Personalization: five reasons why it does not work well
We can map customer behavior online and derive insights from it to apply personalization. But the dividing line between smart campaigns and irritating or even scary content is thin. How do you stay on the right side? A recent study by Inmoment shows that three-quarters of consumers often find personalization rather scary
We can map customer behavior online and derive insights from it to apply personalization. But the dividing line between smart campaigns and irritating or even scary content is thin. How do you stay on the right side?
A recent study by Inmoment shows that three-quarters of consumers often find personalization rather scary. Half of the companies are aware of this and even agree with it. What consumers are about include:
- giving a phone number when they buy a product;
- retargeting on websites that have nothing to do with the retailer;
- systematically receive mails about the product that the consumer has put in the shopping cart, but has not bought;
- Google asking if they want to review a place where they have just been;
- apps that request access to contacts, photos, et cetera.
Does this mean the end of personalization in marketing? Of course not. There is nothing more beautiful than a news feed that only shows relevant messages for you or your favorite store that always comes with the right offer. How often this really works, you can see in your own e-mail inbox. These are the five main reasons why personalization does not get off the ground:
1. Machine learning is not the same as personalization
Personalizing is not a matter of letting go of an algorithm on your data and then everything goes automatically. Machine learning is the engine that makes personalization possible. But you also need to be able to collect current data and process it quickly, for which you need a safe and flexible infrastructure. Before you begin to personalize on scale, the platform, the data and the user feedback must be integrated. Then you can get started with algorithms.
The only way you can apply accurate, interactive and context-driven personalization is to continue to measure consumer feedback on your products. Only in this way do you understand your users, their preferences, hesitations and actions and you can respond to them.
2. Go for the largest segment
Personalization is not a good thing for everyone. What often happens now is that if 75 percent of the users responds well to dynamic content and 25 percent do not, the dynamic content is labeled as a success and is maintained. But you will lose a quarter of your users.
The solution: your systems must remember users. This way you can offer everyone an experience that appeals to them. The 25 percent who are not happy with the dynamic content on the website, for example, you can show the standard version so that you do not lose them.
3. Let go of design for data
The value of personalization does not matter what it looks like or works, but how good the insights are so that you can surprise the user with exactly the right content at the perfect moment and through the best channel. You can derive these insights from machine learning, but they can also be simple 'if, then' business rules.
4. Making assumptions based on one purchase
It happens to everyone: you buy a gift for someone else, such as LEGO for a niece or a PSV shirt for your brother. You do not play with LEGO anymore and you are not a football fan, but the salesman will still email you with recommendations for toys or PSV paraphernalia.
The logic behind this is understandable: promotions are personalized based on your purchase history. The only problem is that the wrong conclusion is drawn about the interests of the customer because only one data point is used. The solution: do not rely on the first interaction but base the personalization initially on a number of personas or choose a data point that is important for personalizing the content, such as the location of the customer. Then ask in the welcome e-mail if the customer wants to indicate what he is interested in, so that you do not bother him with content he does not need.
To personalize relevantly, you need to understand the context of your products and how they fit into the context of your customers. For that you need demographic and location data, but you also want to know what the customer is doing (commuting, working, relaxing), where he is in the customer journey, the motivation to look for a product, earlier experiences with your company, but also the season, the day and time, the role of your product in the life of your customer, et cetera.
5. Communicate too intrusively
Personalization is meant to increase your relevance but in practice often leads to irritation at the customer. Two common examples: endless retargeting while the consumer has already bought the product and with every website visit a popup with the question to become a member of the newsletter while the visitor has already indicated that he is not interested. Here too, the solution lies in recognition: if you analyze the behavior of the visitor, you can assess the chances of a purchase and tailor it to retargeting. And that pop-up will of course not show you if you know that the visitor is not interested in it.
Recognize anonymous click behavior
Many of the reasons described above as to why personalization does not work (optimally) can be traced back to an incomplete customer view. At the time of personalization, not all customer information is available for relevant communication. Often the combination between anonymous online behavior (orientation on websites) and offline behavior (CRM, purchases) can not be made. However, there are possibilities to solve this problem. By registering trackers (e-mail GUID, digital fingerprinting, cookies, login IDs) it becomes possible to recognize the customer behind the anonymous click amount.
Schematically this works as follows:
A visitor visits a website and clicks on links. All clicks on the website, including the defined trackers and tags, are stored in the code in the Google Analytics script based on definitions that are stored in the code. Via the Google Analytics API the click data is retrieved to your own data environment. Here matching takes place between click data and customer data. This matching allows (a part of) the click data to be assigned to your individual customers and prospects. The previously anonymous click behavior becomes (partly) identifiable as a result.
Finally, is identifying your visitors AVG-proof? Of course, you must state in your privacy statement what you record, what you do with the collected data and you must ask the customer for permission as soon as he or she visits the website. But if you communicate honestly and transparently about this, you can try to identify your visitors without any problems.
More relevance, less irritation
Recognizing your website visitor gives you opportunities to enrich customer profiles by combining orientation behavior (on websites) with customer profiles and transactions (CRM). With this customer knowledge you can further optimize the communication towards your customers, prospects and visitors. With the result: more relevance and less irritation.
Header photo: Graeme Nicholl on Unsplash