Listen to somebody working in CRM talk about the challenges of their day-to-day job and chances are you’ll hear two words – productivity and personalisation. Simply put, marketers want to spend less time building emails, but also drive uplifts in clicks and conversions.
This dynamic has been the backdrop for marketers for some years, as they seek to put a tech stack in place that will make their life simpler and more efficient. With advances in AI, businesses are re-evaluating what’s possible; marketing technology is evolving from an era defined by ‘systems of intelligence’ (CDPs, single customer view, predictive analytics) to one defined by ‘systems of autonomy’ and the concept of the machine as ‘co-pilot’.
AI models have a clear role to play in CRM in the classification of creative and context, in understanding customer behaviour, in decisioning, and increasingly in generating content.
What if we can know how our customers’ behaviours are shifting and automatically tailor our campaigns and how we serve them?
-Vivek Sharma, CEO, Movable Ink
In practical terms, for those in the middle of email campaigns or teeing up mobile pushes, Vivek Sharma, CEO at Movable Ink, outlines the scope of AI for running autonomous campaigns.
“What if we can know how our customers’ behaviours are shifting and automatically tailor our campaigns and how we serve them? What if we could adapt our imagery to each customer’s unique tastes and future interests; generate text that represents our brand voice, that speaks to the individual in the right tone; vary the [send] time and frequency?”
“What if we no longer needed a developer to code up our marketing ideas, but it goes straight from the idea in your head to tactics that can be applied across every channel? And finally, what if we can walk away with useful insights that fundamentally change what we know about our customers?”
From a standing start, these ideas may strike some marketers as daunting (particularly those accustomed to ‘batch and blast’ email), either because of the perceived complexity of orchestration or the fact of relinquishing a degree of control to a suite of algorithms.
However, many brands speaking at Movable Ink’s Think Summit earlier this month are already sending dynamic multichannel communications that are influenced by customer behaviour (whether powered by a decisioning model or drawing on first-party data). For these marketers, the increasing involvement of AI, with premiums on brand safety and data privacy, is a logical next step in pursuit of productivity.
Decoupling campaign from creative
The mindset shift for CRM teams is one of “decoupling campaign from creative,” says Rachel Cowlishaw, Director of AI Activation at Movable Ink, adding that email marketers should still be thinking about their templates (for example, how best to mix editorial with promotional or loyalty blocks) but also “managing a creative library of assets, [such as] subject lines, -header text, links, urls…”
Then, according to business rules set by the marketer, an AI model chooses from this library, where creative may sit for 30, 60 or 90 days.
This approach increases productivity because it creates a movable feast. Instead of getting every message, often arbitrarily scheduled, customers receive a mix of the strongest brand-led content and the products or offers a model deems to be of interest to them. The AI models can tailor subject lines to accompanying hero images and a template will automatically update content such as promotion disclaimers.
For batch and blast emails with open rates of 10 to 20 percent, “What about the 80 to 90 percent that didn’t see the email yesterday? It’s a huge missed opportunity,” says Cowlishaw. She explains how a piece of ‘hero’ creative sent to a full file of 5 million customers with a 10% open rate will only reach 500,000 people, perhaps more if a redeploy is done. But if this creative lives for 60 days as one of 50 different hero options in a library, and is automatically redeployed, guided by a model, including to new customers on file, it could be that the reach of this creative is tripled.
An ensemble approach to AI models
For some marketers, this concept of decoupling campaign and creative may sound like a good description of sophisticated, but not ground-breaking, marketing automation. However, looking at an example campaign in Da Vinci, Movable Ink’s AI platform, shows how algorithmic models work in concert to create what can be 1:1 personalisation i.e. enough variants of a particular campaign that some customers will receive a unique combination of creative, from subject line to featured products.
The goal is not 1:1 comms in and of itself, this is simply a by-product of models optimising for short-term revenue (via reinforcement) or longer-term goals (such as category or product discovery and therefore increased AOV).
CEO Sharma outlined four model ‘classes’ that influence a typical campaign in Da Vinci:
- Vision models: these can read and ‘understand’ marketing and product copy, differentiating between brand and promotional content, identifying prices, and analysing and tagging imagery (for example, listing the attributes of a product e.g. a mid-century black leather sofa with wooden legs).
- Generation models: as the name suggests, capable of crafting subject lines and generating reports.
- Insights models: perhaps the area where marketers feel most at home and essentially the modern-day evolution of a direct marketing RFM model, these customer facets will explain the taste of a customer and their preferred categories, their affinity for different types of offers and their propensity to spend. As well as building customer profiles, the platform scores model effectiveness – what is driving uplift and what isn’t.
- Prediction models: creative decisioning, send time personalisation and frequency management.
Anyone who uses the web will recognise these types of models from tech innovators, perhaps in ecommerce or on social media platforms. When they are used to their full extent and together as part of an “ensemble model approach” they can create hundreds of thousands of creative variants.
Brands such as Lands’ End are doing this with their CRM, utilising an entire creative library, personalised to every individual in the customer database. Combined with real-time data it could mean, for example, a brand selling more full-price out-of-season stock, such as shorts in October, because the AI model predicts that certain customers are potentially in the market for their holiday wardrobe and tailors the messaging they receive accordingly.
Of course, the power of AI doesn’t mean personalisation has to always split every hair or lean solely on promotion to produce compelling results.
Creating time to use valuable marketing instincts
It’s easy to paint a picture of AI offering a two-speed approach to CRM. There’s the carefully crafted brand strategy, perhaps with some AI tweaks, and then there’s the fine-tuned conversion-focused promotional campaigns whirring away in the background. Whilst this split (long and short?) is somewhat of a false dichotomy, it may be true to say that AI-powered CRM can free marketers from the tyranny of the overcrowded email calendar.
AI can win back time and, as Movable Ink’s Rachel Cowlishaw puts it, that’s invaluable when there are, “more things to say than time and space to say them.”
The productivity benefit of personalisation was demonstrated neatly at Think Summit by two simple but powerful campaigns run by health and beauty retailer and pharmacy brand Boots, which uses Movable Ink Studio to apply decisioning logic to its email.
Sagi Laniado, Senior Manager, Customer Journey Orchestration at Boots UK, discussed the workflow benefits unlocked by redesigning the brand’s weekly offers email.
Laniado described the email’s format a year ago as an “endless” list of promotions, with “no consistency, no guidelines, no personalization.” Each business unit nominated their best offers for a long email that would go to every customer on a Friday morning, with the same content, no matter what the customer profile or purchase history.
Preparation of the email was time-consuming, too, with Laniado sharing the challenges of “business units pulling out at the last possible minute,” and “offers that become out of stock before send.”
The solution was to use a decisioning model to assign each customer with two categories of offer. “The model will say what each customer is most likely to convert on,” said Laniado, “[and] each customer will see a different order and a different set of offers based on their preferences.”
The result was a 25% increase in click-through rate, a 24% increase in conversion rate and a 90% decrease in production time.
More time back for the journey orchestration team means more time for brand-focused work such as Boots’ recent My Menopause Journey campaign.
This topic is one that Boots is well-placed to advise on. The brand has partnered with ITV and its menopause bus campaign, travelling around the country, and has also launched a menopause healthcare hub on its website. Most of the retailer’s loyalty base is female, and those over the age of 45 represent some of its most frequent shoppers.
“There are a lot of customers who trust Boots as a brand,” says Laniado, “and they’re willing to give us permission to use their data, to give them advice and support with their health care needs and suggest new services and new products.”
Customers signing up to the My Menopause Journey receive an email asking about the symptoms they are experiencing, such as changes in skin condition or changes to sleep, and this data (part of a sensitive consent workflow) is then used to trigger follow-up communications with relevant advice.
The results were described as “pretty amazing” by Laniado, with a 66% open rate and 27% average clickthrough rate described as “massive” and “something we don’t see in any of our marketing emails”.
As Laniado explains, these emails feature very limited content devoted to product. “It’s all support and advice. It’s only copy speaking about what the customer says is relevant and interesting for them to know. And that just proves that you don’t really need to have a lot of offers and discounts on the email in order to get that engagement.”
This is surely the sort of valuable work marketers are best placed to create.
AI may create a ‘1000x marketer’ but the fundamentals of the profession remain
Movable Ink CEO Vivek Sharma believes the productivity gains for marketers can be vast, pointing out that just as software abstractions created the idea of a ‘10x developer’, so too AI can create the ‘1000x marketer’.
“There are six million developers in the EU today, out of nearly 200 million workers. Just think about that. Only three percent of the [working] population. To get software to do something custom, to shape it to your needs, you need to talk to a developer, [to] sit in the queue.”
It’s really all about putting the customer over the campaign. It’s also fundamentally about marketer and machine.
-Vivek Sharma, CEO, Movable Ink
“This next phase will break you free from the shackles of the legacy production process,” Sharma continues. “It’s really all about putting the customer over the campaign. It’s also fundamentally about marketer and machine. As a marketer, you’ll get to work with an AI as a co-pilot that collaborates with you.”
For those marketers still working in campaign silos and relying on other teams to create customer journeys, AI promises much. As it does for those customers who are promotion fatigued.
But, the fundamentals of marketing will remain. As ASDA’s senior CRM and customer data leader Emma-Louise Birch said, sharing details of the retailer’s recent personalisation work, it’s “vital to set clear goals and metrics”, create a testing roadmap, review results with key stakeholders, keep the business engaged and “always celebrate every win”.