Pavel Pola is the CEO of Etnetera Activate and, a public speaker, and trainer.
Pavel Pola is the CEO of Etnetera Activate and, a public speaker, and trainer.
Jump to the topic you are most interested in:
I got into email marketing quite by accident 10 years ago, but I was immediately hooked. Mainly because it helps build long-term relationships with customers, is quite measurable, and covers a wide range of different areas (design, coding, working with data). My favorite topics are deliverability and improving campaign effectiveness.
From the results of the campaigns we can see that segmentation and personalization are the basis for improving the effectiveness of email marketing.
If you send everything to everyone, you may have some impact, but customers today are used to receiving relevant offers and if you send everything to everyone, they will get bored over time and either unsubscribe or stop reading or — worse — label you as spam.
For example, we see really big differences in engagement when we have a well-done subject line or send to a well-defined target audience. And for a campaign to be successful, you need to deliver the right content to the right recipient at the right time. And each recipient is also at a different stage of the buying cycle, you just need to speak to each one differently. The absolute foundation for segmentation is an RFM (recency, frequency, and monetary value) analysis, for example, which gives you a basic view of what your database looks like.
I’m not saying that you need to personalize and segment every campaign. On the contrary, we recommend our customers send the best offer to all their customers from time to time — about once a month, depending on the segment. By the way, this helps to eliminate the risk of recycled spam traps (if they handle bounces correctly).
We recommend identifying the appropriate segments and appropriate places for personalization and leveraging them. For example, it’s a good idea to track the number of campaigns delivered to specific customers — i.e., so that if you over-segment, you don’t end up with someone getting no campaigns and then you have customers who get dozens of them.
It’s about finding the right balance, A/B testing, proper reporting, and that kind of healthy approach — e.g. focus on the segments with the most business potential, the campaigns that can deliver the most benefit. For example, for one drugstore, we identified a segment of customers who regularly buy children’s products but are not members of their loyalty program that offers discounts on these products, so they can save money. If you send the same campaign to people who don’t have children, they are more likely to be put off.
First of all, I would say it’s not just about new technologies. AI alone won’t save us, it’s good to have standard / known techniques and practices implemented as well. As I said before, everyone should have at least RFM segmentation, work with engagement reports, etc.
What’s interesting, for example, are the various cluster analyses that help identify potential customer segments based on buying behavior — this worked before AI.
AI/ML (machine learning) can help with finding new outreach opportunities, recommending the “next best offer” for specific customers, predicting customer behavior, etc. But you need to incorporate all the omni-channel data — i.e. what the customer is buying, how they behave on the web/mobile app before and after purchase, etc. For example, we recommend tracking what the customer is looking for on the web and then helping them make a choice so they don’t go looking to a competitor.
What we have tested — and it works — is when I recommend to a potential churn customer a product that they know and regularly buy and is currently on sale — ideally in the subject line. Then I like to segment by buying cycle and tailor my communications accordingly — e.g. to encourage a second and subsequent purchase. I also really enjoy finding segments that would be good to reach to encourage loyalty — just based on their buying behavior, for example.
Recently, we’ve seen a big increase in robotic interactions — most notably in email opening, where more than 50% of emails are now opened by bots. Apple Mail Privacy Protection is largely to blame for this, but we’re seeing an increasing proportion of robotic opens in B2B as well.
This is a result of more and more customer privacy and protection from unwanted emails (spam, phishing, etc.). This makes it problematic to evaluate campaigns based on these skewed metrics — so it’s important to choose tools that can distinguish these robotic interactions from human interactions and therefore provide better data. Tools that don’t do this can, unfortunately, cause, for example, automations that respond to open/click to stop working, poorly evaluated A/B tests, you’ll have skewed recipient engagement data, etc.
Ideally, use tools that can detect them. Then, in the second row, don’t rely on those biased metrics, but evaluate e.g. actual conversions or the entire customer journey. It’s also a good idea to combine multiple different ways of measuring and connecting data from different systems (e.g. web, mobile apps, transactional data, etc.).
AI is definitely influencing and will influence email marketing. Even on both sides. In the end, it can be a “battle of the bots”, where one prepares targeted content (recommendation engine) and the other evaluates if that content is relevant to that user (spam / relevancy filters).
AI is already helping with segmentation and personalization today, but we are also seeing tools that, for example, design a campaign structure based on a web page with an event invitation that is just perfect. It used to take tens of hours of time to prepare such campaigns, it won’t take long and you’ll have the whole thing done within an hour. As I wrote above, the perfect email is one that offers the right content to the right customer at the right time, and AI can help with that quite significantly.
AI is already being used quite a bit in email marketing today and will be even more so. Firstly, on the recipient side (inbox service providers such as Gmail), where they already use AI for spam selection and for sorting emails into relevant folders (Promotions / Later / …).
Those who have watched e.g. Apple or Google presentations know that it won’t take long for AI to help extract relevant information from email conversations and offer relevant actions straight to customers. So I can even imagine a scenario where the AI will evaluate which offers that arrive in the customer’s email are the most beneficial at a given moment — e.g. when I need to restock the fridge.
On the sender side, I don’t think it’s that widespread yet, with the exception of recommendation models. For content creation, for example, AI is a good tool, but it will take time before it replaces human work. There is still a slight language barrier — even the best models today cannot handle all possible languages, although progress in this area is also very noticeable.
We use it quite regularly. AI helps me brainstorm ideas. It helps me solve various technological tasks — e.g. write an SQL script, find bugs in code, find and fix inconsistencies in data, recommend the next best product, find clusters of customers based on their commonalities, generate images for campaigns, provide A/B test alternatives for subject lines, translate texts into multiple languages, rephrase texts in campaigns to be more catchy, etc.
We use a variety of tools with respect to a given use case. You still need to take into account that we work with customer data, sometimes quite sensitive, and you can’t just put it into freely available tools that you don’t know what they do with the data and whether they are GDPR compliant, for example. So for some clients, we have AI that runs only on their servers, so they are in full control.
I’m a big fan of TED.com — their recommendations for videos I should see pretty regularly hit what I really like or inspire me.
Then I have a few local resources that are my favorites, but unfortunately, those are only in Czech. Unfortunately, most of the other content isn’t that relevant, so I wouldn’t recommend them here.
My favorite phrase is “undelivered email cannot sell”. You can have the most perfect strategy, the best tool, the best AI on the market, etc., but if you don’t have a well-built foundation for perfect deliverability, you’ve de facto thrown everything else out the window.
In terms of success, I mentioned this above — every marketer should ideally “put themselves in the shoes” of their recipient and think about who is the ideal recipient of that particular message. Because what determines success is if you can deliver the right content to the right recipient at the right time (and I’ll add using the right channel that allows them to respond to the message).
