发送内容的时机，在正确的时间把内容发送给正确的人。”

    这种自动优化应用受到了越来越多的“付费社交媒体”——也就是营销微博的欢迎。由于这种品牌化的微博或贴子可以即时显示在我们的Twitter和Facebook的消息更新里，因此不少企业都愿意为它们支付高价。2011年，各大企业花在社交网络广告上的总支出约为52亿美元，其中一部分资金就流向了这个领域。因此，如果有一种应用能够提前自动预测消费者对营销微博的接受情况，它的价值是显而易见的。就像科技博客Mashable所说的那样：“各大品牌将开始优化发布的内容，对此内容进行分析，并将所获得的内容转变成社交广告，通过Facebook的‘付费赞助内容’发送出去。”它有能力把经过优化的微博和Facebook上的贴子直接变成广告，因此有望重塑社交广告的形态，并将成为未来几个月里的大新闻。

    此外还有一些工具极具科幻小说的意味。最新一代的“社交机器人”和“虚拟微博代理人”如今正在Twitter和Facebook上出现。这些工具不仅可以优化内容，甚至还可以自行编撰内容。最简单的Twitter机器人只不过是一组计算机脚本，可以自动追踪使用了某些特定关键词的用户。而最新的社交机器人则迈上了一个新的台阶——它们不仅可以自行参与具有说服力的会话，甚至还经常会吸引大量的粉丝。比如谷歌（Google）的产品经理格雷格•玛拉因为构建了一个名叫@trackgirl的微博机器人，用户们都不知道@trackgirl是个机器程序，这个“女”机器人发布了一条“脚踝扭伤了”的消息，居然就有粉丝给她留言，表示同情。

    The Twitterverse hasn't been entirely kind to United Airlines (UAL). In 2011, the company enjoyed the dubious distinction of ranking worst in social media sentiment among nine major airlines with some 55,000 negative tweets. Things only got worse this spring when United fumbled its merger with Continental's reservation system. Customers stuck in mammoth phone queues turned to Twitter to air their frustrations, registering hundreds of #onhold complaints in a matter of hours.

    In the face of this kind of collective social barrage, what's an airline -- or any company, for that matter -- to do? Social media has gone from a novel way for businesses to connect directly with customers to an often sizable drain on marketing and customer service resources. The sheer volume of social conversations can be mind-numbing. Dell's (DELL) Social Media Listening Command Center -- a crack team of more than 70 specialists headed by its own listening czar -- tracks in the range of 25,000 posts a day focused on the brand.

    Reading, let alone responding, to that many messages isn't always practical. So to keep up with the burgeoning workload, a growing number of companies are taking a step which to some in the field amounts to purest blasphemy: They're taking the social out of social media.

    These companies are turning to automated tools that capture and analyze conversations on social networks, prioritizing which messages really merit attention. They're using auto-schedule and content optimizing apps to assess when and what to post, with an eye toward reaching the most followers while piggybacking on trending topics. In some cases, they're even deploying virtual agents and sophisticated bots in Facebook (FB) and Twitter streams to automatically respond to consumers' queries and comments.

    "The space is nascent but growing," notes social media analyst Jeremiah Owyang of Altimeter Group. "The upsides are obvious: Brands must be able to scale and those that are experiencing an increase in customers using social channels don't want to leave customers behind."

    The downsides, however, are also obvious. Social media's cachet is contingent on real human interaction. In the beginning, at least, Facebook and Twitter enabled consumers and community managers to knife through contact forms and automated help lines, connecting with real people in a public forum and often getting near-immediate results. When authenticity leaves the room, Owyang warns in a recent Mashable article, these channels may become "yet another over-branded corporate medium," losing their primary appeal.

    The question becomes, then, are there ways to automate without losing the human touch?

    Maybe. One area where automation has already made significant inroads is information gathering. Automated listening technology to monitor mentions, specific keywords and even geographic origin of messages on social platforms has in fact been around for years, used by ad agencies to track references to clients and in similar roles. What's new and interesting in this space, however, are tools that intelligently screen these conversations and categorize them based on importance. This next-generation triage technology lets companies cut through the noise on social channels and respond to only the most urgent messages.

    Scheduling is likewise ripe for automation. The ability to schedule tweets, posts and updates in advance is hardly novel (Nor are benefits unknown: Companies that schedule yield three times as many leads, according to marketing platform Hubspot). New smart scheduling apps, however, aspire to take some of the guesswork out of the process. Using algorithms that assess when followers are likely to be online and even what they're talking about, tools send out timed messages for optimum reception (Disclosure: One of the social media systems using this auto-scheduling technology is my company, HootSuite). This sniper-like approach promises to "match what's being said and time content to publish at the right time to the right people," writes analyst Owyang in TechCrunch.

    Importantly, this kind of auto-optimized content lends itself well to the emerging world of paid social media, i.e. "promoted" stories and tweets. Companies pay dearly for these branded updates injected right into our Twitter and Facebook streams -- part of an estimated $5.2 billion spent on social advertising in 2011 -- so a tool to automatically forecast reception in advance has obvious value. As explained on Mashable, "Brands will start with content publishing optimization, analyze this content and re-purpose earned content into social ads ala Facebook's sponsored stories." This potential to turn optimized tweets and Facebook posts directly into ads promises to remake social advertising and will be significant news in the months ahead.

    Then there's the sci-fi stuff: the latest generation of social bots and virtual agents showing up on Twitter and Facebook. These tools don't just optimize content; they make it. In their simplest incarnation, Twitter bots are computerized scripts that automatically follow users who use certain keywords. The newest social bots, however, take this to another level -- participating in convincing conversations and often drawing legions of followers. Google (GOOG) product manager Greg Marra earned notoriety building @trackgirl, who infiltrated the ranks of hardcore runners, even attracting sympathy messages when she "hurt her ankle."