Upcoming Webinar Aug 17: Sunset Endeca and Adopt AI to Exceed Customer Expectations REGISTER NOW

Close Message

Connecting with Customers Through Search

Presented at virtual Activate 2020. Search offers a unique relationship to the digital customer in that they literally spell out their intent. By combining this data with user behavior data at scale, patterns emerge which can be used to improve your connection to your customer. Since taking the first steps with AI, Lenovo has seen a dramatic increase in customer engagement and conversion through search worldwide. We are just at the start of this journey and like many large companies, we have to balance the goals of many stakeholders, product owners and business leaders competing for a finite budget.

Through this journey however, we have been able to demonstrate superior ROI at every step, delivering automation at scale both improving the business results and reducing the cost and time to maintain the site. As we improve and expand the use AI we continue to discover more opportunities and use cases where these powerful tools can help us connect to customers across the globe faster and with more relevance.

Speaker:
Marc Desormeau, Product Owner for Site Search and Product Data Optimization, Lenovo

Intended Audience:
Tech and business leaders, looking to extend and improve the search experience through AI

Attendee Takeaway:
Learn about the Lenovo journey and how we are looking beyond the traditional use case of search

Learn more about how Lenovo uses Fusion.


Transcript

Marc Desormean: Hi, I’m Marc Desormeau. I’m responsible for lenovo.com global site search and strategy, and I’m happy to be here at virtual Activate this year.

Before I take you through Lenovo’s journey with search and with Fusion, I wanted to give you just a little bit of a background on Lenovo. Lenovo started in 1984 with a group of engineers that were looking to bring PCs to the China market. Since then it’s grown into a $50 billion company. It’s a Fortune Global 500 technology company. It’s number one in PC market share. It’s also ranked number one in high performance computing with 173 of the top 500 systems running Lenovo. We have 63,000 employees around the world and we do business in 180 markets, so a fairly complex and broad multinational.

Three years ago, we started our journey with Search and with Lucidworks and, after a very comprehensive vetting process, the choice for Fusion was based on their vision of search as an insights engine, beyond just search. We knew where we wanted to go with Lenovo in ecommerce, that we needed something that would take us beyond just a search box.

Today, search supports 37 languages, we cover 86 countries, and we have use cases that support B2C, SMB, and private sites. We support both direct and indirect sites. We have transactional use cases, purchasing products, and very complex use cases around enterprise branding.

We’ve got a very broad range of products and services that we need to support as well as customer segments. Then, furthermore, being a multinational and having a strong presence in all these countries, we need to be sensitive to different cultural norms and behaviors.

Let’s talk a little bit about the beginning of the journey. Lenovo starts with simple search. We start with basic keyword searches, TrackPoint, server, data. We have to support both content and products. 50% of our searches are two keywords or less.

Now, that trend is shifting. We’re seeing with the adoption of home assistants and digital assistants customers are increasingly becoming accustomed to phrase-based searches.

That’s a challenge in some respects, there’s a lot of data to parse through, but it also provides us much more insights into how customers are engaging with the site and what ultimately they’re trying to accomplish with their visit to lenovo.com.

Let’s talk a little bit about the search progression. We started out much like every search installation deployment, with organic search, and starting with simple aggregation of content. This is sourced from multiple different creative groups, from different data sources, and ultimately the success of the search results depend largely on the quality of the content. This is a process-driven search experience.

Next, looking to improve our search results for our customers, we started applying rules to the search engine. This is a more effective way to narrow the results, but again, is limited in that it’s only as good as the knowledge of each of the people that we work with.

We started working with merchandisers and salespeople around the world, but fundamentally, this is also sensitive to time lags, as typically, analysis is done after the fact and applied, and often it’s more a function of the desires of an individual group within the company and not necessarily reflected what the customers want.

This is a business-driven approach to search results. The next level with signals is where I’d say the magic starts to happen.

As any user experience professional will tell you, you might think you know what customers want. But until you get it in front of customers, until you actually poll your customers and start exposing them to the experience or the product, you don’t understand what it is they’re trying to accomplish.

Signals are the customer-driven level of search. What we’re trying to do here is leverage what customers are telling us and adapt it in near-real time to the events and the circumstances. Signals are also allowing us now to start scaling without human intervention.

This is, I think, a pivotal point and the application of insights to the search experience. Now, again, this is simple, this is the first implementation of signals. We’re starting to use signals and using the weights differentiated by use cases.

We start with some very traditional use cases, product results based off of keywords, content results based off of keywords, and this is where we start applying layers of information and insights to customers.

Segment information, store information, geography, or even getting down to individual users when we can identify a registered user to start trying to present more relevant information. Building on this, we then start getting into more sophisticated, or the next level of insights and starting to use those signals to influence customer behavior. We start tying in typeahead with signals and start making recommendations based off of keywords at the outset.

Now, the power in the shift with this approach is not only are we able to look at historical data and make judgment calls based on the data, but we’re now starting to help the customer understand the journey and we’re starting to lead the customer down the journey. By auto-complete and starting to use typeahead, we’re now taking the next step in this level of implementation to lead the customer down to the most successful or likely successful path.

The next level is starting to make recommendations. When customers are stumped, they can’t find what they’re looking for. They start to use signals to make the best guess recommendations to, again, try to get the customer down the path of the journey.

Let’s take a look at what the results were. The business impact of signals, this was a year-long effort in developing and applying signals. We’ve been able to see a 73% increase in engagement in search. We’re seeing a 93% engagement in click-through rate, or an increase in click-through rate. We see a 35% increase in conversion rate, and importantly, a 34% increase in customer satisfaction. This is, I think, the tipping point for the business in terms of starting to recognize what we as a search team have been saying all along, and, what was the promise of value of Fusion and leveraging that platform.

For us, it was starting to get beyond the search box, starting to say, well, look, we can start not only understanding what customers want, but we can start providing results, improving results in a way that is better than human intervention.

One of the real tipping points for the business, and I’ll talk about this a little bit more later, was when we went from a positive, an increase in revenue contribution per visitor for people who engage with search over browse.

People who engage with search now are spending more money with Lenovo and are more successful at finding their or completing their tasks than just simply browsing. This was a bit of an epiphany for the business team.

As we start presenting the ROI and the investment, we’re able to start using these metrics to demonstrate that, you know,, we’re at the point now where we can leverage the analytics and the insights to provide even better results than a curated response from a human being.

Why is that important? Circumstances this year have been tragic and unprecedented, but with the use of signals, we’ve been able to start driving these changes and reacting to the circumstances at scale and at speed that we simply can’t do manually.

By now, we’re all familiar with this map or a version of it. As we live through these unprecedented times, traditional business processes are to keep up with the rate and change of business at all levels. What we found is, with the application of signals, we’ve been able to adapt and modify the experience based on very local trends.

Simply curating the data and the content manually has just not been able to keep pace. With search driving a significant portion of the overall experience at Lenovo, we’ve been able to match and keep pace with these changes. We’ve seen changes at every country level, and even within a country regional variations that are driving micro-behavioral changes and purchase patterns.

The world is changing and continues to change and old models are simply not able to keep pace.
What that means to us is we have to have the ability to adapt in real time, which has been the strength of these customer insights and data. We’re able to adjust to a shift in purchase patterns. We have businesses buying consumer products. We have consumers buying business products. There’s a surgeon’s home office equipment purchases. We’re having to react to inventory constraints, which then are reflected on customer behavior.

There’s an urgency of need with parents finding themselves suddenly working from home in an environment that was not anticipated. We’ve got children now that have to learn from home. We have to do this at a scale, as a search platform across 1400 stores worldwide in all these different countries.

Being able to react at scale is what’s been, and in near-real time, has been the real power of this and has allowed Lenovo to try to accommodate these critical needs from our customers and react in a timely fashion.

What we’re seeing now is a shift again. We started out with customers focused on the safety and security products, focused on purchasing things that were deemed critical, hand sanitizers, cleaning products, face masks.

These were all sort of the urgent critical needs that customers were purchasing online spurred this shift to online purchases. But now, as people are starting to get used to spending more time at home, purchasing online, we’re starting to see some of these behaviors shift again.

Now they’re starting to move towards purchasing items that are more supporting of an extended stay at home.
There’s been a shift in the focus on comfort. I’ll be honest, this is the first time I’ve worn a shirt in six months. We’re seeing what I thought was an anecdotal, an interesting anecdotal data point. Bazzarvoice’s customers were found to have a decrease in pants sales and jacket sales and 147% increase in year over year sales of pajamas as more people work from home.

There’s definitely this ongoing shift of patterns. As the rules change, and as this crisis plays out, we’re finding that trying to predict these behaviors is very, very challenging.

Being able to quickly aggregate data, and then react to those has been a very important innovation for Lenovo.

Let’s talk a little bit about the shift. Overall there’s been a global shift, a 20%-plus growth in online purchases. This is a trend that we’ve seen happening over the course of several years, the shift from face-to-face brick and mortar to online.

The pandemic has, I think, served to accelerate what was an existing trend and has acted as a catalyst to drive customers that may not have traditionally been comfortable purchasing online to online purchasing. As a result, behavior’s still shifting. We find people are starting to buy products that were traditionally bought in the store and in a brick and mortar, are now having to move online by necessity.

Let’s talk about the impact at Lenovo. While everyone’s seen growth in online sales since March, search growth has outpaced browse growth at a 23% premium.

There’s a lot of reasons for that. We’ve got customers that are likely unfamiliar with our product that are in a position now to have to buy home electronics and office equipment. We’ve got customers that are unfamiliar with the navigation of the site. We’ve got certainly a set of customers that may be unfamiliar with online purchasing and e-commerce as a whole, but are forced now to make their purchases online, likely trained by Amazon and Google. Search becomes a much more familiar way to engage with the site.

Furthermore, when customers can’t find products, they rely on search to help them find the adjacent products. There’s a number of reasons why customers are starting to gravitate to search, and we’re starting to see an increased trend in engagement in search as a way to very quickly find and achieve your goals on the site.

The other significant point I’d made earlier, which has again been the launching point for our discussions with the business as we start looking beyond just search, is this idea that, with the tuning and the optimization of the data and the algorithms, we’ve been able to drive better results than manual intervention through browse. We’re starting to see a positive trend in terms of the revenue contribution for a visitor. That’s significant because this is an area we’ve been focusing on for the last six months and are starting to show gains in this area around the world.

What that’s allowed us to do is then get to this idea, these conversations of beyond just search, beyond just what we would traditionally find in a search results page, and frankly, what most people understand as what a search engine brings to the table. This goes back to our initial evaluation of Fusion as an insights engine.

The goal was always to start joining data in creative and new ways. We’re looking at leveraging the investment we’ve made in the recommendations engine to start using those insights to drive different changes or different experiences across the site, even outside of search. While customers don’t realize it’s Fusion, and fundamentally, the search engine driving these insights and driving these recommendations and this user experience, it’s fundamentally the same technology, the same business process, and the same machine learning and algorithms and tools that have been put in place, and we’re now starting to scale that beyond search.

We’re looking at opportunities to start making recommendations on model page accessories, looking at segment-based recommendations throughout the site, and dynamic navigation flows.

Why be restricted to a particular navigation paradigm when we can react and interact with customers as they touch the different areas of the site? Can we personalize a catalog for customers based on their known history, or based on registration, or based on customers in that segment?

Ultimately, we’re starting to explore how we can use search to start driving a differentiated mobile experience.
When we look at a user experience that’s constrained, either, from a bandwidth perspective or from a form factor perspective, search is an obvious vehicle by which we can very quickly adapt only the most relevant and needed information to that customer’s experience in a session. It’s a very powerful way for us to scale broadly without having to recode and reinvent the navigation on the site or have to redevelop new flows or unique flows.

Let’s face it, personalization at scale is difficult and expensive. What Fusion’s allowed us to do is aggregate huge amounts of data, interpret and then apply that data, and then test our hypotheses.

Fusion gives us a vehicle that provides us quantifiable and measurable ways to improve our connection with customers by understanding the different ways that they engage with the site and being present in a way that’s relevant to the customer where they need us and when they need us.

Thank you for letting me share our story with you. I hope you’ll enjoy the rest of the virtual conference and stay safe. Thank you.

Stay up to date on Fusion