But what are we actually talking about? What is the breakthrough here and what does that mean for you?  

ChatGPT is now the latest tectonic shift in A.I. functionality and truly is that breakthrough. Why?  Because OpenAI has achieved the breakthrough in User Experience.  They didn’t create it, but they did kind of perfect it.  It’s called ad hoc query and the concept has been around since the early 1980s when the relational database boom began (as well as my career, btw). We used to promote ad hoc query at Oracle and it was very real, provided you could speak SQL, or use one of the graphic tools that generated the appropriate SQL. 

Fast forward to the next step in creating the perfect UX. 2011 – Apple launched the iPhone 4s with Siri integrated. For the very first time, millions of us could access the perfect mobile user experience. Hit a button, ask a question, get the answer. Again, ad hoc query, but now Mobile and Voice enabled. 

The very first time I tried this, I recognized THIS was the future. This was the way business applications would be architected, this was the way we would all shop on eCommerce platforms: intelligent, intuitive, interactive. It’s actually what we do here at EasyAsk and Gartner gave us their ‘Cool Vendor’ award for this… but it took another decade for it to be rediscovered and mainstreamed. ChatGPT did that, so ‘thank you’, OpenAI. Combining an Artificial Intelligence/Natural Language query engine with Large Language Models generating comprehensive content for seemingly endless Q&A: Generative A.I.

Like Siri, then OK Google, ChatGPT has the unenviable task of having to ‘index the universe’. So, a common concern is, “is this answer 100% accurate?”. You’ll notice that Chat now qualifies the answer with “to the best of my understanding” or something close. To be clear, this is not a criticism. The tech is fantastic. It’s just always going to be an issue, what content is the answer based upon? They’ve done an amazing job bringing forth a transformative technology and instantly launched the concept of ‘ad hoc query’ forward. We should all be grateful. It IS the perfect User Experience. 

So where do we take this going forward? We, like OpenAI believe in the power of Artificial Intelligence and Natural Language mapped to various business applications and vertical market solutions. The future is very bright to transform the User Experience for Business Intelligence and Analytics, CRM, Medical records, eCommerce and so much more. Congratulations and thank you to OpenAI for paving the road here. I believe it will quickly become a freeway! Off we go. 

The post Artificial Intelligence for Business appeared first on EasyAsk.

When eCommerce site search is working well, it is a dream for your customers; they can find what they’re looking for on the first try.  No need to endlessly click through your site’s categories and attributes.  Unfortunately, eCommerce search is too often scaring customers away.  At a time when years of eCommerce progress has been made in a matter of months, it’s even more important than ever that your site search is up to scratch.  We’ll look at six scary statistics and the ways that you can ensure your site search is not one of the frightening ones…

Scary Statistic #1

70% of eCommerce search implementations are unable to return relevant results,

requiring users to search using the exact same jargon as the site[1]

Yes, that’s right, nearly three-quarters of eCommerce site searches will only return the right products if the customers happen to use the right language.  That’s a lot of lost revenue when those customers become frustrated and abandon the site.

To ensure that your customers can find what they’re looking for, you need a Natural Language search solution.  Natural Language search solutions find different versions of the same terms so that pluralities and tenses can be managed.  For example, it wouldn’t make any difference if a customer used the term “mens”, “man”, “men” or “womens”, “women”, “ladies”.  An intelligent Natural Language search system will understand this automatically, without any input.

Scary Statistic #2

34% of eCommerce site searches don’t return useful results when users search for a model number or misspell just a single character in the product title[2]

This scary statistic raises 2 important issues:

1. Misspelling

A good spell correction system is not just about correcting, but about finding words that occur in the product data which most closely match what the user described.

With EasyAsk’s intuitive search system, your product data becomes the look-up dictionary, meaning that your customers don’t have to match the spelling of your products exactly.

2. Complex Product Numbers

Whilst most good search systems will find common misspellings, they lack the ability to deal effectively with part or model numbers.  Consumers who are taking the time to type in a part number are undoubtedly more ready to buy, so it is even more important that they are able to find what they are looking for.  This is a particular issue for B2B companies.

EasyAsk has the ability to index part and model numbers comprehensively, indexing all variations to ensure that users can find the product, even if they forget or mistype part of the number.

EasyAsk’s ‘Part Number Expander’

A ‘Part Number Expander’ is especially useful for complex part numbers that include a combination of letters, numbers, and other characters.  A user might forget whether characters were separated by a hyphen or a slash, or might omit letters from the end.

The Part Number Expander takes each part number and creates all the different versions of terms, inserting them into the searchable index so that if any of them are searched for, they will match to the product.  The code is broken up into parts and the separators are substituted in all combinations.

For example, if the part number is

123-HC/1345AB

the Part Number Expander would index:

123-HC-1345AB                                     123

123/HC/1345AB                                    123-HC

123/HC-1345AB                                    HC

123 HC 1345AB                         1345

…amongst many other combinations.

If a customer types a part/model number that is slightly different (for example, using spaces instead of hyphens) the correct product will still be found.

JLMWholesale.com demonstrates how Part Number Expander can avoid no results for customers:

The part number for this door handle is 1191-E-3 US32D.  If a customer replaces the hyphens with spaces and omits the last section of the part number, the exact product is still returned.

Scary Statistic #3

If people have a negative experience on mobile, they’re 62% less likely to purchase from you in the future[3]

A negative experience will stay with prospects longer than a positive one, so if you want your mobile customers to become regular, you’ll need to make sure you’re not putting them off with a poor user experience compared to your desktop site.

Does your site have a mobile responsive design, so that the text, images, and menus change according to the screen size?  Furthermore, as your site adapts to a mobile size, do you show relevant categories and attributes, as opposed to just the first ones from the desktop site.

Customers using a mobile eCommerce site are more likely to use voice input, and therefore different word choices and sentence structures when speaking.  Your search system needs to be able to cope with voice input and the long-tail queries that are more likely on mobile.

Scary Statistic #4

46% of e-commerce sites have a “No Results Page” implementation that is essentially a dead-end for users, offering no more than a generic set of search tips[4]

Generating a generic ‘Sorry, no results matched your search’ message could be disastrous.  How many shoppers would stay on your site and how many would go straight to your competitor?

There are better ways to use a ‘No Results’ page, but what if it was possible to avoid searches returning no results in the first place?

It may be that a simple spell correction rectifies a ‘No Results’ search, but it is also important to monitor what EasyAsk calls ‘relaxed’ searches.  A relaxed search is one where the search system has modified the search by intelligently dropping (or ‘relaxing’) a term or terms.

For example, if a customer searched for a ‘Men’s black lace jacket’ and there are no products that match this description, we would want the results to show ‘Men’s black jackets’, rather than ‘Men’s lace jackets’ or ‘black lace jackets’, having relaxed the term ‘lace’.  Some searches will inevitably need to be modified and in this example, we want our search system to recognize that the gender stated in the search query is more important than the fabric.

If a user’s search has been modified (such as by spell correcting or relaxation) it is important to display a message explaining the changes to the original search.  For example, a user may search for “purple leather couches” and see the following message:

This user could have searched for leather couches in the first instance and spent a long time scrolling through the pages of results looking for any in purple.  But by using messaging in this way, the user is informed immediately and is more likely to use the search system again.

So-called intelligent ‘No Results’ pages will suggest searches based on what a user asked for, but truly intelligent search systems will change the search to find relevant products based on what the user asked for, even though there may not be an exact match.

Scary Statistic #5

It’s 5 to 25 times more expensive to acquire a new customer than it is to retain an existing one[5]

A worrying number of businesses still don’t use merchandising tools such as cross-sells and up-sells effectively on their eCommerce sites.  These businesses are leaving money on the table.

EasyAsk’s search solution includes a comprehensive set of merchandising controls.  With 9 predefined promotion types, banners, and business rules, amongst other functions, the business user is in control of the products that are presented to the customer.

Scary Statistic #6

36% of autocomplete implementations on eCommerce websites do more harm than good[6]

Although basic autocomplete suggestions can be found on 82% of eCommerce sites, most sites do not take the opportunity to capitalize on merchandising while users are searching.

As a user types into a search box supported by EasyAsk, however, search suggestions and category suggestions will be displayed as well as the products themselves.

EasyAsk offers ‘Search as you Type’, or SAYT, because it means a whole lot more than just a list of suggested searches.  It means many possibilities for getting the right products in front of your customers.

SAYT progressively searches for and filters through text.  As a user types a query, suggestions, products, categories, and attributes are found and presented with each keystroke.  This allows a user to stop short of typing the entire word or phrase and find what they were looking for quicker.

Search suggestions as well as relevant and popular products, attributes, and categories can all be displayed in an easy to use interface.

Here is an example from EasyAsk customer personalizationmall.com:

The user is looking for a wedding gift and the letters ‘wedd’ have been typed into the search box on the desktop site so far.  The above drop-down panel is displayed without needing to execute the search.  Alongside the search suggestions in the right-hand column, a popular category to browse and 7 products are displayed, based on the first search prediction, ‘wedding’.  Scrolling through the search suggestions using the down and up arrows updates the real-time search and therefore the categories and products displayed.  Personalization Mall could have also configured their search so that it updates just by a mouse hover for a certain amount of time.

Typed characters are shown in bold in the list of suggestions, with the rest of the suggestion in normal type.  Often, autocomplete systems only show suggestions that begin with the typed characters, but we can see here that the suggestions can contain ‘wedd’ anywhere.

Presenting customers with the most relevant search, product, and category suggestions is essential.

Conclusion

When your site search is performing well, customers won’t be scared off, and your business won’t become one of the statistics above.

Getting the right products in front of the customer as quickly as possible shortens the path to conversion, keeping your customers and your bank balance happy.

Follow the tips and best practices in EasyAsk’s upcoming eGuide: “Preparing for the Holiday Season: A Guide for your Ecommerce Site” to ensure your customers find the right products the first time.

[1] https://baymard.com/blog/ecommerce-search-report-and-benchmark

[2] https://baymard.com/blog/ecommerce-search-report-and-benchmark

[3] https://www.thinkwithgoogle.com/intl/en-gb/advertising-channels/mobile/few-tips-speed-your-mobile-site-and-tools-test-it/

[4] https://baymard.com/ecommerce-search/benchmark/page-types/no-search-results-page

[5] https://hbr.org/2014/10/the-value-of-keeping-the-right-customers

[6] https://www.smashingmagazine.com/2014/08/the-current-state-of-e-commerce-search/

The post Six Scary Site Search Statistics appeared first on EasyAsk.

When eCommerce site search is working well, it is a dream for your customers:  they can find what they’re looking for, the first time.  No need to endlessly click through your site’s categories and attributes.  Unfortunately, eCommerce search is too often scaring customers away.  We’ll look at six scary statistics and the ways that you can ensure your site search is not one of the frightening ones…

Scary Statistic #1

70% of eCommerce search implementations are unable to return relevant results,

requiring users to search using the exact same jargon as the site[1]

Yes, that’s right, nearly three-quarters of eCommerce site searches will only return the right products if the customers happen to use the right language.  That’s a lot of lost revenue when those customers become frustrated and abandon the site.

To ensure that your customers can find what they’re looking for, you need a Natural Language search solution.  Natural Language search solutions find different versions of the same terms, so that pluralities and tenses can be managed.  For example, it wouldn’t make any difference if a customer used the term “mens”, “man”, “men” or “womens”, “women”, “ladies”.  An intelligent Natural Language search system will understand this automatically, without any input.

Scary Statistic #2

34% of eCommerce site searches don’t return useful results when users search for

a model number or misspell just a single character in the product title[2]

This scary statistic raises 2 important issues:

1. Misspelling

A good spell correction system is not just about correcting, but about finding words that occur in the product data which most closely match what the user described.

With EasyAsk’s intuitive search system, your product data becomes the look-up dictionary, meaning that your customers don’t have to match the spelling of your products exactly.

2. Complex Product Numbers

While most good search systems will find common misspellings, they lack the ability to deal effectively with part or model numbers.  Consumers who are taking the time to type in a part number are undoubtedly more ready to buy, so it is even more important that they are able to find what they are looking for.

EasyAsk has the ability to index part and model numbers comprehensively, indexing all variations to ensure that users can find the product, even if they forget or mistype part of the number.

EasyAsk’s ‘Part Number Expander’

A ‘Part Number Expander’ is especially useful for complex part numbers that include a combination of letters, numbers and other characters.  A user might forget whether characters were separated by a hyphen or a slash, or might omit letters on the end.

The Part Number Expander takes each part number and creates all the different versions of terms and inserts them into the searchable index so that if any of them are searched for, they will match to the product.  The code is broken up into parts and the separators are substituted in all combinations.

For example, if the part number is

123-HC/1345AB

the Part Number Expander would index:

123-HC-1345AB                                     123

123/HC/1345AB                                    123-HC

123/HC-1345AB                                    HC

123 HC 1345AB                         1345

…among many other combinations.

If a customer types a part/model number that is slightly different (for example, using spaces instead of hyphens) the correct product will still be found.

Ecklerscorvette.com demonstrates how the Part Number Expander can avoid no results for customers:

The part number for this disc brake set is 25-261256-1.  If a customer replaces the hyphens with spaces, the exact product is still returned.

Scary Statistic #3

If people have a negative experience on mobile, they’re 62% less likely to purchase from you in the future[3]

A negative experience will stay with prospects longer than a positive one, so if you want your mobile customers to become regular, you’ll need to make sure you’re not putting them off with a poor user experience compared to your desktop site.

Does your site have a mobile responsive design, so that the text, images, and menus change according to the screen size?  Furthermore, as your site adapts to a mobile size, do you show relevant categories and attributes, as opposed to just the first ones from the desktop site.

Customers using a mobile eCommerce site are more likely to use voice input, and therefore different word choices and sentence structures when speaking.  Your search system needs to be able to cope with voice input and the long-tail queries that are more likely on mobile.

Scary Statistic #4

46% of eCommerce sites have a “No Results Page” implementation that is essentially a dead-end for users, offering no more than a generic set of search tips[4]

Generating a generic ‘Sorry, no results matched your search’ message could be disastrous.  How many shoppers would stay on your site and how many would go straight to your competitor?

There are better ways to use a ‘No Results’ page, but what if it was possible to avoid searches returning no results in the first place?

It may be that a simple spell correction rectifies a ‘No Results’ search, but it is also important to monitor what EasyAsk calls ‘relaxed’ searches.  A relaxed search is one where the search system has modified the search by intelligently dropping (or ‘relaxing’) a term or terms.

For example, if a customer searched for a ‘Men’s black lace jacket’ and there are no products that match this description, we would want the results to show ‘Men’s black jackets’, rather than ‘Men’s lace jackets’ or ‘black lace jackets’, having relaxed the term ‘lace’.  Some searches will inevitably need to be modified and in this example, we want our search system to recognize that the gender stated in the search query is more important than the fabric.

If a user’s search has been modified (such as by spell correcting or relaxation) it is important to display a message explaining the changes to the original search.  For example, a user may search for “purple leather couches” and see the following message:

This user could have searched for leather couches in the first instance and spent a long time scrolling through the pages of results looking for any in purple.  But by using messaging in this way, the user is informed immediately and is more likely to use the search system again.

So-called intelligent ‘No Results’ pages will suggest searches based on what a user asked for, but truly intelligent search systems will change the search to find relevant products based on what the user asked for, even though there may not be an exact match.

Scary Statistic #5

It’s 5 to 25 times more expensive to acquire a new customer than it is to retain an existing one[5]

A worrying number of businesses still don’t use merchandising tools such as cross-sells and up-sells effectively on their eCommerce sites.  These businesses are leaving money on the table.

EasyAsk’s search solution includes a comprehensive set of merchandising controls.  With 9 predefined promotion types, banners and business rules, amongst other functions, the business user is in control of the products that are presented to the customer.

Scary Statistic #6

36% of autocomplete implementations on eCommerce websites do more harm than good[6]

Although basic autocomplete suggestions can be found on 82% of eCommerce sites, most sites do not take the opportunity to capitalize on merchandising while users are searching.

As a user types into a search box supported by EasyAsk, however, search suggestions and category suggestions will be displayed as well as the products themselves.

EasyAsk offers ‘Search as you Type’, or SAYT, because it means a whole lot more than just a list of suggested searches.  It means many possibilities for getting the right products in front of your customers.

SAYT progressively searches for and filters through text.  As a user types a query, suggestions, products, categories, and attributes are found and presented with each keystroke.  This allows a user to stop short of typing the entire word or phrase and find what they were looking for quicker.

Search suggestions as well as relevant and popular products, attributes, and categories can all be displayed in an easy to use interface.

Here is an example from EasyAsk customer personalizationmall.com:

The user is looking for a wedding gift and the letters ‘wedd’ have been typed into the search box on the desktop site so far.  The above drop-down panel is displayed without needing to execute the search.  Alongside the search suggestions in the right-hand column, a popular category to browse and 7 products are displayed, based on the first search prediction, ‘wedding’.  Scrolling through the search suggestions using the down and up arrows updates the real-time search and therefore the categories and products displayed.  Personalization Mall could have also configured their search so that it updates just by a mouse hover for a certain amount of time.

Typed characters are shown in bold in the list of suggestions, with the rest of the suggestion in normal type.  Often, autocomplete systems only show suggestions that begin with the typed characters, but we can see here that the suggestions can contain ‘wedd’ anywhere.

Presenting customers with the most relevant search, product and category suggestions is essential.

Conclusion

When your site search is performing well, customers won’t be scared off, and your business won’t become one of the statistics above.

Getting the right products in front of the customer as quickly as possible shortens the path to conversion, keeping your customers and your bank balance happy.

Follow the tips and best practices in EasyAsk’s recently released eGuide: “Preparing for the Holiday Season: A Guide for your eCommerce Site” to ensure your customers find the right products the first time.

[1] https://baymard.com/blog/ecommerce-search-report-and-benchmark

[2] https://baymard.com/blog/ecommerce-search-report-and-benchmark

[3] https://www.thinkwithgoogle.com/intl/en-gb/advertising-channels/mobile/few-tips-speed-your-mobile-site-and-tools-test-it/

[4] https://baymard.com/ecommerce-search/benchmark/page-types/no-search-results-page

[5] https://hbr.org/2014/10/the-value-of-keeping-the-right-customers

[6] https://www.smashingmagazine.com/2014/08/the-current-state-of-e-commerce-search/

The post Six Scary Site Search Statistics appeared first on EasyAsk.

The EasyAsk team recently visited the Windy City to be a part of IRCE at RetailX. This year’s event merged three shows into one, with the Internet Retailer Conference and Exhibition (IRCE) holding its 15^th annual event. RetailX attracted 20,000 attendees and more than 1,200 exhibiting brands, with EasyAsk proud to be counted among them.

We were able to impress visitors to the EasyAsk booth by demonstrating how outstanding EasyAsk’s Search and Navigation solution is. In case you missed the show, here is what one of the EasyAsk team could have shown you:

The North Face use EasyAsk to power their search so it’s a great example of some of the functionality. This voice search contains references to gender, price and color, as well as redundant words, such as ‘I’m looking for’. EasyAsk has the ability to cope with all of this information and returns the following results:

The EasyAsk engine has understood that ‘wife’ is a synonym for ‘womens’ and brings back only red, waterproof, ladies jackets priced less than $200. Although each of the 11 results are available in different colors, the thumbnail picture presented in the results is always in the red tone.

Contrast this successful search to the same search using Amazon.com:

Amazon’s engine cannot cope with the complexities of this query and the top suggested searches are for ‘I’m red waterproof’ and ‘I’m looking waterproof’.

It is clear that EasyAsk’s understanding of natural language is far superior to that of the eCommerce giant, Amazon. We think this is why the IRCE attendees were so impressed…

It’s worth taking a closer look when it comes to search and merchandising solutions. We’d love to show you further how EasyAsk out-performs Amazon and competes with Google. We can help you help your customers to find the right products, first time.

More information about IRCE can be found at www.irce.com

Take a closer look at EasyAsk’s search and merchandising solutions at www.easyask.com

The post IRCE @ RetailX 2019: How EasyAsk impressed prospective customers appeared first on EasyAsk.

What is Conversational Commerce?

The term ‘Conversational Commerce’ is used to describe the connection between Sellers and Buyers through messaging and chat applications. Many people now like to use chat systems to accomplish different tasks, but until fairly recently, this kind of interaction has predominantly been used for accessing support, such as customer services. But what if online retailers could use Conversational Commerce to provide a better, more personalized user experience for their shoppers?

Gartner research shows that by 2020:

• 30% of digital commerce revenue growth will be attributable to artificial intelligence technologies, such as those that power conversational commerce.

• 5% of all digital commerce transactions will come from a smart machine, such as those that power conversational commerce.

As people become more accustomed to interacting in a conversational way, they will be increasingly likely to want to use the same conversational mode when finding products.

Mirroring the Store Experience

When a customer walks into a physical store, the sales associate can ask a series of questions to narrow down the choice and focus in on the right products for the customer. In a human context, the sales associate has the necessary knowledge about the products to filter down to a few products.

This is what retailers should be trying to achieve online with Conversational Commerce.

When shopping online, a Conversational Commerce system should intelligently suggest appropriate ways of limiting a set of products. You need to make sure that your system has the intelligence to ask your customer the right questions.

Intelligent Questions and Responses

Let’s compare the in-store experience to one that could be achieved via Conversational Commerce:

Store Sales Associate                                                             Conversational AI-powered Search Engine

Just like with the in-store experience, the Conversational Commerce engine asks questions in a logical order and uses the responses to intelligently suggest attributes to refine the search further.

However your customers wish to interact with you, they should have a similar experience. But the Conversational Commerce experience shown above is only possible with the support of powerful, intuitive technology. This kind of advanced interaction requires an AI-based engine that can understand what the next set of attributes (prompts for the user) should be.

Are You Ready?

With the research showing the pattern for Artificial Intelligence to be the biggest influence on eCommerce over the coming years, it is time to make sure that your search system will be able to keep up. Whether using text or voice input, your customers will be progressively accustomed to chat-based interactions and will expect this format when buying online. By adopting an intuitive AI-based search system that can cope with the demands of Conversational Commerce, you will be able to fast-track your customers to the right products every time.

The post What is Conversational Commerce and what does it mean for eCommerce? appeared first on EasyAsk.

Search is essential in eCommerce – it is the gateway. Search is how consumers express what they want and if they don’t find what they want, they can’t buy it. People who choose to search have a far better idea of what they are looking for than those who browse, and are therefore far more likely to make a purchase. This means that understanding the intent of the shopper, however it is worded, is essential to conversion in eCommerce.

This is increasingly important with the proliferation of voice and the adoption of Natural Language searching by the general public. Home Voice Assistants, such as Alexa and Google Home, which were initially used for music, have gone a lot further and are getting smarter. According to new data from Loup Ventures, all Voice Assistants tested had improved in terms of questions answered correctly, with Siri making the most improvement.

From Keywords to Natural Language

Do people generally use Natural Language Search? Most might say, ‘no, not really’. But the truth is that everyone uses AI/NL search everyday, provided you have a computer or Smartphone. Do you ever search on the Internet?   You’re using Google or Bing or maybe Yahoo. All are AI/NL search engines and are incredibly smart. In 2011, Google was hit by a torpedo with the words ‘SIRI’ on it. Google immediately acquired an AI/NL search company and threw out years of keyword search technology: instantly obsolete.

Now people are becoming used to using Natural Language to talk to their smart home devices, but how do they interact with onsite search?

It’s a huge shame, but unfortunately people have become accustomed to the fact that eCommerce Search Engines can’t usually handle queries with more than one or two words, as most of them use obsolete keyword search. Even when using voice search, they assume that they need to think carefully about what the Search Engine is likely to understand, search, then probably rephrase.

But thanks to Home Voice Assistants, when users pick up their phones to use voice search, they’re increasingly likely to speak using Natural Language as opposed to keywords. When people speak, they naturally throw in extra words so an intuitive Natural Language Engine is needed to understand these longer-tail queries.

The beauty of having a true Natural Language Engine is that the customer can be understood as if they were talking to a Sales Associate. By leveraging Natural Language processing, the burden is on the Search Engine, not on the person talking.

If a customer walked into a North Face store, they might say to a sales associate, “I’m looking for a new waterproof jacket for my wife please.” Using these same words in a voice search on thenorthface.com brings back the following results:

The Search Engine has returned 18 highly relevant results.

Similarly, if a customer walked into a Timberland store, they might ask the Sales Associate, “Do you have any ladies field boots please?” In a voice search on timberland.com, exactly the right product is returned immediately:

Both Timberland and The North Face use EasyAsk to power their search. Their online customers can use voice search to describe exactly what they’re looking for as if they were in a store. These major brands have experienced dramatic revenue boosts between 37-70%.

EasyAsk’s search engine understands these types of queries. Will your Search Engine fail as more people search in this way, or will you be ready for the future because your Search Engine understands how people speak? Using a Natural Language Search Engine allows even small businesses to compete with the search quality of Amazon. You don’t need repetitive shopping behavior data to deliver fantastic results. Natural Language levels the playing field.

Encouraging your users to use voice search (perhaps by adding a microphone icon in the search bar) will mean that they provide more information about what they’re looking for. This in turn will mean that your users will see exactly the right results on the first page. Less time spent finding what they’re looking for results in a higher chance of conversion and higher revenue for your business.

The norm for voice interactions is changing. The future starts now.

“EasyAsk understands exactly what your customers are looking for. Delivering the right results, the first time. Everytime.”

Craig Bassin, CEO EasyAsk

The post Using AI-based Natural Language to improve eCommerce appeared first on EasyAsk.

And the implications for eCommerce

Automation of the home environment through Voice

It’s all about the Voice

Increasing numbers of people are using voice, thanks to the growing popularity of home voice assistants such as Alexa. We’re all aware of their invasion of our homes over recent years. They carry out a range of tasks for us daily, but they don’t always get it right. A bit of browsing using the hashtag ‘alexafail’ brings back some fun results:

Whether the voice assistants get it right or wrong, their popularity has boosted Voice technology into the mainstream. But it hasn’t always been about Alexa or Siri. Let’s take a look at how Voice technology has developed over the years with a brief timeline:

How does Voice affect eCommerce?

This proliferation of voice has a knock-on effect for eCommerce, with a study by OC & C Strategy Consultants forecasting a 1,999% increase in voice shopping, reaching $40 billion by 2022. This explosion in revenue will be primarily Mobile Voice UX on Smartphones.

Much of this predicted $40 billion will be attributed to Amazon as people use their devices more for shopping. Having said this, the ability to handle the same voice functionalities becomes even more important for other retailers, as users are increasingly used to voice shopping.

By 2021, mobile eCommerce sales are expected to account for 54% of total eCommerce sales.

Recent statistics even show that for certain brands, more than 75% of transactions are happening from mobile devices.

While more users are using mobile than desktop, conversion rates are dramatically lower for mobile, highlighting inconsistencies between user experience for desktop and mobile. The principle reasons for this m-commerce ‘gap’ are ease of use and product findability. The move towards using voice will allow users to get around that, as it’s easier to do than typing.

Voice recognition does not necessarily provide understanding.

Voice capabilities have been around for a long time, but the ability for the search engine to understand the context of the phrase, the intent of the shopper, was missing.  Now that users are more used to using voice with home assistants like Alexa, voice shopping will increase exponentially.

Businesses who are voice-enabled will see a dramatic increase in conversion rates.

In order to take advantage of these changes in how shoppers are choosing to buy online, you need technology that will understand the way people speak and be able to respond appropriately.

Understanding Language and Taking Action

Imagine that Joe Bloggs speaks to his fridge:

“I need a large glass of milk”

The technology installed in the fridge may be able to understand the request but can’t actually pour Joe that large glass of milk. This demonstrates perfectly a site that enables voice input but doesn’t have AI-based Natural Language Processing in order to intuitively deal with voice requests.

More and more users are expecting to be able to have a conversation when buying online. Using conversational search is therefore going to become more popular. Unfortunately, eCommerce search systems generally aren’t conversational.

EasyAsk’s solutions are on the cutting edge of understanding voice input and conversational search, providing the technology necessary to keep up with the proliferation of Voice.

“With all the fervor around voice-activated applications and machine intelligence, EasyAsk can help end-user organizations and vendors put an Apple Siri- or IBM Watson-like front end on…search applications.”

Gartner Cool Vendor: Analytics

Look out for future blog posts, where we will explore in more detail:

• Using AI-based Natural Language processing to improve eCommerce product search

• What is Conversational Search and what does it mean for eCommerce?

The post The Proliferation of Voice appeared first on EasyAsk.

According to Forrester, B2B eCommerce in the U.S. will hit $1.2 trillion by 2021, seeing a Compound Annual Growth Rate (CAGR) of 7.4% over the next four years. It is therefore essential that B2B businesses optimize the experience for their customers online. B2B sites have typically been known as less usable, but it is time for them to catch up with B2C and make it really easy for customers to do business. B2B eCommerce has unique complexities, which present unique challenges and therefore require a unique set of best practices. We will explore these challenges in our B2B blog series and offer advice and solutions to ensure that your B2B site delivers a superior experience.

Customer-Specific Catalogs

Providing customer-specific catalogs is a unique search requirement for B2B companies and can present unique challenges when it comes to eCommerce. But if it is done well, the personalized experience for your B2B customers will certainly pay dividends.

There are 4 possible situations that a B2B company may need to cater for:

1. The company has products that all customers can see and buy.
2. Certain customers are only authorized to buy from a particular range or category of products. When the customer searches, it is important that they can only view the products that are available for them to buy.
3. Similarly, the availability of products may be based on location. Customers in the North-East might only see products that are stocked in the North-East warehouse. Location of the stock also limits which products a customer should see.
4. There may be custom products produced by a company that are only available to one particular customer. For example, coffee companies need branded cups, so it is important that those custom products are in the supplier’s catalog, but cannot be purchased by anyone else.

Differing needs

Within a typical B2C system, the search system would search the entire catalog and show results, but in the above situations, the catalog indexing process needs to define who can buy which products so that the search system is able to present products in the results correctly. This can become very complex, which is why out-of-the-box search systems can’t handle it. EasyAsk provides a custom implementation for each customer so that any of the challenges described above can be handled.

A paper products manufacturer became an EasyAsk customer and now has certain products in their catalog that are marked with customer IDs, while other products are marked as being global. Their coffee store customer buys branded cups, so when they log in the customer ID is passed to EasyAsk, who pass back only the right products.

The implementation of customer-specific catalogs varies completely between B2B customers, so it isn’t possible to just use defined rules; the flexibility to support all different circumstances is necessary.

B2B Personalization

Being able to search and see personalized results based on pre-negotiated terms is imperative for your B2B customers. They should never see a product that isn’t available for them to buy. Your search system should be aware of product availability for any given customer. Is your search system up to this?

EasyAsk is up to the challenge

EasyAsk offers the only site search and merchandising tool designed exclusively for the rigors and challenges of B2B eCommerce. EasyAsk can be configured for any platform, either commercial or built in-house.

EasyAsk has long served the B2B customer segment and over 200 B2B distributors have chosen us to power their B2B eCommerce sites, including: Aramark, Demco, Kaman Industries, Tacoma Screw, and Crown Packaging.

The post B2B Series Challenge 3: Customer-Specific Catalogs appeared first on EasyAsk.

According to Forrester, B2B eCommerce in the U.S. will hit $1.2 trillion by 2021, seeing a Compound Annual Growth Rate (CAGR) of 7.4% over the next four years. It is therefore essential that B2B businesses optimize the experience for their customers online. B2B sites have typically been known as less usable, but it is time for them to catch up with B2C and make it really easy for customers to do business. B2B eCommerce has unique complexities, which present unique challenges and therefore require a unique set of best practices. We will explore these challenges in our B2B blog series and offer advice and solutions to ensure that your B2B site delivers a superior experience.

The complex world of part numbers

B2B shoppers are much more likely to search using specific product numbers. B2B product numbers can inherently be more complex and therefore be easily mistyped or forgotten. B2B product numbers often take the following forms:

• Groups of numbers, letters and other characters, such as hyphens and slashes
• Long numbers with some kind of structure, e.g. 123 4567 89

If your customers can’t find parts or part numbers on your eCommerce site, they will be calling your customer services, which is expensive to you and inconvenient for them. Worse still, they may take their business to your competitors.

Complex product numbers affect many businesses, but some more than others. Take print cartridges for example. It is important for customers to find the exact cartridges that are compatible with their printer. We used a print cartridge site to search for cartridges for a Kyocera Ecosys printer:

As you can see, the printer model numbers are made up of very similar combinations of letters and numbers. Scrolling through the list revealed 3 times this number of results.

A mechanic who needs a car part for a job the next day is probably not going to be the person typing the part number into a search box. Someone in the office will be asked to order the part, and when they can’t find the part due a slight error in the part number, they have to get on the phone to the supplier. This means it costs the car parts vendor more to service the customer.

The challenge for B2B eCommerce sites is clear, but what can be implemented to avoid the need for calls to a customer service line or using a competitor’s site?

EasyAsk’s ‘Begins Expander’

A ‘Begins Expander’ is EasyAsk functionality that attempts to predict the most common inaccuracies a ‘searcher’ will make when attempting a product number search using EasyAsk’s advanced autocomplete functionality. A Begins Expander can be set up for any field, but is most relevant for a part or product number field. It can be configured to start or finish on any number of characters and it will index the part number as the start parameter, then plus 1 character, then plus 2 characters, and so on until the end parameter. For example, if the product code is 1234567 with the parameters 3 and 7, the Begins Expander would index:

123

1234

12345

123456

1234567

This is an extremely effective technique as it is incredibly helpful to a searcher that doesn’t remember the entire product code, they can just start to type the beginning of the code and EasyAsk’s Search As You Type (SAYT)/autocomplete will show all possible product codes.

                         enasco.com begins to suggest product numbers after 2 characters are typed

EasyAsk’s ‘Part Number Expander’

A ‘Part Number Expander’ is especially useful for complex part numbers that include a combination of letters, numbers and other characters. A user might forget whether characters were separated by a hyphen or a slash, or might omit letters on the end. The Part Number Expander takes each part number and creates all the different versions of terms and inserts them into the searchable index so that if any of them are searched for, they will match to the product. The code is broken up into parts and the separators are substituted in all combinations. For example, if the part number is

123-HC/1345AB

the Part Number Expander would index:

123-HC-1345AB                                                                        123

123/HC/1345AB                                                                       123-HC

123/HC-1345AB                                                                        HC

123 HC 1345AB                                                                          1345

AB

amongst many other combinations.

If a customer types a part number that is slightly different (for example, using spaces instead of hyphens) the correct part will still be found.

Ecklerscorvette.com demonstrates how the Part Number Expander can avoid no results for customers:

The part number for this disc brake set is 25-261256-1. If a customer replaces the hyphens with spaces, the exact product is still returned.

We’ve got you covered

The 2 capabilities above are just a part of the EasyAsk experience. There is no effort to use them; no extra coding is needed because EasyAsk does it automatically. When your data is read the part numbers are broken up based on rules that are built internally. EasyAsk will scan the part or product numbers, and wherever they change from or to alphabetic or numeric, they will be treated as a separate part.

EasyAsk making all the difference

An EasyAsk customer recently encountered problems with product number variations and saw outstanding results when they implemented EasyAsk’s software:

EasyAsk was approached by a library and schools supplier who was a printed catalog business before launching online. Their web system only used the internal product codes, which were just numbers, whereas in the catalogs, each number had a 2-character prefix with a hyphen. So, if the internal product code was 1234, in the catalog it could be rx-1234, kb-1234, depending on the different catalogs that were issued throughout the year.

The company found that a teacher or librarian would look in the catalog, find the product they required, and then when they tried to find the product online, the search failed every time. Because the printed catalog didn’t match what was in the database, the user wouldn’t be able to find what they wanted, and so would call the helpdesk who would take the order over the phone.

When the company implemented EasyAsk software, they provided EasyAsk with their data, plus a list of catalog prefix and suffix codes for those printed catalogs that were in circulation. As part of the build process, EasyAsk indexed product numbers with and without prefixes and suffixes, and with and without hyphens. From the moment that the EasyAsk system went live, customers were able to find the products using the catalog codes and calls to the call center were reduced by half.

EasyAsk made a huge difference in this case, can your website do this?

B2B Experts

EasyAsk offers the only merchandising tool designed exclusively for the rigors and challenges of B2B eCommerce. EasyAsk can be configured for any platform, either commercial or built in-house.

EasyAsk has long served the B2B customer segment and over 200 B2B distributors have chosen us to power their B2B eCommerce sites, including: HD Supply, Aramark, Alphabroder, Demco, Kaman Industries, Tacoma Screw, and Crown Packaging.

The post B2B Series Challenge 2: Complex Part or Product Numbers appeared first on EasyAsk.

No Results, too many results, inconsistency and lack of communication… while there are many reasons for a customer to give up on your eCommerce site’s search the underlying problem is they couldn’t find the products they were searching for.

With average cart abandonment rates hovering around 65%, don’t let your company’s search be the monster that scares business away.

Surely any customer would be scared off by over 50,000 results on Amazon.com for a ladies Halloween costume. The first products aren’t even a good match for the search.

Follow the tips and best practices in EasyAsk’s recent blog series: “Optimizing your eCommerce site for the Holiday Season” to ensure your customers find the right products the first time.

The post Is your search scaring off your customers? appeared first on EasyAsk.