The range of technologies available to CMOs is daunting. This promised land of AI technology for marketers was supposed to become even more of a reality around five years ago, when AI entered the scene as the newest shiny object. And since then, AI-driven platforms and capabilities have come to be viewed by many CMOs as ‘must haves’, even though they’ve rarely lived up to the hype. So why has the promised land not yet been realized?
To answer that question, let’s first take a look at three of the biggest challenges that AI for marketing faces.
Read Also: ‘Leapfroggers’ Using Emerging Technologies To Grow Revenues 5X…
What you put in is what you get out.
Pretty much anyone can create seemingly bulletproof algorithms or machine learning methods that automate marketing. But those algorithms are often only as good as the fuel they are supplied: the data. For example, personalization has long been a target use case for AI in marketing for product recommendations, behavioral targeting, or even one-to-one personalization.
But the effectiveness of AI is constrained by what it knows about a customer. If, as a brand, you can’t provide high-quality, consistent customer profile data, it doesn’t matter how good the machine is – it just won’t work.
Trust in the Machine (or Not)
Marketers are inherently cautious when it comes to handing over control to a machine. And rightly so – being brand-safe is just as important as finding ways to automate and scale. The ‘set and forget’ or ‘black box’ approach of many AI marketing technologies puts the unnecessary risk on a brand’s reputation, as well as its ability to correctly engage with its customer base. Marketers are also rightly concerned by the ‘message overkill’ that some automated approaches can lead to; customers can become weary very quickly if automation leads to cross-channel, cross-device stalking.
The answer here is that often there needs to be human oversight, but how do you balance this with the benefits that automation can bring?
AI for AI’s Sake
Many CMOs feel pressured to invest in AI just because they think they’re supposed to, rather than focusing on the business problem they are trying to solve. There is no doubt that AI can help solve real customer challenges.
Read Also: Norgesmøllene AS, Part of Cernova Group, Moves to the Cloud with Elvenite and Infor
CMOs need to focus on those and pick a solution that can clearly demonstrate a solution and prove a return. It’s also important to start small and then increase investment in these tools once they have shown success, rather than investing huge sums of money upfront without having clear and measurable metrics for success.
NLG, Deep Learning and Optimization
When it comes to language, it seems more emotional rather than scientific or data-driven – but it doesn’t have to be. It is possible to take a data-driven approach to language, which can deliver emotive messages that amplify the essence of your brand. If marketers get their message right, they can not only improve their bottom line, but create an engaged and loyal customer base.
So what does this have to do with AI?
Well, AI makes it possible. Natural Language Generation (NLG) is a field of AI that has been around for decades, but when you combine it with deep learning and optimization techniques, you get a powerful tool that can solve a real marketing problem: how do I get the right message in front of my customers at the right time and make it consistent across channels?
In the last few months, NLG has become more prevalent in the media with the release of GPT-3, a huge pre-trained language model that can be used to generate human-like text.
But it still suffers from some of those challenges that were outlined earlier – it is truly a black box approach without human oversight. The result of this is that while it can generate diverse and creative language for use in marketing, it can also suffer from what we call ‘hallucination.’ This is where the generated language leaves out critical information, or even worse, produces language that is offensive, tone-deaf or completely off-brand.
The answer to this conundrum combines proprietary NLG, deep learning and optimization with human-specified guardrails that ensure all language is on brand and tuned to the needs of each marketing campaign. This combination solves the three challenges for AI: it’s trained only on your brand voice so that every message it generates is on point; it provides scale and automation, but has human oversight and approval workflows built in to give marketers the confidence they need and it solves a real problem; it proves its ROI; and it finds the optimal message to send for each campaign and calculates the impact in dollars and cents.
The Proof – Walgreens Boost Vaccine Rates
This technology has been vital to Walgreens’ efforts during the pandemic to vaccinate as many customers as possible. Walgreens has 50 million customers it wanted to communicate with via email about getting the vaccine. It was more important than ever to get those customers to open the emails and engage. By deploying a combination of NLG, deep learning and optimization, Walgreens boosted email open rates by 30% – meaning 30% more people received info about available vaccine appointments. This protected up to 30% more Walgreens customers from COVID.
Recommended: Mundrisoft partners with a leading US-Based EdTech Company to Accelerate Learning Delivery
The pandemic has changed how brands connect with their customers. It’s created a huge new captive digital audience, as consumers and advertising dollars shift from the storefront to the screen. It’s driven a surge in demand for world-class content. And it’s placed language at the heart of any customer experience strategy. Yes, it’s a more challenging world for marketers than the one we knew in 2019, but one full of opportunity if you have the right tools in hand.
[To share your insights, please write to us at sghosh@martechseries.com]
Comments are closed.