Artificial marketing: AI lessons from big brand successes and failures
Share
Jon Stubley looks at the chatbot, creative algorithm and pre-emptive marketing hits and misses of big brands, and what Australian marketers and brands can learn from them.
Last month, University of New South Wales and Data61 professor Toby Walsh warned that Australia is failing to embrace artificial intelligence (AI) in the business sector despite “punching above its weight” in AI research.
Walsh, the author of It’s Alive: Artificial Intelligence from the Logic Piano to Killer Robots pointed to a 2017 Infosys report on AI which looked at businesses from seven different countries and found that Australia was the least likely to implement AI within its business.
In China, for example, 100% of respondents planned to use AI, but in Australia the figure was just 20%.
This is a little worrying in a country that – despite pushing the idea of being ‘The Innovation Nation’ – is also falling behind in terms of digital competitiveness and skills according to a report from the International Institute for Management Development. There are glimmers of hope in our sector and I believe the media and marketing have a real opportunity to lead the adoption of AI in Australia. The very nature of our (relatively) short term advertising and marketing campaigns mean we are able to more agilely test and tweak and learn in real time, which is essential for the successful adoption of AI.
Since we launched here about 18 months ago, there’s been a real willingness from Australian brands to explore the potential of AI driven advertising campaigns using GumGum’s AI computer vision technology. And on the global stage, larger brands are advancing and experimenting with a diverse range of AI driven applications.
Not all have been successful, but we can watch and take learnings from some of the recent successes and failures.
Of all the possible areas for AI deployment in the media and marketing space, three stand out with clear learnings for marketers:
Chatbots
Forrester research shows that 5% of companies worldwide were using chatbots regularly in 2016, 20% were trialling them and 32% were planning on either using them or testing them in 2017.
Not all of the experiments have been a success, but looking forward, we are certainly going to see them used for more and more sales and marketing applications.
1-800 Flowers was an early adopter of chatbots, launching its first bot on Facebook Messenger in April 2016, followed by Amazon’s Alexa and its own concierge called Gwyn, which was built on top of the IBM Watson platform. The company has revealed that one of the most surprising discoveries is that people like interacting with bots more than humans, particularly on the order entry side: a survey of customers indicate that over 80% would use Gwyn again and the company is happy with conversion rates, with cart abandonment rates lower.
On the flip-side, another early adopter of Facebook Messenger bots, US apparel firm Everlane, has ceased using them, choosing to stick to email instead. No details have been given, but on the retail side of bots, a lack of personalisation and visual imagery is often cited as a barrier.
Lesson for marketers: businesses need to ensure their goals aren’t too ambitious for the available technology or launch into market too soon. Their success will also be predicated on their target audience’s familiarity and previous experience of messaging apps. At this stage in chatbots development, they appear to be more suited to sales versus a pure marketing function.
Creative algorithms
Early this year, Mario Bosaz, Coca Cola’s global senior digital director said he was keen to use automated narratives in the creative process.
Although a long-term vision, Bosaz believes AI has the potential to be used to write scripts and create music; as well as placing on social media and buying media.
Several big agency groups are following suit, with AKQA testing an internal tool built with IBM Watson that searches online platforms to find new audiences for brands, and Saatchi & Saatchi LA running several AI campaigns for brands.
One of these was for the Mirai, Toyota’s car of the future which runs on hydrogen fuel cells. The campaign was written by Watson after Saatchi spent several months training it to piece together coherent sentences. The idea was to produce scripts that would be targeted at likely early adopters of the car of the future: scientists and engineers.
The journey wasn’t entirely linear, but it finally elicited results – albeit ones that were vetted by human copywriters.
Lesson for marketers: Although currently reserved for those with deep pockets and large research and development budgets, AI will increasingly play a part in the creative process for both testing, trialling and creating in the not too distant future, as algorithms become more advanced and precise.
Pre-emptive marketing
Companies like Amazon and Netflix have already found tremendous success by using machine learning for pre-emptive marketing that is making recommendations.
According to McKinsey’s ‘State Of Machine Learning And AI 2017’, Netflix found that customers on average give up 90 seconds after searching for a movie.
When it improved search results, Netflix believes it avoided cancelled subscriptions that would have cost the company $1 billion annually.
The company estimates that its recommendations drive three-quarters of all the videos watched on the service.
Similarly, Amazon drives approximately a third of its business via machine learnings recommendations and uses AI and machine learning across many of its business areas.
Lesson for marketers: While Amazon and Netflix have been able to get ahead of the curve given their deep pockets, the future of marketing will increasingly be predictive, hyper-personalised and responsive.
Brands of all sizes will need to find a way to offer a personalised service across an increasingly complex marketing environment and should be looking at AI as a vehicle to do so.
I’m looking forward to seeing further experimentation with AI, both in Australia and overseas as the technology matures further and machine learning becomes further entrenched in our everyday lives both as marketers and consumers.
Hopefully in the not too distant future, some of the best practice examples will have been developed and deployed here too.
Jon Stubley is MD at GumGum ANZ
–
Further reading
–
Image copyright: nesacera / 123RF Stock Photo
