Over the past years, we have seen the growth of chat apps, social media, and business tools for sharing content and communicating better.
We have also noticed the general annoyance about the number of emails received each day, both in a professional or personal context with commercial communications.
Despite all of this: email is still thriving. In fact, according to a Statistica report, 269 emails have been sent and received in 2017. This figure should get to 333 emails by 2022, that is to say approximately a 23% increase in barely 5 years.
I have worked for several months in CRM marketing and especially emailing and I am currently working in a French data marketing agency. What I learnt is that, indeed, the way brands were using emailing is totally out of date. Emailing is not over yet though, and I am truly convinced that machine learning will very soon transform emailing marketing.
For me, machine learning will have an impact on four elements that are the pillars of emailing marketing. It should help marketers send communications to the right email recipients, at the right time. Most of all, artificial intelligence will provide a better knowledge of these recipients that are actually human beings: emailing marketers will be able to meet the needs of each individual, case by case. According to a PwC and L’Usine Digitale survey*, half of the 240 leaders interviewed are exploiting less than 25% of the caught and analysed data… A reality that might change thanks to machine learning.
A better segmentation to communicate to the right recipients
The actual strategy for most of the brands is to target people based on their very personal information like their age, the geographic area where they live, or their purchase history. The future of targeting is, in my opinion, based on the analysis of their behaviours. Machine learning algorithms will permit to create newly qualified segments, receiving customized communications according to their behaviour pattern.
A very interesting new tool to improve targeting is Tinyclues. This solution helps brands and retailers with huge customers database to sort out this amount of data. Artificial intelligence is able to predict who will be more likely to open, click and buy the product or service. To realize these predictions, Tinyclues is using unassigned customer data, like the name domain of a website address, the purchase history or the link the customer clicked on. The algorithm will then find correlations between the billion of data, mostly unstructured, and learn about it in order to propose a solution.
As an illustration, this short video explains what Tinyclues is doing :
Content: a better knowledge about how to talk to customers
With machine learning solutions, A/B testings on subject lines, body copies and images will not be useful anymore. The artificial intelligence tool will be able to determine which content will perform best in terms of opening, click and conversion rates.
Phrasee explains in the video below how its algorithm permits to generate subject line :
The right timing to send communications
One of the most frequently asked question in emailing marketing is “when should I send this email to my customers?”. According to me, the answer depends on the sector and the typology of clients. However, if a brand sends too many emails, recipients are more likely to unsubscribe. On the contrary, if a brand does not send enough emails, the competitors on the market will take the place.
Machine learning will figure out both the frequency and the timing issue by analyzing the customers’ activity history. It will enable to determine habits, time zones and downtimes in order to adapt to each people individually, according to their preferences.
Personalize the content
Improving the content can go further than finding the right subject line of the image. In order to maximize the results of a commercial email, artificial intelligence will help marketers determine what type of promotion will best perform for each individual (full price product, new products, discounts, free products, free shipping…). The probability to purchase will be significantly increased. Both companies and customers are winners: companies because they will sell more, and customers because they will have communications corresponding to their needs or their wishes.
As a conclusion, it is true that people are receiving too many emails. Commercial pressure is the reality. In order to differentiate, brands need to go further than the first step of personalization (like putting the first name in the subject). Following this objective, AI will help marketers sort out the available data to determine the best messaging, deliver at the best time and including the right offer for each individual.
Therefore, he next challenge for companies is to hire machine learning talent to implement those new AI tools. It will probably be harder for small brands: according to a PwC and L’Usine Digitale survey*, 44% of companies with less than 500 employees do not think about integrating AI in their project. For companies that are already using AI, the human factor is the first obstacle to the development of AI tools: 56% of interviewed companies list the lack of knowledge and 49% the lack of training.
This might in the end build a gap with huge companies that have the means to attract and retain highly qualified talent.