Last week, Einstein went to work for Salesforce.
The company enlisted the name of the smartest person ever to brand its new platform-wide layer of artificial intelligence.
While computer intelligence is not foreign to parts of Salesforce’s clouds, this massive commitment means that advanced computer intelligence will now be embedded throughout the company’s tools.
And the company has made a major commitment to pushing the envelope. It is setting up an R&D lab, and it recently purchased pioneer BeyondCore, a data analysis platform so smart that it tells you the best questions to ask before answering them.
Although Salesforce’s announcement represents the biggest commitment thus far among marketing/sales tools, it is only part of the wave. There may be running disputes as to what processes actually deserve the name “artificial intelligence,” but it’s clear that most marketing and sales tools offer or will offer one or more of these capabilities:
- predictive lead scoring to target the most likely customers;
- predictive marketing to choose the best marketing approach and best time for the approach, for each customer;
- machine vision to scan images and video for brand relevance and other uses;
- natural language processing;
- cognitive analysis for understanding what social posts and other text actually mean; and
- a wide variety of other decision-making steps and insights.
Already, Persado writes its own subject lines and email text, Sizmek’s dynamic creative optimization platform assembles the right ad on the fly, IBM’s Watson is pulling shots for movie trailers and Adgorithms asserts that it is the “world’s first self-driven marketing and ad platform.”
The future is already here; it just needs to be filled in. The question isn’t whether marketing tools and platforms will become more and more self-administering. The question is what is left for marketers to do.
“Today,” retail intelligence platform CEO Kerry Liu Rubikloud pointed out, “a lot of [a markerer’s] role is to manage if/then.” These if/then rules have propelled the wave of marketing tools, determining, for instance, that if someone abandons the site’s shopping cart containing this product, then show them this ad for a sale on that product soon, on some other site.
If/then is being replaced by computer intelligence that generates a model based on past behaviors and results, and then applies it. In the past, for example, a significant number of visitors to this retail site who searched on the site for shoes made a purchase after seeing the daily sale info, so now automatically show searchers for shoes the daily sale. The system’s model replaces the need for the marketer to continually define and implement if/then rules.
Marketers, actually, should be relieved that the platforms are taking over if/then, simply because maintaining those many conditional statements — after analyzing the data to set the best if/then rules — is a losing proposition when you’re dealing with millions of customers who are increasingly expecting personal experiences.
As computer intelligence takes over tactical decision-making, and as marketing trends toward “campaigns of one” targeted at each individual customer, the marketer increasingly assumes the role of goal setter. In other words, in a world of self-driving cars, someone has to tell the car where to go.
“There will always be a need for human intervention,” Liu said.
For instance, he said, assume you want to bring more traffic to your website. An intelligent platform could determine that the most effective move would be ads on selected websites that are frequented by females under the age of 35. It could also determine where those ads should be placed, what they should say or show, and what prices you should pay, based on previous best results.
But even in this scenario, a human needs to decide that no more than $10,000 will be spent on this effort and that the goal is to bring in more traffic.
Persado SVP for Global Customer Success Ryan Deutsch compared a marketer’s irreducible role to that of a pilot in a plane that frequently flies on autopilot.
His company employs machine intelligence to generate emails and other marketing text from scratch, according to guidelines and previous successes. But even so, he noted that the marketer’s role will remain as the autopilot’s overseer.
For example, Deutsch said, Persado’s platform might determine that the most effective subject line in an email campaign for Labor Day weekend shouldn’t mention either Labor Day or the special 30-percent-off holiday sale. But the CEO of the company undertaking the email campaign may understandably want to know why.
The email campaign is on a kind of autopilot, but the marketer has to assure everyone along for the ride that the autopilot knows what he’s doing.
The marketer would explain, for instance, that the data shows a subject emphasizing exclusivity (“just for you”) is more effective than one about Labor Day or the sale.
Living in the material world
This points to another role for the digital marketer in this post-Einsteinian environment: negotiating with the real world.
At least until IBM’s Watson and his brethren can walk down the street and operate as a human would in a meeting, the marketer is the liaison with reality.
The intelligent marketing platform may have determined that, based on past results, “customer exclusivity” is a winning theme for digital marketing. But, as Liu pointed out, a human needs to design and throw the in-store launch party that celebrates the “just for you” theme.
Another facet of the real world is its unpredictability, and marketers will also continue to be crucial — at least for the foreseeable future — in helping intelligent systems handle the occasional “black swan” event. These out-of-the-ordinary occasions, like a news event that makes a particular product undesirable or a particular campaign unappealing, will still have to be managed by human marketers who are overseeing the autopilot. At least for now.
As for the marketers’ assets of imagination and empathy, it remains to be seen if these are outside the realm of computer intelligence.
Imaginative renderings that fit a pattern are the first to fall. Persado and other platforms are setting the standard for writing marketing copy that has proven to be effective, for instance, but the history of advertising is written by creative approaches that broke the mold.
However, it would seem that, if systems like Persado and Watson can create models of what has worked before, they can create counter-models to break the mold — at some point. For the time being, inspiration remains in marketers’ corner. “We aren’t anywhere close to AI [having intuition],” Liu claims.
And empathy is similarly a dicey proposition. Empathy could help platforms determine, for instance, when the “creepy” line is crossed, separating targeted marketing from over-the-top marketing, because one human has a sense when another human’s invisible boundaries have been crossed. While models can show diminishing returns from certain intrusive marketing, marketers are the best ones to know when non-quantitative limits have been crossed. Again: at least for now.
But the ultimate protection for the marketer is probably the fact that the makers of these intelligent systems don’t want to put the people who buy them — the marketers — out of a job. Because of this fact, it’s likely that, even as the marketing platforms become smarter and more self-reliant, vendors will continue to design marketers into the workflow.
As Persado’s Deustch told me, “I’m a marketer.”
“I didn’t join Persado to get rid of marketers.”