The true impact of AI on marketing

This article examines exactly how artificial intelligence and machine learning might affect the various aspects of marketing and specifically, how marketers can get the most from this emerging technology

Most of us can position ourselves somewhere on The AI Fear Continuum.

  1. What’s AI?
  2. AI will never be a threat to my professional or personal life
  3. I have general concerns about artificial intelligence
  4. I’m worried about AI replacing me in my job
  5. I’m worried about AI being used in cyber-attacks, spying and data gathering
  6. I’m very worried about both of the above

For nearly 100 years, science fiction writers have prepared us for the worst as they explored scenarios from Hollywood 1927’s Metropolis to adaptations of Isaac Asimov’s I, Robot and Philip K Dick’s Do Androids Dream Electric Sheep (famously the basis for Blade Runner). We’re ready to be tricked (Alien), enslaved (The Matrix), killed (West World)  or even have our whole species wiped out (Terminator).

Conversely, news stories about AI feel overwhelmingly positive, informing us that AI is saving the day in sectors as diverse as health care, law and climate change. Marketing professionals have two key points to consider; claims that AI can help to make marketing more creative and effective, and claims that it will make many of our job roles surplus to requirements.

This article is for those marketing professionals at Continuum Level 4: with concerns about how AI will impact their jobs. In light of ongoing speculation and discussion, it’s time to pinpoint which specific roles AI can play in marketing, and whether it is a saviour or a threat.

AI: now and next

Right now, key claims and considerations around AI in marketing include:

  • People hates ads and want more meaningful - personalised - communications from brands
  • Companies are already competing to use AI to give more personalised customer experience
  • “In the next five years, marketing without AI will be no marketing at all” (Malcolm Frank, Cognizant)
  • AI will help behavioural planning through predictive analytics
  • AI will aid Conversion Rate Optimisatiom (CRO)
  • Marketers don’t yet know how to use AI to impact KPIs
  • The majority of marketing employees are concerned about the impact of AI on job losses

These are all worthy thought-starters, but don’t leave us closer to knowing how AI will affect our jobs on a day-to-day basis. Despite growing fear in the industry that AI could replace humans in a range of roles, it’s rare to find detailed evidence that backs that up.

As a content marketer with a background in journalism, nothing I’ve seen so far has me afraid I might lose my job. So let’s look at the facts.

What is AI good at?

Speed

AI is great at reducing the time it takes to do computation or even creative tasks. Reduced time can mean reduced costs, lower overheads and increased margins. The Washington Post has already published more than 850 automated new articles, and Reuters recently announced that they are building a tool to ‘help’ journalists by analysing data, suggesting story ideas and even auto-generating some sentences.

The aim, for now, is to support journalists, not replace them.

Volume

Related to speed, is volume. Due to the speed of computations, AI can analyse vast amounts of data much more quickly than a human being. It can also create vast amounts of content, from simple text to video, AR and VR.

Accuracy

Because output from AI is based on more data and more ways of using that data than an individual can crunch, its computations and recommendations will be more accurate.

In this respect, things AI can bring to marketers’ tool kits include:

  • Improved language/speech recognition
  • Better ad targeting
  • Predictive customer services
  • Website design recommendations
  • Content recommendations and curation

This will impact those working in paid digital, web design and CRO, though it remains to be seen whether a leap could be made from speeding up marketing processes and making them more accurate, to actually replacing humans in the workplace.

What does AI struggle with?

As a writer, I was intrigued (terrified) to learn that one of the big US news brands was using software to turn financial reports into articles some years ago. They had done so without announcement and found that no one could tell the difference. Then I was told about Articoolo - a web-based app that writes articles based on inputting a single keyword. I immediately tried it and found that the content was broad, packed full of platitudes, lacked a sense of rhythm or tone of voice.

Not quite what I was after.

It’s important to remember that what might be useless today, could become good enough to rule the world tomorrow. As a writer, I feel safe for the time being.

Further proof that writers’ jobs won’t be in jeopardy any time soon can be found in the AI that wrote a sci-fi screenplay. While fun, funny, even moving, it is total nonsense.

Considered side-by-side with the robo-reporters of Reuters and The Washington Post, it is interesting to think about why AI can do one but not the other? Is it because one is ‘analysis’ in which the mimicry of previous forms allows success, while the other is ‘creative’ involving original thought?

How marketing can get the most from AI

An important thing to consider is that many marketers don’t know how to get the most from AI. It’s chicken and egg: you don’t need it until you’re using it, but you won’t use it until there’s a need. While I - as a marketer - don’t fully know how to get the best from AI, I find myself asking: well, what can it do? The answer is usually, ‘anything, what do you want it to do?’ And so the conversation chases its tail.

This is the key: marketers need to say what we want and see if AI can do that. Necessity is the mother of invention, so what do you need?

What could AI help marketers with?

Branding

Well-developed AI could look at a product, analyse market data and audience insights and suggest a brand name, slogan and brand guidelines - including colours and photography.  But this is a long way off.

Measurement

Most marketers are forced to pick one or more KPIs, even though each might not tell the whole story. Not all UPVs are of equal value, so how to do we create a meaningful picture of marketing performance? By the time you’ve added in dwell time and bounce rate, clients are often saying, ‘yes, but what does that mean?’ with any patterns hidden by averaging.

AI can be used to crunch huge amounts of data, to give meaningful measurement.

Attribution

Content marketers in particular don’t want to admit that pound-for-pound ROI is like kryptonite to us. The web has dozens of ‘how to measure content marketing’ articles, and the fact that they’re all different suggests that no one’s licked it in a way that satisfies Marketing Directors and their bosses.

A/B Testing and optimisation 

Wouldn’t it be great if you had an AI to constantly A/B test your web content, to optimize that content for a given KPI? It could analyse every single interaction and improve anything from readability and formatting to link additions and design - even adding in sections.

Personalisation

This has been mentioned, but personalisation at scale is impossible - if it isn’t, then you don’t have many customers. So the ability for AI to know who you are when you interact with any channel, and tailor the journey for you, would be a boon.

It goes hand-in-hand with the above, a personalised journey that is constantly being optimised for every individual - dealing with infinite complexity, second to second.

Targeting

Paid digital managers spend lots of time nipping, tucking and tweaking variables to make the most of spend. This could all be automated.  AI would be able to analyse huge amounts of data to target bespoke audiences that are more sophisticated than advertising channels (Facebook, GDN) can offer. For example: you may want to access a specific cross-section or sub-set of a broad option in Twitter.

AI could also create target lists for PR and influencer outreach instantly.

The end of subjectivity and client feedback

Marketing success can often be diminished by subjective opinions of non-experts. 100 stakeholders - non-writers and non-designers - all inputting into design and copy until you have creativity killed by committee. Even creatives enjoy the power of data when it can prove that their approach is the right one.

But there are still battlegrounds where no hard data exists and we must argue the toss with clients: photo choice, tone of voice, what to include, what to cull, aspects of design. AI could argue on our behalf. In fact, would it mark the end of client feedback?

Content

Maybe I’m in denial, but I feel like we’re a way off from AI being able to create content with the extensive literary tool kit of a human writer: nuances of humour, metaphor, satire and imagination may always be beyond the reach of AI. The point is almost moot as the aforementioned media outlets are already using AI to create news items, but so far, only in stat-heavy content like financial forecasting and even sports results.

Where AI would be useful is in the creation of briefs: scouring data for gaps and opportunities, building a picture of what an article needs to cover as well as the usual meta-description, H1 tags, URL and keywords. Tools such as CognitiveSEO, SEMrush and Ahrefs show we’re well on our way to this. We just need the ‘create article’ button.

Consumers are emotional, so what is essential is for AI to understand what drives emotions and therefore how to control them - and we’re getting close.

SEO

SEO teams have access to vast amounts of data, from Google Analytics and competitor sites to those mentioned immediately above. Each can be compared and construed in an almost infinite number of ways - which takes time. Also, objectives and strategy are important factors in driving output.

An AI would be able to take an objective, analyse a universe of data, interpret it to find the best/easiest opportunities and recommend watertight solutions for maximising them: content, links, keywords, audiences.

Video

Imagine not having to pay thousands for an explainer video. You just upload your script with directions: fade in, move up, and so on. Then, you instantly get an animated video back. Even with tweaking, the time saving would be immense.

Social media

AI is already woven into the fabric of social media, from relatively benign algorithms that show you what you’re into, to bots and fake accounts. The ability to create profiles that become almost-instant influencers, by getting them to mimic the behaviour of successful influencer best-practice, is a marketer’s dream.

If people like the user experience, they may even start to prefer them to real humans. The success of chat bots goes some way to suggesting that this will be the case.

Project management

There are already lots of great project management tools, but could an AI ever fill the role of a project or account manager? Can AI be as flexible, adaptive and diplomatic as an account manager? Or would everyone simply ignore the AI account manager like we ignore other things we know are just a computer? Automated alerts, bots and adverts, for example.

Design and build websites

Software such as Wix is allowing tech-novices to design whole websites, but we’re already at the next stage: AI creating websites from drawings, and even websites that design themselves.

Ideation

Could an AI ever come up with a brilliant marketing concept? Just Google ‘best marketing campaigns 2017’ (or whatever last year was), and ask yourself if an AI could have come up with those ideas. The chances are that most campaigns were informed by data, maybe even lots of data, but takes the human touch to turn it into an actual idea.

Could an AI have come up with Heineken’s brilliant Worlds Apart ad?

Or the Fearless Girl of State Street, NY, in celebration of National Women’s Day?

Value, disruption and ‘cutting through the noise’

All modern cars look the same to me, because science defines what is safe and fuel efficient. As such, the Mini Cooper and the Fiat 500 cut through.

In a world where super-intelligent algorithms define everything, will our marketing all look alike? Would two separate car insurance companies end up creating identical campaigns and collateral?

If so, it will only reinforce the need for a human element: creativity, left-field, out-of-the-box thinking, humour, silliness and strokes of utter genius. Articoolo is already tackling this with its content creation tool, giving users the choice of readability or uniqueness.

Print media declined in the new millennium as web offered value, novelty and innovation. By 2010, print was back in vogue because so few marketers engaged with print. It was the new novelty, the badge of honour in an uncrowded market, and many brands re-embraced magazines - from Rolls-Royce to Airbnb.

The list above may give you a renewed sense of career longevity, or it may make you feel more ‘redundant’ than ever, but the most important way for marketers to make the most of AI is not just ask what it can do, but ask it whether it can do what we need it to do.