The AI personalization no Product Manager is talking about
A trending technology doesn’t have to be implemented if it doesn’t need your product. AI is a great trend. It doesn’t mean that you implement an AI in your product. But how companies are using AI and how AI is shaping the future of products should be learned.
Companies like Careem are hiring for AI. Not only the engineers, but also the product managers. Before implementing AI tools in your product, you need to understand how they will improve it.
Engineers look at these tools as time savers. In some cases, these tools are saving costs. They are excited, so they want to try them.
It is the job of product people to look at AI from the consumer perspective. They need to find the use cases that are truly helpful to the consumer and business needs.
For most consumers, AI is just an add-on. They take AI just as a generator of some content to play with. They imagine AI in the shape of a chatbot or an assistant. Now they also see that the word ‘AI’ is used as a marketing slang.
In my recent interviews for hiring product managers for new startups, I have realized that most of the product managers are thinking of AI mainly as a co-pilot. They love ChatGPT and tools that generate ideas and media for them. There is no doubt that these tools have become part of our daily toolkit.
And here comes the bias. Product managers use these tools for their work. That’s why they want to implement these tools in their products. They think that their products should have a similar experience where an assistant should help them solve problems and complete tasks.
I am not against AI. I am not against using ChatGPT or Grok. I am against associating AI with just a particular set of tools. It is limiting your imagination.
At the same time, the engineers in the same organizations are talking about the other AI tools that can be ‘game changers’. They want to implement those tools that can simplify complex tasks. Tools like Next Best Action by Amazon are excellent. They are some of the amazing engines that help generate the recommendations for the users.
In the search for adding personalization to the product, we have found that more than 60% efforts have failed in recent years. These products were stable; they were already in the market. Transactions were being made through these products. Startups were doing their business. But to improve their product and company, they wanted to implement AI personalization. And most of the time, the results were terrible. Consumers hated them.
People have experienced bizarre recommendations. For example, you buy an iPhone through a platform and the next day you are offered to get another iPhone at a cheaper price, while we all see that an iPhone is something most people would buy once a year, max.
It is not a failure of technology; it is a failure of product management. They are failing to understand how these tools should help in their products and what a user will see in different cases.
Some apps want to be a “super app“ or ”everything app”. They are in the best position to create personalized experiences.
If you opt for a personalized experience, it won't be very easy. It will never be simple. Otherwise, you will be trying to sell me an iPhone every day.
Things can be suggested based on location and time. See the patterns in them and see what they mean. Consider their history of purchases, categories, and especially the sub-categories, things in the cart, previous feedback, and returns and cancellations.
At the same time, they should use the information from their existing tools, such as wallets and subscription status. And then based on these data, they should align suggestions with their business goals, like sales, price drops, packages, or special offers.
Building this system is not easy, it can be extremely complicated. You are not dealing with data; you are dealing with a real human. You cannot factor everything to create a recommendation at every time. You cannot ignore most of the important events as well. That’s why you should start small.
Take the example of driving assistants in cars. The car industry has evolved over the past three decades. They have started focusing on one case, like automatic emergency braking. Slowly added more things like lane assistance. And over the years of research and experiments, now we have autopilots. Still not perfect, but exceptionally reliable.
The same approach needs to be taken for your products. Due to the data and the AI tools, the process can be a lot faster today. It won’t take decades. But the process should be the same.
A straightforward way to start this personalization is by asking the user what they want as they use your product. Using these data points, see how these personalization tools can help. They shouldn’t be recommending things to everyone, but just to a specific set of audiences.
Start by giving users control over their experience. The user shouldn’t see a blurred line between what they want to see and what you want them to see. This is one of the reasons why people are addicted to WeChat. The level of control it gives to the users helps them build confidence in the app.
When it comes to sharing more recommendations, many consumers may experience choice paralysis. You show so much information to these visitors and make it difficult for them to make decisions. A simple user experience is always the best one. That helps the user easily navigate to the transaction.
You should use this data not for personalized recommendations. Use it on different UI elements to understand them better. Try different icons for different segments. Try a different language or tone of message. Play with colors.
The more you understand your audience, the more you will understand how to tailor your product to their needs.