How to Create a Great Search Experience
Chris Bechtel VP of Growth – Inbenta
About this episode
Today we speak with Chris Bechtel about the importance of search on ecommerce websites.
Benefits of a quality search
How to implement a good site search
Which functionalities to look for in a site search
You might also want to listen to…
- Episode #33Simple TVVickmar Martinez & Yenny Velazquez
Podcast #33 – Artificial Intelligence is the Future of Customer Service
- Episode #32CrowdlinkerAram Melkoumov
Podcast #32 – Data access will open up the possibility for AI to grow
- Episode #31NeoenergiaRenato Suplicy
Podcast #31 – Keep the Human at the Center of your Digital Strategy
- Episode #30PATJohn Ball
Podcast #30 – Reasoning Engines and the Truth About AI
Welcome to the Future of Customer Service Podcast. I’m Andrea Palten from Inbenta and I will be interviewing customer support and service professionals to see what is currently working well, what issues they’re trying to overcome, and the future success of customer service.
Hello, today we have Chris Bechtel from Inbenta with us, and we will be chatting about all things search, but specifically we’re going to be talking about e-commerce and search functionalities on e-commerce platforms. So, Chris, thank you so much for being here and I’m going to go right in and let’s start with the basics just for those people that maybe don’t really know what we’re talking about. What is your definition of search on an e-commerce website?
Well, great question. Thanks for having me here, Andrea, and happy to be here and answering some questions for all of you. I mean, of course, you’ve all probably used search many, many times, but obviously on a website, particularly for e-commerce, you know, a search is really about finding a product typically. Now, in many cases, you could have folks trying to find information about support, returns, and all of those other things which Inbenta can also help to do. But in this case, we’re talking about the functionality of a product search or a large category or a catalog of searches. Now, typically we’re talking about 30% of users to your website. An E-commerce site will perform a search and the most important thing, of course, is for those users to get the best possible search result instead of the dreaded zero results page or hundreds of irrelevant results. Now, most searches kind of function sort of based on a keyword, and at Inbenta we go far beyond just keywords and we’re using semantic search and that’s going to deliver a far greater quality of a result.
Okay, and I do want to dive into later about the functionalities and exactly what a semantic search is. But for talking about the basics a little bit more, since you just said something quality. So, I’m curious what makes a good quality search.
Well, so you can kind of imagine, particularly if you’re searching, you may not know, obviously, you know. We want to try and replicate online, that kind of experience that people would have in a store where you have a large store, you’re greeted with a friendly salesperson that’s not too intrusive but there to help you find what you’re looking for. In some cases, as a shopper, you have maybe a very clear idea of what you want, or you may have a somewhat unclear idea and you’re just browsing. In either case, you want to be able to find things that are interesting to you and are appropriate for your tastes or your needs and your trends. So, rather than just sort of a keyword type-driven search where you’re going to just look for handbags, for instance, well, there are lots of different types of handbags of sizes, materials, prices, those that have reviews, those that are frequently purchased.
All of those are things that you might be able to ask a salesperson in a conversational experience, hey, can you show me, what’s the most popular item right now? What’s the thing that everybody’s buying? What do you think is right for my body type? What about the colors that I like? How can I find those things without having to browse through endless pages of sort of broad results that don’t really give you much of anything? So, ideally, you want to be able to find products or services that directly meet your needs and just like you would have that experience in a store.
Yeah. So, what should happen? Let’s pretend. Let’s go with your handbag example. So, I type in pink leather handbag and you’re the store owner, you’re the search person or the person that set up the search and you do not have this. You do not have pink leather handbags. What should be brought up in the search if you don’t have that?
Well, I mean, just like what a quality concierge or salesperson would tell you in a store is, oh, you’re looking for that color. I don’t have that color, but I do have this color. These colors might look good on you. I mean, it’s all about ultimately the journey and we’re talking about customer experience and the move towards digital experiences because of the way trends are moving in terms of behavior. Obviously, we have a global pandemic that’s affecting retail. Retail has been affected even prior to that with the movement of e-commerce and trends and people buying online. So, we’re really trying to create that same sort of experience online but now there’s a lot of benefits actually of an online experience that you actually can’t get in the store because if you think about it online, you can actually expose the entire inventory of a store. Maybe even what’s beyond that can be actually shown in the front of a store.
You can actually expose everything that’s available in the stock room per se. With digital tools, you can allow people to really search and drill down. So, the type of experience that Inbenta enables is for someone to have a very specific search and then further narrow down that search with proactive questions such as, of these would you like to see the most highly reviewed items? And this is data that we can directly get that a salesperson in a store won’t be able to know off the top of their head but with the power of the data, we can in real-time return these products that you found here are the most highly reviewed items. Here are the items that are actually most frequently purchased.
These are all the types of things that, again, you can be able to drill down so you can really find the products and then mix and match and shop and compare in a much more easily findable way. All of that, ultimately in e-commerce, we’re talking about conversion rates and that’s the most critical component to revenue. For most e-commerce websites, there is an average site-wide e-commerce conversion rate, which means visits to purchases is anywhere from one to 3%. So, if we can boost that e-commerce conversion rate to five or 6% depending on the AOV or average order value and the frequency of purchases that improvement in conversion rate can mean pretty substantial dollars.
And we also know retention. A lot of times, like most customers, especially in more of a spontaneous e-commerce purchase, if you’re kind of feeling in the mood to buy something and you’re looking, and you can’t find what you’re looking for, you may not never come back to that website again. If you have a frustrating experience and you can’t search and you can’t find you may just never come back again. So, it’s pretty critical for an e-commerce brand to provide that great search experience. Especially in many cases, e-commerce brands are paying money through paid media, social advertising to get people to come to that website. If a fewer number of them convert because they land on the site and when they try and search for something, they can’t find what they’re looking for and they don’t have a really great experience in doing so they may leave and not come back again, and you’ve wasted that ad dollar and ultimately you’re really impacted your overall revenue.
Yeah and nobody wants that.
Nobody wants that.
For our listeners that are in charge of implementing search on their e-commerce platform you and I talked offline before we started hitting record. There are three different ways you can go about implementing site search. Can you tell us about the three?
Yeah. So, typically, almost every e-com platform or content management solution is going to have a built-in search. All of those searches are all based on keyword-driven searches. Now, depending on the level of detail in the product catalog in terms of keywords that are in there and the number of products. Again, if it’s a fairly simple site with just a handful of products a keyword search might be sufficient It may certainly do the job. Now, another way is you can build it in-house. Maybe there are varying third-party technologies or other things that you might be able to implement. But of course, that means you’re going to have to maintain this and that can become quite costly and a difficult burden. In many cases, probably many e-commerce companies don’t have a particular person that’s dedicated to be doing that work.
Then the third, of course, besides just sort of building a homegrown solution or using the built-in CMS or e-com platform search is to purchase a dedicated search solution from a third-party vendor. That’s where Inbenta comes into play and this is where you’re really going to get a super powerful out-of-the-box powerful AI-powered semantic search engine. Now, what do we mean by that? So, AI traditionally means there’s a machine learning component and that’s one component of AI. At Inbenta we also use what we call symbolic AI and that means essentially the simplest explanation and forgive me, because I’m the head of growth here. I am not a machine learning or linguist or knowledge engineer, which we have on our team. But what I would explain to you essentially, is that using symbolic, everything in language, we basically translate into symbols and in the world of conversational AI, essentially each topic is boiled down to intent.
Essentially what Inbenta has been able to do with our combination of symbolic AI and machine learning and our Inbenta lexicon, which is a database of hundreds of thousands of pre-existing meanings and relationships. We can quickly index your site, your knowledge base, your FAQ, your product catalog and very quickly using our semantic search technology within days to provide a very high-quality result. So, in many cases for a lot of other machine learning-based search engines and others, you have to give it a lot of training data, meaning all the various ways in which people would ask a question or ask something. But `Inbenta because of our power of search, that’s not necessary. So very, very quickly, we’re going to upgrade your existing keyword-based search to a powerful, semantic-driven search engine. One of the things that are going to do the machine learning part is it’s going to learn the most common questions and there are options to show and highlight those very common questions.
So, it’s almost like the suggested questions where it’s ultimately going to highlight the most popular products, the most popular product searches at the top of the page. I’ve seen some examples where companies have tried to implement Google search onto their website, for example, and what shows up on a Google search that’s been implemented into a company is other searches and even ads. I’ve seen companies have a Google search on their page with ads for their competitor. Not something that you probably want to do. So, that’s why we obviously recommend upgrading a keyword-driven search to a semantic-driven search. Because like I said before, it’s going to give you the ability to find, sort, and have recommended products that are much more likely to meet the need of the customer and are much more likely to generate a search to cart action and therefore increased conversions and increased revenue.
Yes. Which is good. So, I want to ask you something about the semantic search. So, Nordstrom, for example, sell shoes, they sell women’s clothing, all kinds of stuff. They recently, when COVID hit started selling masks. So, does the semantic search then, can it get smart, real fast when new terminology like COVID, COVID-19 pandemic, masks, COVID mask, all those different words all of a sudden are new in the lexicon for everybody. Is that then something that the tool can quickly pick up on?
Yes. Now the other thing that’s important, probably for everyone to understand who is maybe new to the world of semantic search and conversational AI is the phrase NLP which we’re talking about natural language processing. Its cousin is natural language understanding (NLU). So, you may see out there these terms NLU and NLP, and that’s exactly what Inbenta specializes in is natural language processing. The ability for a human to communicate in a search or in a chatbot with the machine in their way in which they naturally express themselves, in their natural language. We need the machine to be able to understand what the intent is, what the customer is asking about. In the case of masks, for example, and as I mentioned before Inbenta has this huge lexicon.
There are a number of ways in which we could actually handle that mask situation. If there are terms that are universally applicable that have meaning like that, they would be added to our global lexicon. We have a team that is continuously improving that, and that same knowledge can include a lot of already pre-existing industry-specific knowledge. Again, all of the e-commerce knowledge about returns and cancellations and tracking, and where’s my shipment, and all of that standard type of language is already pre-trained. But if for a specific organization, I think here’s one example. We have a customer who has some products, and their products are actually animal names and so if a customer is asking about a kangaroo, let’s say, they’re not actually asking about the animal kangaroo. They’re asking about the product that’s called kangaroo.
So, therefore exactly within our platform and our team, we have as I mentioned, knowledge engineers that work with our clients to keep this implementation super simple and really fast. You have the ability to train the system for that particular customer that when people use the name kangaroo, it actually does not mean the name animal. It equates to this specific product that this customer has. So, even with the term mask, yes, there are ways in which we can absolutely understand those terms and direct the customer. In some cases, we might have to ask a clarifying question. Are you talking about a mask like this or are you talking about a mask like this? Now, giving the customer the option to then choose rather than the terribly frustrating answer, which is zero results in search, or I’m sorry, I don’t understand your question neither of which are acceptable.
Yes. I actually, personally, when I search and I get so many answers and they’re not really related, it is really frustrating for the user. Like you said, earlier, I leave, and I go to a different website and buy from them.
Exactly. So, obviously, there is a whole science to conversational AI dialogues meaning in many cases you don’t want to ask yes or no questions. You ask clear questions and give in many cases multiple choice answers to customers that allow that a customer chooses the right answer and drill down further into that topic. Now again, we’re talking a little bit about, which we haven’t addressed yet in this podcast, but essentially is at Inbenta we can layer a conversational AI or a chatbot on top of the search experience.
So, within that chatbot, you can have the same sort of dialogue that you would just type into search, but it’s more of a conversation. The results would pop out from the chatbots. You can see your options, you can filter, you can narrow the search, you can ask the chatbot some questions and you can give it some direction so that the chatbot, which can also be proactive, just like a good salesperson and ask you questions as I mentioned before, like, would you like to see the most popular items? Would you like to see the most highly reviewed? Those types of things. All of that is important in having a great customer experience.
Got it. Got it. So, my last question for you, Chris today is, and you’ve already touched upon this, but I’d love to summarize. Which functionalities are the ones to look for when you’re trying to implement the perfect search for your e-commerce site? You already mentioned semantic search using NLP technology. Can you summarize them up, the functionalities?
Yes. I mean, I think that is absolutely critical right. Number one. I mean, ultimately the most important thing in all of this is are your humans understood i.e., are the customers understood? And therefore, the answers they are looking for, the products they are looking for are found and returned in an easily branded high quality, elegant user interface that is easy to use, that’s intuitive, that’s flexible, and personalized. Because what we’re really talking about, and one of the big opportunities of course, in e-commerce now that we have all of this data is more and more personalized experiences. You also want to look for a vendor that, of course via API calls and webhooks can enrich the conversation with other data like I mentioned, frequently purchased products, related products, all of those things.
So, you want to look for a vendor that has already pre-built all of these types of integrations so that conversations be enriched. You want to look for a vendor that’s proven its ability to ingest quickly all of your existing data sources because that’s the other thing. Sometimes this data is going to reside, data i.e., your product catalog, but you might also have other information such as reviews of those products. Maybe there’s an FAQ or other knowledge, and you want to unify all of that into one search and one conversational experience. So, I think you definitely want to work with a vendor that can unify all of that data from a bunch of different sources and then make it available in a search easily. You also want to work with a vendor where the AI is explainable, meaning you understand why it came to the result that it did.
Machine learning as I mentioned before, is something that’s important, but in many cases, it can also be something that can delay an implementation quite a long time, because for machine learning to really work, you have to give it a lot, a lot of data, which means there’s a lot of time and money in the setup and maintenance. So, you want to work with a vendor that has the ability to get to market quickly and really deliver what I would call rapid time to value because again, depending on your size as an organization, even the largest e-com companies, everyone still has multiple responsibilities and limited resources since you want to work with a product that has a team behind it, that’s going to make your job implementing it and your job maintaining it very easy.
Awesome. Thank you. This was really good. I really appreciate your time and for anybody that wants to get more information, obviously, you can go to the website at inbenta.com. Schedule a demo, even try it for free, and then you can see what it’s actually like. If they want to talk to Chris, Chris, where can they find you LinkedIn or?
They can find me on LinkedIn. I have been on Twitter, not for a little bit. You can find me there @chrisbechtel. Those are the best places and, of course, you’re welcome to reach out, fill out a contact us on our website, happy to talk to anyone one-to-one. I spent a lot of time in the e-com space. Really understand of course how to acquire, retain, and get customers to then refer others. I think that’s really the key here and that’s what we’re really doing at Inbenta. Happy customers make customers that are more likely to buy, more likely to refer others, and therefore accelerate your revenue growth and that’s what we’re here for. So, thanks to everybody for tuning in today. Look forward to continuing to give you knowledge and resources.
Thanks so much for tuning in. This podcast was brought to you by Inbenta. Inbenta symbolic AI implements natural language processing that requires no training data with Inbenta’s extensive lexicon and patented algorithms. Check out this robust customer interaction platform for your AI needs, from chatbots to search to knowledge centers and messenger platforms. Just go to our website to request a demo at inbenta.com. That’s I-N-B-E-N-T-A.com and if you liked what you heard today, please be sure to subscribe to this podcast and leave us a review. Thank you.
Interested in Customer Service Automation?
Download our ebook!
Three smart solutions that can make an impact on your AI-powered customer service automation
Case studies related to each solution
Tips to choose the best technology to automate your customer service