You are here: Home » Management » Features » Strategy
Business Standard

Browse less, find more

More and more online businesses are engineering targeted suggestions to help consumers make quick purchasing decisions

Ankita Rai 

You don't like a helicopter salesperson, but you like to be addressed by name when you enter your favourite store. You have little time to browse through the products randomly piled across the shopfloor, but you hate to be prompted along the way. You are a fastidious shopper, but you don't mind relevant suggestions while negotiating the aisles at the store.

This is one area where brick and mortar businesses have had an edge over - the art of gently nudging the towards things that might not have been on her shopping list in the first place. Now have closed that gap - the smart ones are working overtime to incorporate data-driven merchandising to provide targeted and relevant offerings that help visitors make a purchase. They are providing navigation that enables consumers to easily refine product choices by important product attributes. In short, e-retailers have entered the domain that was once the exclusive preserve of brick and mortar stores: that of guided discovery.



Tools to drive engagement
DRIVE ENGAGEMENT
LIVE CHAT: There’s no better way to get a query resolved than by asking someone while you are on the site  
ON-SITE PUSH NOTIFICATIONS: Stores have peak hours and dull hours. Use tiny, little nudges to make virtual walk-ins aware of offers
VOICE OF THE CUSTOMER: satisfaction surveys post purchase are a sure shot way of getting feedback and insights
SELLING INTO NON-PRIMARY CATEGORIES: Use product adjacency analysis to create a basket of products that are typically bought together. For example, electronics gadgets and graphic novels. Research shows that the same demographic tends to purchase them
SEGMENTATION: Crunch data to create profiles of the customers. For example, a person who buys electronics and comments on technology blogs can be tagged as part of a segment ‘early technology adopters’. Other such segments can be created
With inputs from Avlesh Singh, CEO, WebEngage and Dipayan Chakraborty, director, Delivery and Client Relations, TEG Analytics

Mind you, the drive towards guided discovery - or narrow personalisation as some experts call it - is nothing new. It has actually taken a few steps beyond deploying the rudimentary push tools in recent years. Says Kanika Mathur, MD, Razorfish India, "Guided discovery has moved beyond creating cookie cutter newsletters, which contain deals around the logged in customer's 'browse' history at a portal. The onus is on 'one-on-one' offers to a customer, which are created by leveraging deep data that is available about a and delivering a personalised experience across all touch points, including web, mobile, tablet, in store." No wonder, smart companies are experimenting with virtual assistants who play the role of the store manager or sales person, typically seen otherwise in brick and mortar stores.

Guided discovery takes on particular importance for retailers seeking to draw millennials to These shoppers make fewer clicks; they are aware that retailers have the ability to collect and use personal information to customise the shopping experience. While they don't want it taken too far, there is an expectation that this information can be enhanced for a better online shopping experience.

Now look at the picture from the retailer's point of view: Given that shopping online is now a mainstream activity, the challenge is no longer getting new customers but rather driving incremental money from the existing base of shoppers. "The time has come to focus aggressively on technology, particularly on multichannel and personalisation technologies," says an Oracle White Paper, "How to Win Online: Advanced Personalisation in "

The obvious question: as a retailer why should you want to do 50 per cent of the customer's job? Experts say you can increase your conversion rate and average order value if you engage the target within first few seconds of her logging in on your website because beyond that her eyes are trained to wander off. Also she is more likely to buy when she is presented with less but accurate choices.

Look globally and you will find the answer in numbers. Take Amazon and Netflix, for instance. These companies get billions of visitors at their sites every month and are able to guide those visitors successfully. Amazon gets 35 per cent of its sales from the recommendation it makes. Similarly, 75 per cent of purchases on Netflix, a movie streaming company, are courtesy guided search or recommendation.

Now take the case of US-based big data start-up True Fit, which takes cues from Amazon and Netflix analytical models to help consumers figure out whether or not the apparel they want to buy online will fit them. It leverages the growing database of the world's top apparel, footwear and consumer fit data to help consumers, brands and retailers find each other.

Data curation also offers greater insights about popularity and trends. Retailers worldwide have begun increasingly crowd sourcing data via forums like Pinterest. "Curation of social data (foursquare, Twitter, YouTube activity) is leading to interesting initiatives," Mathur adds. "For instance, Japanese casual wear brand Uniqlo successfully rolled out Pop Up stores at various locations based on the number of people who were commenting on social shares of outfits tried online or in store in a particular area. Such strategies not only influence the offers but also the entire interaction design of the consumer experience." Mathur adds.

While that sort of sophistication is still some way off for online in India, work is already underway.

How it works now
At present, most companies in India are using traditional recommendation engines to display items frequently bought and push promotions. Bangalore-based Myntra customises its home page according to a customer's profile. So for every visitor the home page is different. Says Prasad Kompalli, chief officer, Myntra, "We try to match the physical experience a is used to in a brick and mortar store. We use personalised recommendations and emailers. The open rate of these customised emails is three times more than regular emails."

Prasad says the large assortments offered by companies can be both an advantage and a disadvantage as the may not be able to find the product she is looking for. To help consumers navigate and find offers, has a personalised offer button with every listed product. "We have built in-house algorithms based on how many times a consumer added a product to her wish-list or cart. With data analytics, our conversion rate, which is calculated as the number of orders divided by number of visits, increased between 30 and 50 per cent in 2013."

Delhi-based Jabong uses a hybrid model, a combination of content and collaboration, to enable guided discovery. Says Praveen Sinha, co-founder and managing director, "The key enablers to guided discovery are, first, the recommendation engine, which is based on multiple logic (someone who bought this, also bought this), and second, fine-tuning of search engine, and third, the history of purchase."

Snapdeal, a Dehi-based marketplace, uses data gathered from its website and through e-mailers to address buyers. "We use analytics differently for both these channels. On the Snapdeal site, for instance, if two customers come they would see very different home pages. Their home screen listings get decided from what they browsed in their last visit to the site," says Ankit Khanna, VP, product management, Snapdeal. The portal collects data from user behaviour including navigation, preferences and clicks on its systems. "We are crunching 15 million data points every three hours," says Khanna, adding that all the analytics is done in-house. Search is the other place where the portal has implemented machine learning.

poster boy Flipkart says to enable guided discovery, e-retailers must first think like consumers. "It is important to show the right products to the user the moment she enters the site," says Saranagati Chatterjee, VP (products). "Depending on the entry point, we work on providing right suggestion in the search results. There are tools like 'filter' and 'compare' to show relevant merchandising and recommendations.'' On Flipkart, the approach varies according to the category of product a user is browsing. "Electronics is one category where consumers need hand-holding before making a purchase decision," says Chatterjee. With consumers now shifting to mobile phones and tablets for shopping, Flipkart enables offline browsing to help consumer save on data usage. "It is even better to collect consumer data on cellphones because it comes with a unique device identity," Chatterjee adds.

"Guided discovery can also work as a delightful discovery because sometimes the consumer is not sure what she is looking for. In such case e-retailers should offer adjacent products such as a bouquet of flowers or car rental deals along with an evening dress," says Sanjay Sethi, CEO, Shopclues.

Online deal sites have entirely different challenges when it comes to guided discovery. The key here is local services. While selling deals, it is important to understand the deal structure that gets the best response. Says Ankur Warikoo, regional head, APAC, emerging countries, Groupon, "We plan to bring location-specific smart deals to India. It identifies who is browsing the website from which location. So the consumer gets relevant deals in real-time."

For a gifting site like Giftease.com, guided discovery becomes complex because the is usually buying the gift for someone else. Hence, recommendation engines that use cookies or IPs for making personalised recommendations are of limited use. "We use past transaction data like the occasion, relationship and item attributes like price, category to refine our recommendations," says Vivek Mathur, CEO, Giftease Technologies.

Social shopping platform LimeRoad.com uses scrapbooks and a flip-book magazine to helps users browse through various products on the site. Users have the option of seeing a live feed of products that other users have viewed. "Our scrapbooking community has curated more than 60,000 scrapbook looks, which are 100 per cent unique to our site," says Suchi Mukherjee, CEO & co-founder, adding that recommendations can help increase the average ticket size of the purchase between 75 per cent and 100 per cent. "conversion is always high if the recommendation tool is working properly, though unlike in gadgets and books, recommendations in the lifestyle segment are mostly led by the visual appeal of a product," she adds.

How to make it better
Most recommendation engines are based on collaborative filtering, which looks at the affinity between two products in the same category or between two categories. It has limitations because it only looks at past purchase history of a consumer. So if a consumer bought a striped shirt or searched for it, she will keep getting recommendations even if she has closed that deal. So just using internal data doesn't work. "To make the right kind of guided discovery, one must combine internal data generated inside the company with external data," says Srikant Sastri, co-founder, Crayon Data. "External data can be generated by trailing the internet reviews and social networking for movies, books, shopping, travel etc. Building cross-category connection on the basis of external and internal data can make recommendations richer and powerful."

Traditional recommendation engine often works like spam. Guided search should work on pull mechanism rather than push mechanism. So if there is a shopping application, you should ideally get the personalised choices only when you open the app.

The other issue is dealing with cart abandonment. One way to tackle this is to offer real-time coupon codes - such as a 20 per cent off if you shop in the next 15 minutes. "E-retailers need real-time, onsite and personalised coupons. With the help of analytics, it is easy to know how many people visited the site, from which location and when they bounced off," points out Avlesh Singh, co-founder & CEO at WebEngage, which provides on-site engagement tools to major e-retailers such as Flipkart, and Jabong.

Webengage enables websites to target users leaving sites through leave intent-based targeting that detects mouse movements of visitors. A pre-configured survey/notification pops up the moment a user is approaching the browser's close button or moving to a different tab in the browser window.

Komli Media, a digital advertising technology platform, uses a client's product catalogue feed to create personalised ads to retarget users. For instance, a user expresses her interest in buying a product through com but does not make a transaction. Komli's remarketing demand side platform monitors this and when the visits any social site next time, the ad of this particular product pops up alongside her profile. This increases the company's chance to convert the visitor into a Says Ashwin Puri, VP, remarketing & mobile, Komli Media, "We match user's browse history and use that data to surface relevant ads."

These are just some of the ways smart e-retailers are using technology and the data at their disposal to hang on to customers and urge them to spend more. The thing to remember here is to stop before you become pushy. Experts contend that while analytics can help in targeting the consumers and customising options for her, e-retailers must understand that customers want to be guided, but they don't want to sacrifice the experience of shopping. Also remember, narrow personalisation is not just something you sprinkle on your website landing page. It goes beyond inserting a name at the top of a web page or e-mail. It should be a business -one that takes into account not only who the is but what she likes, what she doesn't like, when she wants to hear from you, and when she doesn't. To be most effective, guided discovery is ideally managed from a single integrated platform that makes it easy for you to deliver a relevant experience across all interactions, through the entire lifecycle.

RECOMMENDED FOR YOU

Browse less, find more

More and more online businesses are engineering targeted suggestions to help consumers make quick purchasing decisions

More and more online businesses are engineering targeted suggestions to help consumers make quick purchasing decisions You don't like a helicopter salesperson, but you like to be addressed by name when you enter your favourite store. You have little time to browse through the products randomly piled across the shopfloor, but you hate to be prompted along the way. You are a fastidious shopper, but you don't mind relevant suggestions while negotiating the aisles at the store.

This is one area where brick and mortar businesses have had an edge over - the art of gently nudging the towards things that might not have been on her shopping list in the first place. Now have closed that gap - the smart ones are working overtime to incorporate data-driven merchandising to provide targeted and relevant offerings that help visitors make a purchase. They are providing navigation that enables consumers to easily refine product choices by important product attributes. In short, e-retailers have entered the domain that was once the exclusive preserve of brick and mortar stores: that of guided discovery.

Tools to drive engagement
DRIVE ENGAGEMENT
LIVE CHAT: There’s no better way to get a query resolved than by asking someone while you are on the site  
ON-SITE PUSH NOTIFICATIONS: Stores have peak hours and dull hours. Use tiny, little nudges to make virtual walk-ins aware of offers
VOICE OF THE CUSTOMER: satisfaction surveys post purchase are a sure shot way of getting feedback and insights
SELLING INTO NON-PRIMARY CATEGORIES: Use product adjacency analysis to create a basket of products that are typically bought together. For example, electronics gadgets and graphic novels. Research shows that the same demographic tends to purchase them
SEGMENTATION: Crunch data to create profiles of the customers. For example, a person who buys electronics and comments on technology blogs can be tagged as part of a segment ‘early technology adopters’. Other such segments can be created
With inputs from Avlesh Singh, CEO, WebEngage and Dipayan Chakraborty, director, Delivery and Client Relations, TEG Analytics

Mind you, the drive towards guided discovery - or narrow personalisation as some experts call it - is nothing new. It has actually taken a few steps beyond deploying the rudimentary push tools in recent years. Says Kanika Mathur, MD, Razorfish India, "Guided discovery has moved beyond creating cookie cutter newsletters, which contain deals around the logged in customer's 'browse' history at a portal. The onus is on 'one-on-one' offers to a customer, which are created by leveraging deep data that is available about a and delivering a personalised experience across all touch points, including web, mobile, tablet, in store." No wonder, smart companies are experimenting with virtual assistants who play the role of the store manager or sales person, typically seen otherwise in brick and mortar stores.

Guided discovery takes on particular importance for retailers seeking to draw millennials to These shoppers make fewer clicks; they are aware that retailers have the ability to collect and use personal information to customise the shopping experience. While they don't want it taken too far, there is an expectation that this information can be enhanced for a better online shopping experience.

Now look at the picture from the retailer's point of view: Given that shopping online is now a mainstream activity, the challenge is no longer getting new customers but rather driving incremental money from the existing base of shoppers. "The time has come to focus aggressively on technology, particularly on multichannel and personalisation technologies," says an Oracle White Paper, "How to Win Online: Advanced Personalisation in "

The obvious question: as a retailer why should you want to do 50 per cent of the customer's job? Experts say you can increase your conversion rate and average order value if you engage the target within first few seconds of her logging in on your website because beyond that her eyes are trained to wander off. Also she is more likely to buy when she is presented with less but accurate choices.

Look globally and you will find the answer in numbers. Take Amazon and Netflix, for instance. These companies get billions of visitors at their sites every month and are able to guide those visitors successfully. Amazon gets 35 per cent of its sales from the recommendation it makes. Similarly, 75 per cent of purchases on Netflix, a movie streaming company, are courtesy guided search or recommendation.

Now take the case of US-based big data start-up True Fit, which takes cues from Amazon and Netflix analytical models to help consumers figure out whether or not the apparel they want to buy online will fit them. It leverages the growing database of the world's top apparel, footwear and consumer fit data to help consumers, brands and retailers find each other.

Data curation also offers greater insights about popularity and trends. Retailers worldwide have begun increasingly crowd sourcing data via forums like Pinterest. "Curation of social data (foursquare, Twitter, YouTube activity) is leading to interesting initiatives," Mathur adds. "For instance, Japanese casual wear brand Uniqlo successfully rolled out Pop Up stores at various locations based on the number of people who were commenting on social shares of outfits tried online or in store in a particular area. Such strategies not only influence the offers but also the entire interaction design of the consumer experience." Mathur adds.

While that sort of sophistication is still some way off for online in India, work is already underway.

How it works now
At present, most companies in India are using traditional recommendation engines to display items frequently bought and push promotions. Bangalore-based Myntra customises its home page according to a customer's profile. So for every visitor the home page is different. Says Prasad Kompalli, chief officer, Myntra, "We try to match the physical experience a is used to in a brick and mortar store. We use personalised recommendations and emailers. The open rate of these customised emails is three times more than regular emails."

Prasad says the large assortments offered by companies can be both an advantage and a disadvantage as the may not be able to find the product she is looking for. To help consumers navigate and find offers, has a personalised offer button with every listed product. "We have built in-house algorithms based on how many times a consumer added a product to her wish-list or cart. With data analytics, our conversion rate, which is calculated as the number of orders divided by number of visits, increased between 30 and 50 per cent in 2013."

Delhi-based Jabong uses a hybrid model, a combination of content and collaboration, to enable guided discovery. Says Praveen Sinha, co-founder and managing director, "The key enablers to guided discovery are, first, the recommendation engine, which is based on multiple logic (someone who bought this, also bought this), and second, fine-tuning of search engine, and third, the history of purchase."

Snapdeal, a Dehi-based marketplace, uses data gathered from its website and through e-mailers to address buyers. "We use analytics differently for both these channels. On the Snapdeal site, for instance, if two customers come they would see very different home pages. Their home screen listings get decided from what they browsed in their last visit to the site," says Ankit Khanna, VP, product management, Snapdeal. The portal collects data from user behaviour including navigation, preferences and clicks on its systems. "We are crunching 15 million data points every three hours," says Khanna, adding that all the analytics is done in-house. Search is the other place where the portal has implemented machine learning.

poster boy Flipkart says to enable guided discovery, e-retailers must first think like consumers. "It is important to show the right products to the user the moment she enters the site," says Saranagati Chatterjee, VP (products). "Depending on the entry point, we work on providing right suggestion in the search results. There are tools like 'filter' and 'compare' to show relevant merchandising and recommendations.'' On Flipkart, the approach varies according to the category of product a user is browsing. "Electronics is one category where consumers need hand-holding before making a purchase decision," says Chatterjee. With consumers now shifting to mobile phones and tablets for shopping, Flipkart enables offline browsing to help consumer save on data usage. "It is even better to collect consumer data on cellphones because it comes with a unique device identity," Chatterjee adds.

"Guided discovery can also work as a delightful discovery because sometimes the consumer is not sure what she is looking for. In such case e-retailers should offer adjacent products such as a bouquet of flowers or car rental deals along with an evening dress," says Sanjay Sethi, CEO, Shopclues.

Online deal sites have entirely different challenges when it comes to guided discovery. The key here is local services. While selling deals, it is important to understand the deal structure that gets the best response. Says Ankur Warikoo, regional head, APAC, emerging countries, Groupon, "We plan to bring location-specific smart deals to India. It identifies who is browsing the website from which location. So the consumer gets relevant deals in real-time."

For a gifting site like Giftease.com, guided discovery becomes complex because the is usually buying the gift for someone else. Hence, recommendation engines that use cookies or IPs for making personalised recommendations are of limited use. "We use past transaction data like the occasion, relationship and item attributes like price, category to refine our recommendations," says Vivek Mathur, CEO, Giftease Technologies.

Social shopping platform LimeRoad.com uses scrapbooks and a flip-book magazine to helps users browse through various products on the site. Users have the option of seeing a live feed of products that other users have viewed. "Our scrapbooking community has curated more than 60,000 scrapbook looks, which are 100 per cent unique to our site," says Suchi Mukherjee, CEO & co-founder, adding that recommendations can help increase the average ticket size of the purchase between 75 per cent and 100 per cent. "conversion is always high if the recommendation tool is working properly, though unlike in gadgets and books, recommendations in the lifestyle segment are mostly led by the visual appeal of a product," she adds.

How to make it better
Most recommendation engines are based on collaborative filtering, which looks at the affinity between two products in the same category or between two categories. It has limitations because it only looks at past purchase history of a consumer. So if a consumer bought a striped shirt or searched for it, she will keep getting recommendations even if she has closed that deal. So just using internal data doesn't work. "To make the right kind of guided discovery, one must combine internal data generated inside the company with external data," says Srikant Sastri, co-founder, Crayon Data. "External data can be generated by trailing the internet reviews and social networking for movies, books, shopping, travel etc. Building cross-category connection on the basis of external and internal data can make recommendations richer and powerful."

Traditional recommendation engine often works like spam. Guided search should work on pull mechanism rather than push mechanism. So if there is a shopping application, you should ideally get the personalised choices only when you open the app.

The other issue is dealing with cart abandonment. One way to tackle this is to offer real-time coupon codes - such as a 20 per cent off if you shop in the next 15 minutes. "E-retailers need real-time, onsite and personalised coupons. With the help of analytics, it is easy to know how many people visited the site, from which location and when they bounced off," points out Avlesh Singh, co-founder & CEO at WebEngage, which provides on-site engagement tools to major e-retailers such as Flipkart, and Jabong.

Webengage enables websites to target users leaving sites through leave intent-based targeting that detects mouse movements of visitors. A pre-configured survey/notification pops up the moment a user is approaching the browser's close button or moving to a different tab in the browser window.

Komli Media, a digital advertising technology platform, uses a client's product catalogue feed to create personalised ads to retarget users. For instance, a user expresses her interest in buying a product through com but does not make a transaction. Komli's remarketing demand side platform monitors this and when the visits any social site next time, the ad of this particular product pops up alongside her profile. This increases the company's chance to convert the visitor into a Says Ashwin Puri, VP, remarketing & mobile, Komli Media, "We match user's browse history and use that data to surface relevant ads."

These are just some of the ways smart e-retailers are using technology and the data at their disposal to hang on to customers and urge them to spend more. The thing to remember here is to stop before you become pushy. Experts contend that while analytics can help in targeting the consumers and customising options for her, e-retailers must understand that customers want to be guided, but they don't want to sacrifice the experience of shopping. Also remember, narrow personalisation is not just something you sprinkle on your website landing page. It goes beyond inserting a name at the top of a web page or e-mail. It should be a business -one that takes into account not only who the is but what she likes, what she doesn't like, when she wants to hear from you, and when she doesn't. To be most effective, guided discovery is ideally managed from a single integrated platform that makes it easy for you to deliver a relevant experience across all interactions, through the entire lifecycle.
image
Business Standard
177 22

Browse less, find more

More and more online businesses are engineering targeted suggestions to help consumers make quick purchasing decisions

You don't like a helicopter salesperson, but you like to be addressed by name when you enter your favourite store. You have little time to browse through the products randomly piled across the shopfloor, but you hate to be prompted along the way. You are a fastidious shopper, but you don't mind relevant suggestions while negotiating the aisles at the store.

This is one area where brick and mortar businesses have had an edge over - the art of gently nudging the towards things that might not have been on her shopping list in the first place. Now have closed that gap - the smart ones are working overtime to incorporate data-driven merchandising to provide targeted and relevant offerings that help visitors make a purchase. They are providing navigation that enables consumers to easily refine product choices by important product attributes. In short, e-retailers have entered the domain that was once the exclusive preserve of brick and mortar stores: that of guided discovery.

Tools to drive engagement
DRIVE ENGAGEMENT
LIVE CHAT: There’s no better way to get a query resolved than by asking someone while you are on the site  
ON-SITE PUSH NOTIFICATIONS: Stores have peak hours and dull hours. Use tiny, little nudges to make virtual walk-ins aware of offers
VOICE OF THE CUSTOMER: satisfaction surveys post purchase are a sure shot way of getting feedback and insights
SELLING INTO NON-PRIMARY CATEGORIES: Use product adjacency analysis to create a basket of products that are typically bought together. For example, electronics gadgets and graphic novels. Research shows that the same demographic tends to purchase them
SEGMENTATION: Crunch data to create profiles of the customers. For example, a person who buys electronics and comments on technology blogs can be tagged as part of a segment ‘early technology adopters’. Other such segments can be created
With inputs from Avlesh Singh, CEO, WebEngage and Dipayan Chakraborty, director, Delivery and Client Relations, TEG Analytics

Mind you, the drive towards guided discovery - or narrow personalisation as some experts call it - is nothing new. It has actually taken a few steps beyond deploying the rudimentary push tools in recent years. Says Kanika Mathur, MD, Razorfish India, "Guided discovery has moved beyond creating cookie cutter newsletters, which contain deals around the logged in customer's 'browse' history at a portal. The onus is on 'one-on-one' offers to a customer, which are created by leveraging deep data that is available about a and delivering a personalised experience across all touch points, including web, mobile, tablet, in store." No wonder, smart companies are experimenting with virtual assistants who play the role of the store manager or sales person, typically seen otherwise in brick and mortar stores.

Guided discovery takes on particular importance for retailers seeking to draw millennials to These shoppers make fewer clicks; they are aware that retailers have the ability to collect and use personal information to customise the shopping experience. While they don't want it taken too far, there is an expectation that this information can be enhanced for a better online shopping experience.

Now look at the picture from the retailer's point of view: Given that shopping online is now a mainstream activity, the challenge is no longer getting new customers but rather driving incremental money from the existing base of shoppers. "The time has come to focus aggressively on technology, particularly on multichannel and personalisation technologies," says an Oracle White Paper, "How to Win Online: Advanced Personalisation in "

The obvious question: as a retailer why should you want to do 50 per cent of the customer's job? Experts say you can increase your conversion rate and average order value if you engage the target within first few seconds of her logging in on your website because beyond that her eyes are trained to wander off. Also she is more likely to buy when she is presented with less but accurate choices.

Look globally and you will find the answer in numbers. Take Amazon and Netflix, for instance. These companies get billions of visitors at their sites every month and are able to guide those visitors successfully. Amazon gets 35 per cent of its sales from the recommendation it makes. Similarly, 75 per cent of purchases on Netflix, a movie streaming company, are courtesy guided search or recommendation.

Now take the case of US-based big data start-up True Fit, which takes cues from Amazon and Netflix analytical models to help consumers figure out whether or not the apparel they want to buy online will fit them. It leverages the growing database of the world's top apparel, footwear and consumer fit data to help consumers, brands and retailers find each other.

Data curation also offers greater insights about popularity and trends. Retailers worldwide have begun increasingly crowd sourcing data via forums like Pinterest. "Curation of social data (foursquare, Twitter, YouTube activity) is leading to interesting initiatives," Mathur adds. "For instance, Japanese casual wear brand Uniqlo successfully rolled out Pop Up stores at various locations based on the number of people who were commenting on social shares of outfits tried online or in store in a particular area. Such strategies not only influence the offers but also the entire interaction design of the consumer experience." Mathur adds.

While that sort of sophistication is still some way off for online in India, work is already underway.

How it works now
At present, most companies in India are using traditional recommendation engines to display items frequently bought and push promotions. Bangalore-based Myntra customises its home page according to a customer's profile. So for every visitor the home page is different. Says Prasad Kompalli, chief officer, Myntra, "We try to match the physical experience a is used to in a brick and mortar store. We use personalised recommendations and emailers. The open rate of these customised emails is three times more than regular emails."

Prasad says the large assortments offered by companies can be both an advantage and a disadvantage as the may not be able to find the product she is looking for. To help consumers navigate and find offers, has a personalised offer button with every listed product. "We have built in-house algorithms based on how many times a consumer added a product to her wish-list or cart. With data analytics, our conversion rate, which is calculated as the number of orders divided by number of visits, increased between 30 and 50 per cent in 2013."

Delhi-based Jabong uses a hybrid model, a combination of content and collaboration, to enable guided discovery. Says Praveen Sinha, co-founder and managing director, "The key enablers to guided discovery are, first, the recommendation engine, which is based on multiple logic (someone who bought this, also bought this), and second, fine-tuning of search engine, and third, the history of purchase."

Snapdeal, a Dehi-based marketplace, uses data gathered from its website and through e-mailers to address buyers. "We use analytics differently for both these channels. On the Snapdeal site, for instance, if two customers come they would see very different home pages. Their home screen listings get decided from what they browsed in their last visit to the site," says Ankit Khanna, VP, product management, Snapdeal. The portal collects data from user behaviour including navigation, preferences and clicks on its systems. "We are crunching 15 million data points every three hours," says Khanna, adding that all the analytics is done in-house. Search is the other place where the portal has implemented machine learning.

poster boy Flipkart says to enable guided discovery, e-retailers must first think like consumers. "It is important to show the right products to the user the moment she enters the site," says Saranagati Chatterjee, VP (products). "Depending on the entry point, we work on providing right suggestion in the search results. There are tools like 'filter' and 'compare' to show relevant merchandising and recommendations.'' On Flipkart, the approach varies according to the category of product a user is browsing. "Electronics is one category where consumers need hand-holding before making a purchase decision," says Chatterjee. With consumers now shifting to mobile phones and tablets for shopping, Flipkart enables offline browsing to help consumer save on data usage. "It is even better to collect consumer data on cellphones because it comes with a unique device identity," Chatterjee adds.

"Guided discovery can also work as a delightful discovery because sometimes the consumer is not sure what she is looking for. In such case e-retailers should offer adjacent products such as a bouquet of flowers or car rental deals along with an evening dress," says Sanjay Sethi, CEO, Shopclues.

Online deal sites have entirely different challenges when it comes to guided discovery. The key here is local services. While selling deals, it is important to understand the deal structure that gets the best response. Says Ankur Warikoo, regional head, APAC, emerging countries, Groupon, "We plan to bring location-specific smart deals to India. It identifies who is browsing the website from which location. So the consumer gets relevant deals in real-time."

For a gifting site like Giftease.com, guided discovery becomes complex because the is usually buying the gift for someone else. Hence, recommendation engines that use cookies or IPs for making personalised recommendations are of limited use. "We use past transaction data like the occasion, relationship and item attributes like price, category to refine our recommendations," says Vivek Mathur, CEO, Giftease Technologies.

Social shopping platform LimeRoad.com uses scrapbooks and a flip-book magazine to helps users browse through various products on the site. Users have the option of seeing a live feed of products that other users have viewed. "Our scrapbooking community has curated more than 60,000 scrapbook looks, which are 100 per cent unique to our site," says Suchi Mukherjee, CEO & co-founder, adding that recommendations can help increase the average ticket size of the purchase between 75 per cent and 100 per cent. "conversion is always high if the recommendation tool is working properly, though unlike in gadgets and books, recommendations in the lifestyle segment are mostly led by the visual appeal of a product," she adds.

How to make it better
Most recommendation engines are based on collaborative filtering, which looks at the affinity between two products in the same category or between two categories. It has limitations because it only looks at past purchase history of a consumer. So if a consumer bought a striped shirt or searched for it, she will keep getting recommendations even if she has closed that deal. So just using internal data doesn't work. "To make the right kind of guided discovery, one must combine internal data generated inside the company with external data," says Srikant Sastri, co-founder, Crayon Data. "External data can be generated by trailing the internet reviews and social networking for movies, books, shopping, travel etc. Building cross-category connection on the basis of external and internal data can make recommendations richer and powerful."

Traditional recommendation engine often works like spam. Guided search should work on pull mechanism rather than push mechanism. So if there is a shopping application, you should ideally get the personalised choices only when you open the app.

The other issue is dealing with cart abandonment. One way to tackle this is to offer real-time coupon codes - such as a 20 per cent off if you shop in the next 15 minutes. "E-retailers need real-time, onsite and personalised coupons. With the help of analytics, it is easy to know how many people visited the site, from which location and when they bounced off," points out Avlesh Singh, co-founder & CEO at WebEngage, which provides on-site engagement tools to major e-retailers such as Flipkart, and Jabong.

Webengage enables websites to target users leaving sites through leave intent-based targeting that detects mouse movements of visitors. A pre-configured survey/notification pops up the moment a user is approaching the browser's close button or moving to a different tab in the browser window.

Komli Media, a digital advertising technology platform, uses a client's product catalogue feed to create personalised ads to retarget users. For instance, a user expresses her interest in buying a product through com but does not make a transaction. Komli's remarketing demand side platform monitors this and when the visits any social site next time, the ad of this particular product pops up alongside her profile. This increases the company's chance to convert the visitor into a Says Ashwin Puri, VP, remarketing & mobile, Komli Media, "We match user's browse history and use that data to surface relevant ads."

These are just some of the ways smart e-retailers are using technology and the data at their disposal to hang on to customers and urge them to spend more. The thing to remember here is to stop before you become pushy. Experts contend that while analytics can help in targeting the consumers and customising options for her, e-retailers must understand that customers want to be guided, but they don't want to sacrifice the experience of shopping. Also remember, narrow personalisation is not just something you sprinkle on your website landing page. It goes beyond inserting a name at the top of a web page or e-mail. It should be a business -one that takes into account not only who the is but what she likes, what she doesn't like, when she wants to hear from you, and when she doesn't. To be most effective, guided discovery is ideally managed from a single integrated platform that makes it easy for you to deliver a relevant experience across all interactions, through the entire lifecycle.

image
Business Standard
177 22