写在开头：
人工智能在营销领域的运用绝对是未来的趋势所在，但是具体会如何发挥在该领域的影响。其实我和这篇文章的作者观点是一致的，下面一起来看看文章到底都说了什么吧。
It's clear artificial intelligence will play a role in the future of marketing, but what will that role encompass? Contributor Daniel Faggella makes a few insightful predictions.
很明显，人工智能将在未来的市场营销中扮演一个角色，但这个角色将包括什么？贡献者Daniel Faggella提出了一些有见地的预测。
While the creative process itself is unlikely to be wholly automated by machines, many agencies are aware that some elements of their work might be overhauled or redefined by machine learning tools.
尽管创意过程本身不太可能完全由机器自动完成，但许多机构都意识到他们工作的某些基础可能会被机器学习工具超越或重新定义。
In this article, we’ll discuss what kind of marketing agency work might be augmented or automated by machines (including examples of startups working on these marketing applications already) — and we’ll explore what these marketing agency AI trends and examples might lead to over the next five years.
在本文中，我们将讨论可能通过机器扩大或自动化的营销机构工作类型（包括已经在这些营销应用中开展工作的创业公司的例子） - 我们将探讨在未来五年这些营销机构人工智能趋势和示例可能会产生什么影响。
What kind of marketing agency work will artificial intelligence affect?
什么样的营销机构工作将会被人工智能所影响？
Roughly speaking, tasks that might be considered low-hanging fruit for automation are generally tasks with a clear set of inputs, a clearly defined set of tasks and a clearly defined set of outputs. These roles might be considered to have limited context. Here are a few examples:
粗略地说，可能被认为是对自动化容易实现的任务通常是具有一组明确输入，明确定义的一组任务和一组明确定义的输出的任务。这些角色可能被认为具有有限的上下文。这里有一些例子：
Pulling data from varied sources into coherent, insightful, written reports for humans to use for decision-making.
Optimizing the timing of when ads run, calibrating in real time for the lowest-priced clicks or lead conversions.
Coming up with thousands of permutations of ad copy or ad images for a campaign. It should be noted that this is only possible when initial directions and examples are provided to a machine, giving it a limited range and variety to select from in its creative variations (e.g., visually similar images, color filters, re-wording of template sentences).
将来自不同来源的数据转化为连贯、有见地的书面报告供人类用于决策。
优化广告投放时间，实时进行校准以获得最低价格的点击次数或潜在客户转化次数。
成千上万的广告文案或广告图片排列在一个广告系列中。应该注意的是，这只有在向机器提供了初始方向和示例时才是可能的，从而使其具有有限的范围和多样性以从其创意变体中进行选择（例如，视觉上相似的图像，滤色器，模板句子的重新措辞）。
The tasks above are limited to a specific, often rather small, sets of data (inputs) and involve working with and manipulating data in predictable ways. In fact, they are areas that have often already been automated to some extent with rules-based logic rather than artificial intelligence. From my more than four years interviewing experts on the impact of AI on the job market, I’ve found these are common qualities of automatable work.
上述任务仅限于特定的，通常相当小的一组数据（输入），并涉及以可预测的方式处理和操作数据。事实上，这些领域在某种程度上已经通过基于规则的逻辑而不是人工智能来实现自动化。从我四年多的专家访谈中了解人工智能对就业市场的影响，我发现这些都是自动化工作的共同特质。
Some other tasks within a marketing agency involve varied context — not simply access to a handful of data sources but wide contextual knowledge of many aspects of the business — and information that isn’t easily quantifiable in a single data source.
营销机构内部的其他一些任务涉及不同的背景 - 不仅仅是访问少数数据源，还涉及业务许多方面的广泛背景知识 - 以及单个数据源中难以量化的信息。
Examples of this varied-context work might include:
这种不同背景的工作的例子可能包括：
Knowing how to work with team members with varied personalities. It’s exceedingly difficult for machines to coordinate people and resources to reach abstract company goals — such as revenues, profit, growth, or new positioning for the company. Tasks like in-person hiring and firing, quarterly planning, running sales teams, and general management, are all generally outside the capacity of AI tools — at least in the near term.
了解如何与具有不同个性的团队成员合作。机器协调人员和资源以实现抽象公司目标非常困难 - 例如收入，利润，增长或公司的新定位。像人员招聘和解雇，季度计划，运营销售团队和一般管理等任务通常都不属于AI工具的能力 - 至少在短期内。
Sales and account management.While it might be possible to automate some business development tasks (scraping the internet for possible leads and contact info, automatically sending and customizing outreach emails and so on), it will be incredibly hard for machines to take over the processes of “feeling out a fit” with a new client and of understanding and managing their needs in order to garner their trust.
销售和账户管理。虽然有可能自动化一些业务开发任务（抓取可能的潜在客户和联系信息的互联网，自动发送和定制外联电子邮件等），但是机器接管流程将是难以置信的与新客户“感觉适合”，理解和管理他们的需求以赢得他们的信任。
Determining the positioning and strategic direction of the agency. In the decade ahead, machines may streamline many processes and allow for better decision-making by humans, but machines are unlikely to create a vision for the future of an agency. Human leaders have varied goals (finding better work-life balance, striving for work with higher profit margins, reaching a certain level of annual revenue), and the broad context involved in strategic planning (understanding your industry, your political climate, the tax ramifications of your business choices, competitors and so on) is beyond machine abilities for the time being. While machines may be able to pull in and present data from many sources, the multicontext tying together of this information into a set of coherent future plans is not going to be taken away from humans any time soon.
确定该机构的定位和战略方向。在未来的十年中，机器可以简化许多流程，并允许人类做出更好的决策，但机器不可能为机构的未来创造愿景。人类领导者有不同的目标（找到更好的工作与生活平衡，争取更高利润率的工作，达到一定的年收入水平）以及战略规划所涉及的广泛背景（了解您的行业，您的政治气候，你的商业选择所带来的税务影响，竞争对手等）暂时超出了机器的能力。尽管机器可能能够从许多来源获取数据，但将这些信息整合成一套连贯一致的未来计划的上下文能力并不会那么快的从人类手中夺走。
While the tasks above may be augmented by machines, and aspects of those tasks might be supported by automation, they’ll require talented human team members. Hiring, selling, strategic planning and other varied context work is unlikely to be wholly automated in the near future.
虽然上述任务可能会被机器增加，并且这些任务的各个方面可能会受到自动化的支持，但他们需要具备优秀的人类团队成员。招聘，销售，战略规划和其他各种背景工作不可能就在不久的将来完全自动化。
To gain perspective from real marketers in the field, I spoke with Simone Knapp, a senior account executive for RIS News, who has worked in agency and PR roles for nearly 20 years. “It’s not robots, it’s about being able to tailor a proposal to a client that’s meaningful and customized,” she said. “Can I know what they care about and automate the ideal promotions to reach their needs?”
为了从该领域的真正市场营销人员获得观点，我与RIS新闻的高级客户经理Simone Knapp进行了交谈，他曾在机构和公关部门工作近20年。“这不是机器人，而是关于如何为客户定制一个有意义和定制的建议，”她说。“我能否知道他们关心的是什么，并自动完成理想的促销活动以满足他们的需求？”
She described what she sees as the two sides of the AI coin for agencies and advertising publishers:
AI facing the agency or account manager (CRM, Process Reminders, Data Cleansing, Sales Enablement).
AI facing the customer (Marketing Automation, Advertising Optimization)
她描述了她认为代理机构和广告发布商的AI这个硬币的两面：
面向代理或客户经理的AI（CRM，流程提醒，数据清理，销售支持）。
面向客户的AI（营销自动化，广告优化）
It’s likely that both of these categories will involve discreet AI tasks of limited context and that these tasks will be among the first to which agencies will apply AI technologies.
很可能这两个类别都将涉及有限背景下的谨慎AI任务，并且这些任务将成为机构将首先应用AI技术的任务之一。
What these trends might imply for the future of marketing agency work
这些趋势可能暗示营销机构工作怎样的未来呢？
We predict that — as with other industries — marketing agencies will gradually change the way they work in the coming half-decade. Increasingly, new agencies will grow as efficiently as possible, using technology (instead of people) wherever possible. Larger firms will follow suit. We don’t expect to see massive layoffs in large marketing agencies in the next few years, but we expect growth to require a smaller human staff, resulting in less need to hire.
我们预测，与其他行业一样，营销机构将在未来的五年中逐渐改变他们的工作方式。越来越多的新机构将尽可能高效地发展，尽可能使用技术（而非人员）。大公司也会效仿。我们预计未来几年内大型营销机构不会出现大规模裁员的情况，但我们预计未来对新增人员的需求会变小，从而减少雇用。
However, it should be noted that many current marketing agency AI applications are in their infancy. Report generation, ad copy creation, and to some degree, event programmatic advertising, are in no way mainstream — and, in part, this has to do with their current return on investment (ROI).
但是，应该指出的是，许多现有的营销机构AI应用程序还处于初级阶段。报告生成，广告文案创建以及某种程度上的事件程序化广告绝不是主流 - 部分原因在于它与当前的投资回报率（ROI）有关。
At some point over a decade ago, most creative agencies began drafting and presenting designs (for logos, posters and so on) on computer programs like Adobe Illustrator, and not on plain paper or graph paper. Why did this shift happen? Because at some point the computer design programs became so easy to use and so capable in their features that it became clear: “Doing this the old way will mean losing time and money.”
在十多年前的某个时候，大多数创意机构开始在Adobe Illustrator等计算机程序中起草和展示设计（用于徽标，海报等），而不是在普通纸或方格纸上。为什么会发生这种转变？因为在某些时候，计算机设计程序变得如此简单易用，并且功能如此强大以至于变得清晰：“以旧的方式做这将意味着浪费时间和金钱。”
Artificial intelligence is not immune to these same phases of technology adoption, and, at present, there are very few AI use-cases for marketing agencies which are so fleshed out, so capable and so easy to use that any rational agency must adopt them. Many AI tools require a good deal of bespoke integration, while other tools still require a robust background in AI and/or data science in order to actually derive value from the software.
人工智能对技术采用的这些相同阶段无法避免，而且目前市场营销机构的AI使用案例非常少，这些机构如此充实，功能强大且易于使用，因此任何理性机构都必须采用它们。许多AI工具需要大量的定制集成，而其他工具仍然需要AI和/或数据科学方面的强大背景，才能真正从软件中获取价值。
An AI adoption timeline
AI采用时间表
Agencies who want a finger on the pulse of how AI might impact their industry should be looking to see what applications large agencies and large brands are beginning to adopt en masse. Simone put it to us this way:
那些想要了解AI如何影响其行业的机构应该关注大型机构和大品牌开始采用哪种应用。Simone这样说：
Some of this is may be a year or so out. There were some beta products I’ve seen that seem interesting, and I’m sure there are kinks to work out — but my guess is that this is coming sooner than most people think.
其中一些可能会一年左右。有一些我看到的测试版产品看起来很有趣，我相信有一些可以解决的问题 - 但我的猜测是，这个产品比大多数人想象的要快。
As with any industry that’s largely dependent on human resources, large agencies will have to ensure that their PR efforts around AI involve making work better for existing employees, not laying them off. Even if large firms eventually realized that layoffs would be inevitable (which doesn’t seem to be the case given the state of AI now but might be in the next five years), they would have no reason to convey this reality to their droves of nervous employees, who need reassurance that all is well (even if at some point all is not well).
与任何主要依赖人力资源的行业一样，大型机构必须确保他们在人工智能方面的公共努力包括为现有员工提供更好的工作，而不是让他们脱离现实。即使大公司最终意识到裁员是不可避免的（鉴于现在AI的状态，但可能在未来五年似乎不是这种情况），他们没有理由将这一现实转达给他们紧张的员工，他们需要保证一切安好（即使在某些时候一切都不好）。
Predictably, we see representatives of large firms reinforcing the job security of their staff.
可以预见的是，我们看到大公司的代表加强了员工的工作安全。
“If automation matures, we will embrace it — we will not use it to replace our workforce,” Dave Meeker, VP of Isobar’s US operations, told Digiday.
“如果自动化成熟，我们会接受它 - 我们不会用它来取代我们的员工队伍，”Isobar美国业务副总裁Dave Meeker告诉Digiday。
As Simone aptly stated, there do seem to be kinks to work out, and it’s not possible to guess exactly which artificial intelligence applications will become the norm for marketing agencies.
正如Simone恰当地指出的那样，看起来似乎有什么问题需要解决，而且无法准确猜测哪些人工智能应用程序将成为营销机构的常态。
We hold that data collection, data processing, and optimization problems (like those listed at the beginning of this article) are the most likely AI applications to first gain widespread use in marketing agencies. It may be another two or three years before these tools make their way to small and mid-sized agencies.
我们认为，数据收集，数据处理和优化问题（如本文开头所列的那些问题）是最可能首先在营销机构中广泛使用的AI应用程序。这些工具可能还需要两三年时间才能送往中小型机构。
说明：
原文作者
Daniel Faggella
Daniel Faggella is an email marketing and marketing automation expert with a focus on the intersection of marketing and artificial intelligence.He runs TechEmergence, a San Francisco-based market research and media platform for artificial intelligence and machine learning applications in business.
网友评论