Generative Chatbots in Marketing

At Creative Oak, we have started making chatbots to help our clients' marketing. Why are we, you might think? We do this because there is great potential in exploiting the latest possibilities in generative artificial intelligence when it comes to customer management and contact. We have made a little post about it here!

What is a Chatbot?

You probably know what a chatbot on a website is — you've probably used one yourself (and maybe you have one lying around on your website already). But just to get everyone on board, we'll go right through this.

A chatbot is really just a small window on your website that customers can write to in natural language and that provides information about the services you provide. The whole point is that instead of having to navigate your website (which can be more or less difficult), customers can go one place -- your chatbot -- for answers. Of course, there are different chat solutions — sites that require in-depth technical support often have people on the other end who have deep knowledge of products and technology and can therefore help the customer.

However, if you do not provide technical support in your chat, it is typically not necessary to have people in the chat. Instead, you can use a chatbot. Most former chatbots are what we call “rule-based”. It really just means that their creator has programmed some predefined responses into them that they come up with if the customer writes a message that matches what they are looking for. It can be smart for a lot of things (we'll look at the pros and cons later) -- and they're still widely used around the internet.

There are also some tasks they are very bad at solving. They are typically not very good at seeing what is asked of them, and they can do nothing but what they have been asked of! Therefore, you need to be able to guess in advance everything a customer could come up with to ask the chatbot about, and that can be difficult. Fortunately, with generative AI, we can now remedy these problems.

Chatbots History

Although, we talk a lot about chatbots now, after ChatGPT has stormed forward, it's not a new technology at all. One of the earliest and most famous chatbots in history is Eliza. It was developed in 1966 by Joseph Weizenbaum at MIT, and functioned primarily as a simulation of a psychotherapist, more specifically in the role asking questions based on the user's previous statements. Its best-known script, DOCTOR, used simple algorithms for pattern recognition and rephrased the user's statements as questions that were presented back to the user. Although Eliza was quite primitive and had no real understanding of conversation, it was able to mimic human conversation well enough that many users attributed human-like emotions and motives to it. Another famous chatbot is Anna, IKEA's virtual assistant, launched in the early 2000s. Anna was designed to answer customer queries on IKEA's website, help find products and provide information about stores. The chatbot, like Eliza, used algorithms to recognize keywords and phrases to provide relevant answers.

‍New possibilities with Generative AI

Generative AI, which knocked the door on every business with ChatGPT in 2021, is laying the foundation for the chatbots of the future. This solves many of the limitations inherent in rule-based chatbots.

Chatbots based on generative AI can adapt to a much greater extent to what the customer writes and how the customer expresses himself. Because they are based on large-scale AI language models, this type of chatbots have a built-in understanding of language and can dynamically generate responses no matter what the customer writes to it. However, this also presents some challenges. It's harder to guarantee what the chatbots actually answer -- where before all responses were rule-based and predetermined, the new ones are generated “on-the-fly,” making it harder to guarantee a specific response.

Fortunately, we can set up some framework on which, what the generative chatbots base their responses on! We do this in Creative Oak's own chatbot, among other things, by allowing the chatbot's responses to be based on, and limited to, the content we have on our website. That way you get the best of both worlds — our chatbot, Oakly, can respond to everything between heaven and earth, but it also has direct access to the content on our website. For example, ask them what services we provide!

Of course, there are advantages to a rules-based chatbot — which is why they are still widely used. Briefly, the pros and cons of different chat solutions can be said like this:

Generative AI

Advantages:

• Provides an automated solution that works 24/7
• Can answer all types of questions (including in different languages)
• Can have direct and dynamic access to specific professional knowledge

Disadvantages:

• It is more difficult to guarantee a specific answer to a question (which is why they are rarely used for very critical tasks, such as in public administration)
• Still have ignorances, and shouldn't be used as a search engine
• Can't immediately (as a human being) ask colleagues for advice or directly escalate a problem for the customer

Rule-Based Chatbots

Advantages:

• These chatbots are programmed, and are therefore much faster at providing answers
• The rules guarantee correct and predictable answers
• Is generally the cheapest option

Disadvantages:

• It's harder to answer diverse questions
• Pre-programmed responses only
• It's harder and more expensive to update these chatbots with new response options

Chat solution with people

Advantages:

• There is a real person behind the screen
• Gives a better impression to the customer that they are talking to a real person
• Even generative chatbots are still far less intelligent than humans

Disadvantages:

• The price is much higher than that of other chat solutions
• May (rarely) be available 24/7

Price level

When purchasing a chat solution, the decision is largely about time and money saved versus the quality and value delivered to customers. Here, in particular, you need to consider what challenges your company needs to solve and what requirements you have for how this is done.

Generative Chatbots lie in a kind of sweet-spot between rule-based bots and a chat solution with humans. If you do not need in-depth technical knowledge, but need a dynamic chat solution that is more accessible to different types of customers, generative chatbots are a really good solution for you!

You can see prices for our generative chatbots hereto.

Generative Chatbots as Marketing Tools

As we have already mentioned, generative chatbots are a powerful tool for integrated marketing, especially when it comes to customer management and contact. Because a generative chatbot can deliver dynamic content that responds to all types of inquiries while always being online, it can help give customers the impression that your company is always available for help — even if you're on vacation or sleeping.

In addition, through the chatbot, you can collect data about what your customers are interested in, what they find difficult to find on the website and what questions they have about your services, without you having been in direct contact with them. In our chatbot solution, of course, we collect data for you so that you have access to and overview of it.

Generative Chatbots for Internal Use

Another use of generative chatbots that larger companies often make use of is internal chatbots. Because they can have dynamic access to various resources — including internal ones, generatives are a good way to create an easy entry into many different documents and information that can be difficult or cumbersome to access.

In addition, generative chatbots can also be used as a sparring partner for various creative or time-consuming work. For example, in Creative Oak we use an internal chatbot to help write social media posts and blog posts like these.

Have you become interested in hearing how we can help with a chatbot for your business? You can find an overview of our solutions hereto and you are always welcome to contact us.

‍