Not Every Task Can Be Automated

Jan 10, 2013


      Bookmark and Share

The explosion of digital media launched after the first internet bubble burst ten years ago has given rise to a vast and intricate landscape of technologies each promising to simplify the increasingly complex and fragmented tasks involved in digital marketing. The number of companies in this space is staggering. LUMA Partners has famously mapped technology vendor markets for display, mobile, search, social, and video. Forbes contributor Feinleib recently traced a map of the Big Data market.

All of these vendors share an identical core value proposition - to automate and thereby expedite tasks that would normally require significantly greater investments of time and manpower were they to be executed manually. Most of these vendors are able to deliver on this proposition through innovative deployments of new and existing technology. The best vendors in this space have enabled marketers to devote more of their efforts to creating better and more effective programs for their organizations.

However recently I had an experience with a company that gave me pause. The vendor proposed a solution that would automate core social media tasks. With the relentless pace of this automation craze I am now concerned that we may be trying to automate too much.

The vendor approached me about six months ago with a solution aimed at automating digital influencer outreach. The user simply uploaded content to the platform through a simple interface and then selected the users to whom the content was to be sent. The digital influencer would receive an email that allowed him or her to login to a separate dedicated user interface where he or she could either elect whether to use the content using an embed code or hypertext link. Where the digital influencer chose to use the content, placing the corresponding embed code or hypertext link on his or her asset(s), the platform could track the number of impressions and interactions generated by the placement. The platform also allowed marketers to provide incentives (e.g., coupons and so forth) to digital influencers.

What I found troubling was that the solution sought to automate relationships. The problem with this is that relationship building is, in neuroscience terms, a "top-down" process. This means that it requires a complex combination of attention, expectation, learning, and memory, which no machine has ever been able to successfully replicate.

For example, what if none of the digital influencers whose contact information I've uploaded into the platform uses my content? How would I know why they opted not to use my content? Should a machine be trusted to make such a subtle determination? Ultimately I would have to return to the manual relationship building process in order to successfully resume communication. Moreover many of the digital influencers might find my sending them automated transmittals in the first place to be somewhat remiss.

There are of course ways in which technology can be used to streamline digital influencer outreach such as maintaining annotated databases of digital influencers and maintaining a journal that records results of outreach efforts (a Salesforce.com for digital influencer outreach if you will). However the core tasks of relationship building, as well similar tasks that require a human cerebral cortex that can bring past experience to bear on present and anticipated events, should be delegated to human beings.