With the advent of AI in Marketing Automation, user personalization has taken the digital world by storm. Yet, digital media houses and publishers haven’t really taken complete advantage of that.
Although leading publishing houses like Forbes, CBS Interactive, Venturebeat, and so on have taken initiatives to personalize content for their users, most of the other publishers don’t do much about user personalization yet.
Today media houses and publications strive harder than ever to be discovered by users on social media. Most of these media houses, garner a third of their web traffic via Facebook or Twitter. This is one of the best examples of how powerful a tool social media can be. But, at the same time, it also shows how publishing houses are losing a hold on their readers. Here’s an article on how facebook’s new “Instant Articles” feature has the potential to eat into the revenues of publishers.
Be it any sort of website, nothing is more valuable than direct traffic. Today direct traffic equates proportionally to brand value as well as trust on the brand. The crucial point here is that Publication houses thrive on brand value and trust. WSJ, CNN, Huffington Post, I can keep counting, but you get the point. If you know the name, they can be trusted; if you don’t, then they don’t exist.
Now let’s come back to our topic, “Publication and User Personalization.”
How is it better than other traditional methods of engaging readers?
Why A/B testing or rule-based targeting are not better than AI personalization for publishers?
That’s because Artificial intelligence for publishers has been added to the mix. Didn’t get my point? Don’t worry, we are going to discuss everything.
So let’s dig in.
AI Personalization vs Rule-based Marketing
Today, most of the media publishers, blogs and websites promote their content using rule-based marketing.
In layman terms, rule-based marketing is a process where you serve content based on rules, conditions and triggers. For example, if A follows Rule 1, show him X. Or, if A follows Rule 2, show him Z.
Traditional rule-based targeting is based on clear-cut scenarios. If you follow such a strategy, you know who to target and what content you need to serve to them.
Rule-based targeting can be very effective if you know your high-value readers and you are focussed on optimizing for their experience.
Sounds simple right? Find your important readers, set rules and provide them content. Alas! If only it was this simple. Because, every now and then a group of diverse users will appear that do not fit into any pre-conceived bucket.
Here is an example:
- First step is to understand the opportunity at hand, for each segment of readers, user or customers you want to target. Like are there enough readers in this segment who will subscribe to our paid channel?
- Then, if opportunity exists, you need to set up flows and funnels. A lot of manual work has to go in. Setting up rules,magnets and optimizing it, all of them have to be done manually.
- And then you need to wait, variate and wait some more, until folks start to trickle in.
But today, customer’s preferences changes by the minute. One minute they are reading about the U.S. elections, the next, their interest peaks towards the Syrian war. You can’t set rules for things that change on the fly.
If the example does not clear your doubts, I hope this will.
Limitations of Rule-based Targeting
Under rule-based targeting, content is delivered based on conditions and rules. Visitors are assigned a unique identity (a cookie or tag in most cases). These cookies identity is used to track their web journey on the host’s website.
This tactic is suitable for websites with less than 100-200 pages or smaller website, where the user can’t have a diverse behavior pattern. But as the website grows, the rules become complex and unmanageable.
There are two major limitations that make rule-based targeting ineffective for large-scale publishing houses:
- Manual input: Rule-based targeting does not have the ability to learn on it’s own. For every action, a group or an individual has to set various rules for triggers to work. Any admin or marketing team has to devote a lot of time towards monitoring, defining and implementing rules, making changes to them on a regular basis to fit in a new spectrum of customers. In the end rule-based targeting consumes a lot of your time.
- Not enough data: Since rule-based targeting is based on assumption and scarce data points like demographic, age of user, location, a few landing pages, etc, it’s not enough to deliver a truly personalized experience to the user. Because of such limitation your ROI and campaigns always fall short. There will always be a bucket of users that will show different interests and backgrounds that will be totally off our consideration.
AI Automation vs A/B and Multivariate Testing
A/B testing or mutlivariate testing is considered to be a major part in user engagement and digital marketing. Ask any marketing guru about their marketing strategies, he or she will surely mention A/B testing.
But A/B testing again has a lot of manual input to start with. Here is a basic example of how you can put in place a simple A/B test.
- First step is to pick up parts of your website that can be targeted for A/B testing. Identify pages and post, put them in a collated list.
- Then you’ll need to get an in-house web developer or subscribe to a 3rd party A/B testing tool like VWO or Optimizely to actually make changes.
- Once changes are done, you’ll be running tests for 2-4 weeks, to study the impact the experimental version makes.
- After ending the experiment, you’ll have to analyze the results. Use the data, to make changes, that too, only if the tests worked in the first place.
- And then, repeat the whole cycle again.
Now let’s look at the limitations that A/B testing possess, apart from the example.
Limitations of A/B Testing and Multivariate Testing
- A/B testing is slow and expensive: The cost of A/B testing is pretty expensive and the whole process takes a lot of time to yield results. We are talking at least 4 weeks for a single test to conclude. One has to decide whether they can build an in-house A/B testing platform or are they willing to pay for A/B testing softwares (VWO or Optimizely). Both end up becoming a costly affair. Secondly, there are 2 more factor deciding the cost of A/B testing.
- The parts and elements to target on a certain website. If it’s a site wide overhaul, we would need a service that is 100% compatible with your website or in most cases, a dedicated web developer that just looks at A/B test requests.
- Actually creating those alternate versions. These alternate versions might not perform as well as you expect them to. If that happens then you will have restart from scratch, and this leads to a fall in your ROI.
- A/B testing is always an ongoing process: At a point, you will start noticing that your experiments are working. You will see Version B is working better than the original version, Version A. But that does not mean it will work the same way in the coming 6 months or more. User preference will keep on changing and you have to adapt accordingly. Therefore most experts in A/B testing speak about continuous experimentation. Like, if Version B is doing great, let’s build up on it and release version C and observe it is better. If yes, than repeat the process, if no, than come up with a better version. It’s a endless loop that you’ll trap yourself into.
Such limitations is why we need to move beyond traditional marketing techniques and delve into the tech that removes all these limitations and lets us concentrate on things that really matter, like producing higher quality content.
Artificial Intelligence Personalization for Publishers
Fortunately with the use of AI and smart-automation these things can be handled automatically.
Advances in the artificial intelligence sphere is expanding fast and finding uses in a wide range of industries. And one of the industry that benefits the most out of AI and NLP (Natural language processing) is publication/media houses.
In a more broader spectrum, all the niches which build their business on content consumption has a great scope of growth and customer engagement with the help of AI.
Today, AI is not only used to develop chatbots or personal assistant but also as an engine that keeps a watch on user activity, behavior and engagement.
- By monitoring user activity, AI engine mimics a marketer, understands what the reader likes and dislikes. From that information it makes recommendation and amendment accordingly. This always leads to higher engagement and retention rates for media houses as well as individual publishers.
- It also eliminates the manual intervention part and takes matter into it own hands. All sort of contents that is served to the user has the potential to be personalized via the AI. Whether it’s an email blast, a push notification, a chat message, SMS, etc, everything can be personalized directly via AI.
- Even Rules-based targeting and A/B testing is done on the fly. And because an AI engine works 24/7 on top of the media house, any sort of changes or recommendations are made swiftly.
As a person who works in a media house, you know how much time it can save you, if you are backed by such an AI engine.
For example, Boomtrain’s own AI engine, is powerful enough to decide what channels and medium to target while interacting with users. Not just that, it optimizes and personalizes content on the fly on your website as well as content on others channels (email, push, in-app). Communication frequency and optimal delivery times are also calculated on the fly to improve opens rates and CTORs for your campaigns..
With the world getting smarter day by day and the world’s largest brands like Apple, Google, Amazon, Microsoft, betting big on AI and NLP, we are getting closer to a day where all our fixes will be handled by AI.
Bonus: Did you know Boomtrain specializes in AI Personalization and market automation for media houses. You can check it out here.