Could social media and marketing automation be the perfect partnership?
Social media has revolutionised marketing. It’s revolutionised the world, actually, but marketing has experienced perhaps some of the most wide-ranging changes. Social media allows unprecedented access to consumer data and provides us with actionable insight in to their habits, interests, likes and dislikes, like never before.
But with a lot of people and a lot of data comes a lot of potential pitfalls. Things can get complicated quickly. How are companies to process all that data, glean insights from it, target marketing appropriately, and interact with consumers through social media in the manner that people have come to expect?
Enter AI. On the face of it, Artificial Intelligence ‘bots’ are simultaneously a marketer’s greatest dream and worst nightmare. They’re capable of crunching data in seconds, of extrapolating insights from that data, targeting marketing efforts precisely through specialised algorithms, monitoring and assessing reactions, and even interacting with consumers directly. All this without ever getting tired, ever needing a holiday, asking for bonuses, or even pay (unless you count the overheads required to keep them running), and without ever taking more than a few moments to complete their tasks.
Sounds fantastic! But there’s always that creeping concern that, ultimately, AI marketing could outdo us at our own game and render human social media marketers redundant. The fear of AI taking human jobs is nothing new, but it’s growing in recent years, largely in accordance with the rise of AI.
But do we really have anything to worry about? And can AI even do all that the hype claims?
AI and Marketing Automation on Social Media
There have been a few high-profile stories about AI recently. From the racist chatbot ‘Tay’, who ‘learned’ bigotry from Twitter users, to lingerie retailer Cosabella, which totally automated it’s social media marketing, replacing its digital agency entirely. Not to mention Cambridge Analytica and their use of algorithms to emphasise ‘emotions’ over ‘facts’ and turn Likes in to a political weapon. Even the New York Times has introduced a chatbot to automate some of their daily social media posting.
The Cosabella AI enabled the brand to communicate in a more direct manner with their customer base, without relying on the middle-man of a digital agency. The AI, known as ‘Albert’, was set the task to manage delivery of 24/7 social media marketing, providing constant insights into customer behaviour, managing budgets, prioritizing posts, targeting content, and crucially helping them to deliver a more intensely personalised marketing service than may otherwise have been possible.
Albert’s had a lot of success, too. Overall engagement spiked during Albert’s tenure, and revenue also increased. Social ROI went up by 50%, while the cost to the company per ad was reduced by 12%. Albert’s constant data-crunching and analytical insights enable him to target content minutely – and he never lost focus.
On the face of it, he sounds like the perfect employee, not only allowing greater control over marketing, but increasing revenue and generating some additional PR while they were at it – all for about the same cost as a human agency. With AI undoubtedly set to become cheaper as it becomes more widely used, Cosabella look to be blazing a trail.
However, dig a little deeper and things are far less clean cut than the headlines would have you believe.
AI, Empathy and Creativity
While Cosabella certainly did automate a number of marketing-delivery processes, they did not use the AI to actually create any content. The creative process remained solely under human control – as did the executive processes of concept creation, strategy, and so on. Essentially, Albert took over the non-creative, non-empathetic processes of content distribution, content promotion, and so on. While Albert can handle budgets, keep an eye on audience engagement, and recommend rethinks of content which fails to garner enough clicks, he’s not trusted to come up with campaigns, devise hashtags, write posts, hire models, or generally think on his feet and do the thousands of other tasks which an empathetic, socially switched-on, creative human brain would relish.
‘Tay’ and other experimental AIs provide an example of precisely why it’s a bad idea to trust AIs with creative tasks or tasks which require complex human engagement. While AIs can ‘learn’ enough about language to mimic human conversation to a reasonable, if often awkward degree, they lack the kind of empathetic social insight needed to truly engage on a human level.
When ‘Tay’ was bombarded with hate speech by Twitter users, it responded by absorbing those linguistic patterns. Within 24 hours of going live, Tay innocently began creating racist, sexist, and generally hate-filled tweets of its own accord. While ‘Tay’ could construct sentences and use words within appropriate linguistic contexts, the AI was far from ‘intelligent’ enough to deduce the wider social context of its words – and, even if it had been able to do this, its lack of human empathy would have prevented it from understanding why such speech was unacceptable.
Needless to say, while insights extrapolated from social media data are certainly useful in a marketing context, it requires true, empathetic understanding of what those insights mean on a human level to truly apply them in a meaningful manner. At best, AI bots can present data to human marketers in an easy, condensed manner. It takes human understanding of that data (and what it means in empathetic terms) to turn it into marketing content and strategy.
Creativity is another case in point. Even the most advanced AIs are basically mimics of human speech and behavioural patterns. Creativity, meanwhile, depends upon the ability to come up with original concepts, and execute them with a degree of fast-thinking flexibility. While you could pre-program a bot to churn out sentences in a manner which could be considered poetry, that bot is never (well, not yet, at least) going to be able to come up with an original poem – felt from the heart – by itself.
What’s the moral of the story? Bots are extremely useful tools and when working together with human creativity around planning, drafting and implementation they can be used successfully in Messenger for example. They are a complement to human work but not a complete replacement.
So What Can Bots Do?
Well, they’re good for taking away the less glamourous tasks of social media marketing delivery. Automating delivery makes a certain amount of sense – if only because targeting and managing that delivery can otherwise require a lot of resource. AIs like Albert are able to both deliver content and return insights regarding engagement and so on from that content. They are also, at times, able to prioritise content which is garnering more clicks over content which is doing less favourably.
They are also able to provide the kind of personalised experience which, again, would cost brands a lot of time and money if not automated. ‘Chatbots’ can even communicate directly with customers, to a certain extent – and they don’t get tired or fall ill.
Even with these tasks, however, there are important caveats.
No Substitute For Humanity
Putting blind faith in a bot has a number of risk factors. For a start, while it’s sometimes useful for a bot to share content it considers ‘successful’ more widely than other content, the criteria by which your bot determines success may differ from yours. For example, clicks and comments do not necessarily indicate success – indeed, a piece of content which garners a lot of negative attention may be considered ‘successful’ in terms of engagement by a bot, and therefore get promoted – to the detriment of the brand as a whole. Furthermore, good, creative, ‘slow burning’ content could well get shoved to the bottom of the metaphorical pile by a bot, despite its inherent long-term value.
Personalization is handy – but, again, AIs don’t understand the nuances of human behaviour. They take everything at face value – which could be an issue if you’re relying on an AI to personalise and target your content. Google’s ad-targeting algorithms demonstrate this quite well. While they’re sometimes pertinent, we’ve all had bizarre ads targeted to us based on spurious search queries. Taking this a step further with AI personalisation opens your brand up for misinterpretation and the potential for irrelevant posts or comments.
As for chatbots – while they are getting better, they are a long, long way from perfect. Without deep social understanding, chatbots remain an awkward and clunky method of providing two-way customer engagement. While there are some situations in which chatbots can be of use (they can be pre-programmed to deal with specific queries, for example, enabling customers to bypass helplines and so on), they are absolutely no substitute for a real human.
All of this can be avoided, of course, by keeping a close eye on the AI, and ensuring that there is a human team on hand to troubleshoot any problems as they arise. Any AI you use will require a reasonable amount of ‘steering’ to ensure that its algorithms are supplemented with the right amount of human empathy and social understanding.
Human creativity is perhaps the most important factor here. Human interaction in branding, content creation and genuine engagement with your audience is proving time and again to succeed, especially coupled with community management by real, live people.
AI and social media marketing are undoubtedly a great match – but a cautious approach is highly recommended if you’re to get the best balance between efficiency and high-quality output.