Skip to main content

Hope, hype and humans: the foundations of AI in marketing

(Image credit: Image source: Shutterstock/PHOTOCREO Michal Bednarek)

What do you think about when you think of artificial intelligence (AI) in marketing? For most, it’s hype and hope. Hype, that AI can transform the workings of any marketing department in any company. And hope, that by bolting-on AI to your marketing stack, you’ll be able to overhaul how your marketing team manages customer data, and in doing so improve the experience for your users.

The reality though, is somewhat different. Hype has, in most cases, been driven by vendors wanting to push AI products to companies regardless of whether they need them, or whether they’ll benefit from the technology. AI isn’t suited to every company, particularly when it comes to marketing. And that hope of easily ‘bolting on’ AI to a marketing stack just isn’t achievable. The impact (for your team and your customers) isn’t instantaneous, and the technology itself cannot provide quick-fix transformations for your team’s workflows and customer data management process.

However, get it right, and adopting AI in marketing can deliver impressive results. According to Forbes Insights and Quantcast research, AI enables marketers to increase sales (52 per cent), increase in customer retention (51 per cent), and succeed at new product launches (49 per cent).

So that’s the hype and the hope. What’s missing from the AI equation is the third ‘h’: the human factor. Adding AI into the marketing stack requires incredibly careful attention to deep details of data modelling and data hygiene. This involves human training and effort, and ongoing investment in people as well as technology. An AI platform must be useable by individuals across an organisation, for instance, and must be configurable and customisable to ensure a streamlined integration with a marketing team’s existing IT stack.

Companies need to ensure they utilise AI technologies and models that make the most sense for their business and marketing strategy. If not, they won’t succeed at AI in marketing, nor be able to address and overcome some of the most common challenges with customer data management. ‘Failing’ at AI isn’t uncommon: a quarter of businesses responding to a 2019 IDC survey reported a 50 per cent fail rate on AI projects.

Unifying data

With so many failures, adding AI into the marketing mix can seem daunting. Compounding this is the fact that AI is a relatively new concept, and awareness of what the technology can deliver is limited. However, one major area where AI can benefit marketers is in managing digital experiences. And at a time when digital, multi-channel experiences are the new normal, without the support of AI and other emerging technologies in managing those experiences, those brands which fail to digitalise will be left behind.

This is particularly true of the past few weeks; a period that has highlighted just how crucial a seamless, relevant and engaging online customer experience is. If they’ve not done so already, businesses really must move to a digital-first way of managing customer experiences across multiple digital platforms, and unifying these.

Digital experiences in the spotlight

With lockdown restrictions and physical footfall decreasing dramatically (or in the case of many retail businesses, completely), customers instead have relied on their smartphones and laptops. The digital domain has been the only way for consumers to buy goods, access and manage their finances, educate their children, enjoy entertainment, pay bills and use any other service that used to involve some in-person element.

As customers flip from one digital channel to the next they’ll want (and increasingly, expect), for the company in question to provide a joined-up experience. When it works well, users will likely not realise – it’s when it a company doesn’t deliver on this that frustrations (resulting from a poor customer experience) will arise.

A consumer might, for example, pay a lot of money for a concert ticket and sign up to the ticket site’s mailing list, only to be sent recommendations for the same concert they’ve already spent a lot of money on. Or, a shopper browsing and buying men’s clothing on a fashion website might repeatedly be shown recommendations for women’s clothes when they next log on. Or, a consumer trying to authorise a transaction online might call a customer service helpline, only for the representative on the other end to have no knowledge of the customer or the transaction. Or, a customer shopping in Australia might be shown out of season stock on the UK version of a brand’s website. Or, a nut allergy shopper trying to do an online grocery order via their smartphone might be sent push notifications for special offers on Easter confectionary containing nuts. The list goes on!

The right tool kit

Many businesses will hope that by throwing AI into the mix, these customer experience issues will instantly disappear. However, some AI tools will fail to deliver – and the business in question will subsequently ‘fail at AI’. What’s needed is an AI-capable personalisation tool that utilises data from across different channels and from third-party sources as well as data available in house.

Things like geolocation, visit frequency, device and system used to browse your site, pages, viewed, browsing behaviour all need to be taken into consideration. Tools must then be available to marketers to analyse this data, to better understand it, use it to make predictions, inform decision making about what personas consumers might fit into, and then action this to display the content and offer the experience most appeal to those individuals. One human marketer doing this for one consumer is possible. Offering this one-to-one experience to thousands or millions of customers is not.

AI technologies can enable one-to-one personalisation at massive scale and in real or near-real time. Such a tool will gather customer, product, order and behavioural data from retail points of sale, and e-commerce order management, email marketing and analytics systems. It can then perform data hygiene, de-duplication, and standardisation, enabling marketing teams to gain an accurate, complete view of each and every customer.

That’s the tech part; now for third ‘h’ – those humans. To avoid an AI fail, these tools have to be straightforward to use by all team members – including those hesitant about the adoption of AI. It should also be possible for businesses to use customer data platforms alongside other applications, dependent on the experience and knowledge of marketers.

Google Marketing Platform, for instance, features Google Analytics, which is already used by many marketers who wouldn’t consider themselves AI specialists. Google Cloud, meanwhile, includes tools that allow users to analyse data using AI. Alternatively, if members of your team have stronger analytical skills and are familiar with AI, tools like BigQuery ML or AutoML have clustering and prediction tools built in. An AI-capable customer data platform should be able to work alongside these other applications, allowing a business user to choose the approach that works for its team.

Businesses should organise their marketing strategy, stack and organisation around AI and with a focus on the specific outcomes they want to achieve. Falling for hype and hoping the technology can simply be forced in are what lead to those AI fails. By taking this human-centric approach to AI integration, companies will be able to maximise the benefits of such technologies, unifying customer data across digital and physical technology silos, providing meaningful insights into the data, and orchestrating consistent, relevant customer engagement at every touchpoint. After all, it’s these human experiences that define brands.

Omer Artun, Chief Science Officer, Acquia

Omer is Chief Science Officer at Acquia and the founder of AgilOne. He holds a Ph.D. in Computational Neuroscience and was a consultant with McKinsey & Company, consulting high-tech and retail companies on strategy development.