Curious about a product recommendation chatbot?
Well, you’ve landed on the right page!
You see, at Landbot, we are all for stopping the nonsense of thinking of bots in the limiting terms of customer support. That’s a thing of the past.
Given their immediate reaction time, chatbots are quickly becoming the go-to strategies for both sales and marketing. In the light of these developments, we put together this article to help you find the value and opportunities in using a chatbot for one of the essential sales processes — product recommendation. Better yet, we will teach you how to build a product recommendation chatbot of your own!
Personalized Product Recommendations: Are They Worth the Hassle?
eCommerce is one of the most competitive markets out there. Truth be told, it was competitive before the pandemic hit… the post-COVID status is closer to outright bloodshed and survival-of-the-fittest types of scenarios.
The preference for uniqueness, personalization, and speed wasn’t news even before. However, today, it’s a straight-out demand. No middle ground. No understanding on the side of the customers. Businesses are asked to deliver quality products and services with speed and efficiency, all the while treating each customer as an individual.
The reports and statistics shout the same truth:
- Accenture reports that 91% of customers claim they are more likely to shop with brands that provide personally relevant offers and recommendations;
- A survey by SmartHQ shows that 72% of shoppers say they ONLY engage with personalized messaging
- According to a study by Epsilon, 80% of consumers are more likely to shop with a brand that offers a personalized experience;
- Statista states that 90% of consumers in the US think of marketing personalization as very or somewhat appealing;
You fight to keep up, or you start lagging.
It’s not much of a choice when 89% of online businesses are already investing in personalization. Retail is the leader! 79% of businesses in the retail sector specifically are investing in a variety of personalization Tools — more than any other industry.
So, if you are selling products or services online, you better start considering ways to apply real-time automation that doesn’t trample your personalization efforts into the bones of your conversion funnel.
Why Choose a Product Recommendation Chatbot?
Personalization can take up many forms.
However, chatbots have the crucial advantage of real-time reactions, adaptability as well as the added comfort of conversational format. All three factors combined make building relationships and offering individual treatment achievable regardless of the size of your business.
Bots use the power of conversation to make customers feel comfortable and quelch their need for “instant satisfaction” by tending to their queries immediately.
When it comes to product recommendations, they eliminate the necessity for shoppers to scroll down an endless product page to find what they need. Instead, they use the conversation as a filtering system, bringing the product to the customer.
What’s more, chatbots can also be designed to remember previous interactions and purchases to offer personalized recommendations based on the buyer’s preferences and chat history. Having somebody to advise shoppers along the way helps bring them to the final stage of the sales funnel more smoothly.
You might think that customers don’t like sharing data, but providing a name or preferences feels much less intrusive within a conversation. More importantly, it allows you to observe, in real-time, in which part of the conversation users lose interest and fix the flow of the dialog accordingly. A product recommendation chatbot also detects how shoppers react to new products, pricing, discounts... Data that usually gets lost on a static website.
Clicks give you just the numbers; conversations enable you to dig deep and analyze the sentiment behind the actions.
How to Build a Product Recommendation Chatbot
Many businesses still shy away from bots. Some of the most significant factors influencing them to stay away are the notions of expense and complexity associated with chatbot implementation.
However, the times have changed, and the avalanche of no-code tools crowded the digital market, making it easy for small and medium businesses to implement sophisticated conversational solutions without coding.
Landbot chatbot builder is one of those no-code chatbots development tools — an intuitive chatbot flow builder and messaging manager that enables you to create website chatbots, Facebook Messenger bots, WhatsApp chatbots, as well as custom solutions.
This tutorial shows how to build a product recommendation chatbot for your business without coding or the need to employ complex artificial intelligence solutions.
1. Decide on Your Product Recommendation Chatbot Type
Before anything else, you need to decide which kind of chatbot you want to build.
You need to consider:
Communication Channel: A chatbot can assist your customers on your website, inside your native app, or on their favorite messaging apps like Messenger or WhatsApp. The channel you choose will dictate the visual freedom you can exercise in terms of design and product presentation.
Base (rule-based or NLP): You can either choose an AI chatbot that uses natural language processing to understand user requests or opt for the more structured rule-based chatbot. Each bot approach has its own advantages. NLP bots provide a more natural conversational experience but are harder to create and tend to fail more often. Rule-based bots give you more power over user experience as they lead them down a pre-determined path. Although, there’s also an option to combine NLP elements with rule-based ones. In such case, for example, you can leave the shopper a free had to ask any question (e.g., Do you have any vanilla-flavored protein supplements?). The NLP software will analyze the question and direct the customer down a rule-based path designed to address this topic.
Recommendation system: Not all product recommendation chatbots are the same. You can model them in different ways based on your needs.
- The simplest of approaches is for your bot to simply serve as a conversational interface. Instead of navigating your website, the bot can simply offer options and guide them to their personalized offering in a speedy manner. After a shopper selects a product, you can start again by recommending similar or “commonly-purchased-together” products.
- Another option is to design the bot to ask a series of qualifying questions that allow the assistant to present customized recommendations. For instance, a sports-wear bot can ask questions about gender, preferred sport, favorite colors to make a recommendation. This bot is handy when you are selling customized products only. For example, BuddyNutrition built a recommendation bot that assessed their customers’ health, fitness, and nutrition habits and proceeded to create a personalized nutrition package on the spot!
- The third option is allowing the shopper to ask questions freely. However, in this case, the use of AI and natural language processing is inevitable.
2. Create Your Account & Access the Canvas
In this tutorial, we will be using Landbot builder to put together a basic product recommendation bot. If you are not signed up yet, register for free and enjoy the free trial.
After registration, you will be welcomed by the central dashboard, from where you are only a click away from accessing the bot builder. There are a few different ways you can go about it:
Landbot interface works on the concept of conversational block connected by arrows which you can create and move around as you wish in a drag-and-drop fashion.
For the purposes of this tutorial, we chose “The Honest Amish” beard products brand as our digital guinea pig. So, the first step to start is by redesigning our welcome message.
3. Create Welcome Message
Click on the message field to change the text and on the little EDIT icon in the GIF to change it to another GIF or image. Of course, the image/gif is not compulsory, and you can get rid of it if you deem it unnecessary. However, visual elements do give an edge to the conversation.
Next, we decided to ask the shopper for their name to make the conversation more personal. Again it’s not necessary to know one’s name to suggest a product, but it showcases an essential feature of the builder - variables.
Think of a variable as a category name that stores a specific category of information. This way, you can make your bot leverage user responses in real-time by implementing the variable in the dialog:
You can check how it’ll work in practice by clicking first SAVE then PREVIEW in the upper right corner of the builder:
In the “Name Block,” the variable is set automatically to @name. That is just to make your job easier. That is why the builder offers different types of question blocks - each comes with a preassigned variable to save time (e.g., email block has preassigned @email variable). However, you can create any new variables you deem useful or important.
The important thing is, you can use variables to categorize and export user data to either a Google Sheet or any other CRM tool you are using.
4. Present Shopper with Options
There are two ways you can present your products and product categories. You can use the traditional BUTTONS block:
Which will take this form on the frontend:
Or, you can use the “Picture Choice” block, which was designed with product showcase in mind. It enables you to display product categories in an image carousel. As you work on creating the product carousel in the backend, you will be able to see a preview of each new “cart” you make:
Better yet, by clicking on the “Extra Options” of the cart you are working on, you can add more information about the product or product category you are displaying:
In the front end, the categories will appear as follows:
As you maybe have noticed, you can directly connect the images with the category/product URL so the user can leave the conversation and go directly to the relevant category/product.
Alternatively, after the customer selects a category, they are presented with the products within that category. Here, the same principle applies. You can let shoppers proceed to the product page, offer more information on the selected product or even let them add the product to the cart.
5. Upsell and Cross-Sell
The best about product recommendation chatbots is that the conversation doesn’t end with the potential customer selecting one product. When the visitor shows interest in a product or decides to add anything to the cart, the bot can react immediately by suggesting other products or better versions of the selected products:
6. Leverage Integrations
For your product recommendation ecommerce chatbot to work seamlessly, it needs to be integrated with your eCommerce and the tools you use to manage it.
Like flow building, Landbot significantly simplifies the integration process, making it accessible to no-coders (or making it really fast for developers).
One of the most popular integrations is Google Sheets, which, when set up, allows you to insert, update, or retrieve data from a Google Spreadsheet in real-time.
You can learn how to set it up in minutes:
You can store and update information with users every visit, personalizing every exchange. In practice, if your bot collects the user’s name and purchases storing them in the sheet, the next time the bot identifies the user (through cookies or other identifying factors such as email), the conversation can pick up where it left off.
Though that is just the tip of the iceberg!
You can integrate your bot with Airtable, SalesForce, Google Analytics, or use Zapier to integrate it with any tool, including product and CRM databases.
7. Use Conditional Logic for Hyper-Personalization
Using conditional logic is another smart way to avoid complex AI integrations but still create a hyper-personalized experience.
Conditional logic is based on the simple:
- IF this is TRUE, then answer A
- IF this is NOT TRUE, then answer B
The good thing is, Landbot builder makes conditioning simple so you can create more complex conditional chains without much effort. (BTW: Conditional chain is a condition based on a series of answers)
Based on the answers, you can suggest a specific product that fits the requirements. For instance:
IF Beard Length under 1 inch AND dandruff status: yes AND beard dry THEN Product X
IF Beard 2-3 inches AND dandruff status: no AND beard dry THEN Product Y
You can repeat the principle for each instance. Having “consulted” their needs with your bot, customers are more likely to be confident in making the purchase. It’s the equivalent of talking to a real-life shop assistant in a local store.
8. Design & Choose Format
Last but not least, your bot should represent and serve your brand in a way that suits your image as well as drives sales. Hence, you shouldn’t underestimate choosing the right design and format.
To adjust design elements such as bot title, avatar, font, color, background, etc. Save your bot and click on the “Design” option in the header menu. Here you will have options to select a design template or customize the bot as you see fit:
To control the format of your bot, click on the “Share” option of the header menu and select one of the four formats available:
- Full Page
- Live Chat
To embed it on your site in the selected format, all you need to do is copy-paste the code on your website and tadaaaaa! It’s done!
What About Product Recommendation on WhatsApp?
A website bot is cool mainly because of the absolute power it gives over the features you can include within the conversation. However, when you decide to create a product recommendation bot on WhatsApp, you will be confined to a very text-oriented interface.
But don’t worry, not all is lost!
Instead of buttons, you can use numbered menus and drop handy tips to help customers get through the chat quicker.
You can always check out our Guide to Chatbot Design for WhatsApp for more tips to bypass WhatsApp UI limitations.
All in all, having a WhatsApp recommendation chatbot for eCommerce is not unattainable (especially, when the best chatbot software of 2022 is at your tips). Considering its ample client base and freedom from desktop interactions, working around its limitations is worth it.
In any case, for more complex product chatbots, you will need to make use of NLP… and if that’s the case, be sure to check out our guide to creating an NLP chatbot with Landbot & Dialogflow.