Chatbots are not only helping your customers to find what they need and get help with their questions, but they can also collect a lot of useful insights into what customers are wondering about when they visit your site. Can you imagine a scenario that would lead to an e-commerce brand realising that they need to use a chatbot though? Let's take a look!
A, B and C work at an e-commerce brand, which is struggling with high spend, low conversion rates and a high number of repetitive enquiries to customer services. They get together one afternoon to discuss this. It is a strong case of ‘worst case scenario’, but can happen if little attention is paid towards understanding and preventing high bounce rates and dissatisfied customers.
A: So, how’s it going with the online performance, E-commerce Director?
B: We’re spending a tonne of cash on getting visitors to our website, but conversions have barely increased. Maybe 0.3% since we decided to increase the budget and make changes to the website.
A: Hmmm, that doesn’t sound promising. How are things on your end, Customer Service Director? How’s the general customer satisfaction rate?
C: We’re honestly so swamped with calls and emails right now, that my staff are working very hard but haven’t been able to answer everything. The abandoned call rate has increased by 12%, and we have a backlog of 600+ emails, growing by the day. I only have 12 FTEs!
A: So it’s all going to pot, basically? What is the main reason the customers aren’t buying though, and why do we seem to have so many enquiries? What are the common issues?
B: Couldn’t tell you that, we don’t have an efficient way of determining this.
C: I’ll let you know when we’ve cleared the majority of the backlog!
A: Have we thought about different ways to gather more insights?
Chatbot has entered the room
🤖 Insights, you say? I can help you to gain insights into what your customers are wondering about!
A: A chatbot?! Aren’t you just a programme that answers FAQs?
🤖 Aren’t you just a senior executive that doesn’t understand your own customers?
A legitimate challenge faced by many organisations is whether they have enough insights into many areas of their business, leading to them becoming reactive and not proactive. We’ll therefore look at the role of a chatbot, or proactive digital assistant, in giving potentially frustrated visitors a voice.
Note the use of the word proactive here as well; if you’re working with conversion rate optimization and conversion marketing, then this will be right up your alley, although the notion of giving customers a voice should take priority here.
Gaining customer experience insights that matter
To first understand this, we need to rewind a bit and take it back to customer experience (CX). A good CX is what any business should strive for, although CX tends to be more defined by the result of the interactions between a business and customers. As more and more of our world moves online, the human touch is gradually disappearing, but is mitigated by having various methods for people to interact with a company. There are many different touchpoints for customers, including apps, the website, email newsletters, surveys, text messages after a call etc, all of which carry possibilities to engage and interact with customers, and collect insights.
What needs to be pointed out though is that a number of these methods may be reactive, rather than proactive, and likely to collect insights after an event occurs, and not before. This can leave the customer feeling that they’re two steps behind; they’ve had to find ways themselves to resolve an issue, and then may be asked for their feedback afterwards. This could end up affecting the overall CX in a negative way. Ironically, the only way to know this would be to collect feedback.
Better customer insight in real time
There are various methods for gathering insights live on a site, although have you ever thought about the possibility of using a chatbot to gather insights? Few tools will give you the same rich data that a chatbot gives, as it’s live on the page and appears to every single customer, having the potential to create a dialogue with hundreds or even thousands of people.
While the nature of the dialogue may end up going beyond what the chatbot can handle and require the human touch (especially for complaints and support issues), all interactions are important insights into the burning issues that customers face. Static FAQ pages can answer a wide range of FAQs, but are more general in their nature.
In addition to this, they tend to be not very interactive (unless they’re designed well), and as suggested above, more reactive than proactive. A proactive bot could place more emphasis on understanding why the user seeks the FAQ section, so that recurring themes can be clustered, mapped out and explored in more detail.
It’s also possible to trigger a dialogue that asks the user whether they’ve found what they’re looking for upon exit intent, which may go a long way to explaining high bounce rates. These are just a number of different methods used to give visitors a voice, kind of like an actual employee in a shop would.
Turning customer data into marketing insights
If an item/product/service isn’t selling as well as it should, then surely you would want to know why? Granted, we can’t get everyone to answer a questionnaire or survey, but even if 1,000 out of 10,000 people answer the survey, then that’s still pretty positive. You never know what insights can be derived from this sample; what if 5 out of 100 people who had bought a certain product have had the same issues with it? Worth knowing, right?
For quality control issues, which do invariably happen, you can even link to forms or contact details for letting the appropriate teams know. The most important thing is that you can encourage users to leave reviews, submit feedback and recommend products and services, so that these can also be recommended to others.
Encouragement is key, and it’s good to be proactive about this as well; returning users are usually returning for a good reason, such as to buy more or to make a complaint/report an issue. Anything that can potentially impact future revenue must be seen as a good thing, right? This is why collecting insights is key to understanding both users’ buying behaviours and pre- and post-purchase concerns, as it will help to alleviate both hesitations and frustrations.
Customer insights tools - using a chatbot for data analytics
A good way of looking at this is thus; in order to find and identify solutions, you need to use good tools to diagnose the problem. There’s no shame in admitting that what you’re offering is a tool, as long as it’s a tool that actually adds value. However, one important point to remember is illustrated by this very apt quote: “a fool with a tool is still a fool” (Grady Brooch, 1989). Therefore, use the tool wisely to collect important insights, which can be highly beneficial when looking at ways to improve the overall customer experience.
In conclusion, insights gained through dialogue will enable the collection of valuable data from every step of the purchase journey, from the browsing stage to post-purchase, with many steps in between.
So, how should this conversation have really gone?
A: How’s it going, E-commerce Director?
B: Great! Steady increase in conversions and average order value, thanks to more relevant offers towards visitors, which many users who gave us feedback asked for. We’ve also seen good engagement and conversions by sending customers their shopping cart contents by email. A number of visitors have told us that they haven’t found what they’re looking for while browsing using the search function, so we’re currently working on improving this.
A: Fantastic! How about on your end, Customer Service Director?
C: It’s going well, although users are mostly telling us via the chatbot that they’ve not received their packages, or that there’s something wrong with their item upon delivery, and they need to return it. We’ve therefore placed links to check deliveries and to return items in the chatbot, which has reduced emails to customer services by 30%.