Become Amazon in your space with AI/ML

0

Prosenjit Sen

Amazon.com, a company founded in 1994, has a market value of approximately $1.3 trillion, net sales of more than $465 billion in 2021 and, according to Digital Commerce 360, accounts for more than 40% of sales e-commerce in the United States. And it’s in a growing industry, with online sales up 50.5% since 2019.

It is essential that the self-service tool can extract answers directly from your original datasheets, user guides and other reference documents. AI/ML systems can deliver such capabilities with high accuracy.

Amazon has achieved this dominance by providing the customer with the ultimate shopping convenience: all you have to do is go to its site, type what you are looking for in the search bar, browse through the possible purchase options, then buy with one click. Once you order, have it delivered in a day or two, get full visibility to track the order’s journey, and resend it anytime. It is this ultimate convenience that has earned Amazon unparalleled loyalty around the world.

B2B e-commerce: explosive growth in the face of complex challenges

B2B e-commerce is now experiencing equally explosive growth, with companies from high-tech to industrial parts increasingly selling online.

blog-ProsenitSen-electronic sales chart

Source: US B2B Market Forecast Report 2022, Digital Commerce 360.

Start visualizing the future of B2B e-commerce and share critical data with colleagues and customers.

In 2021, online sales across B2B e-commerce sites, hookup portals and marketplaces increased 17.8% to $1.64 trillion, from $1.39 trillion in 2020. Report on US B2B market forecast 2022 by Digital Commerce 360.

U.S. manufacturers grew their combined digital sales 12.9% to $4.104 billion in 2021 from $3.634 billion in 2020, according to the 2022 edition of DC360’s The Manufacturing Report. B2B e-commerce now accounts for a larger share of all US manufacturing sales, so US manufacturers are turning digital commerce into a mainstream channel.

In this B2B commerce space, providing Amazon-level convenience is even more important. Business buyers are all busy professionals, often under tremendous pressure. They need to be able to buy online quickly, easily and on their own, no matter how complex the products.

According to Sapio Research, the top three challenges these enterprise buyers face are limited product data, inaccurate product information, and overly long and complicated checkout.

SapioResearch-blog-ProsenjitSen

Source: Sapio Research

B2B customers need to address these issues today

The level of customer frustration is high. They have little time and need to find products based on complex specifications on a site with millions of products. However, the main tool they have today is search, which is unsuitable for this space. As a result, they often need to call a helpline, use voice, live chat, or email site support, which is time-consuming and expensive for the B2B website.

Buyers should be able to specify their product specifications in English, without having to fill out an elaborate form or wade through hundreds of pages looking for the right items. They must obtain the exact product or part numbers, as well as the necessary spec sheet, drawing, installation instructions, etc. And if they miss to enter something in the spec, they should be sure that they will be prompted for the missing information.

Customer support requests also include installation, configuration and troubleshooting of specific configurations.

They also involve finding replacement parts, matching or conforming parts and mating parts. Whatever their need, they expect it to be easily accomplished through your merchant site using self-service tools.

Put your entire product catalog online

Rather than previewing just a few products, showcase your entire product catalog, even if it has millions of items. Make it easy for customers and distributors to find all the information necessary for customers and distributors, i.e. products or parts that meet their specifications, specific features, prices, promotions, spare parts, compatible parts, compliance and regulations, opinions, etc. The buyer must be able to compare several products in order to make a decision. Even compare your products with a competitor’s products.

The limits of existing technologies

1—The customer is looking for a product with a complex specification: Consider a customer looking to purchase a router with a specific specification: “I’m looking for a router with 2 WAN/LAP ports, at least 300 Mbps system throughput, WAAS-enabled, and priced under $4,000.”

Search technology cannot handle such a complex query. The search finds unstructured documents based on keyword density. In the case of this query written as an English sentence, you need to be able to get information from structured and unstructured data and then perform mathematical calculations on the information such as “less of 300 MBPS” or “price between $2,000 and $4,000”, then put together the pieces of information to find the answer.

You need a Deep Learning and Natural Language Processing (NLP) based system that can get the data subsets from structured and unstructured data and then perform intelligent mathematical calculations on it.

Also: “I need a router that supports IEEE 802.11a/b/g/n/ac/ax standards and offers 1148 + 4804 + 4804 Mbps speed. I also need a separate guest network.

2—Customer is looking for product features, installation steps, or setup instructions — and you must provide the exact answers: “How to configure QoS on Linksys routers?”

Existing search technology presents the customer with links to documents containing the requested information such as data sheets, user guides, product guides, etc. But going through these documents takes more time than the client is willing to spend. The customer today needs specific answers, for example, in the case of the query above, the exact configuration steps and perhaps an accompanying visual from the relevant user guide.

In your B2B commerce site, you may have a million products with a million data sheets and a million user guides. It’s important to be able to extract the exact answer (into paragraphs, images, tables, lists), display it with the exact formatting as in the source document, and then provide a link to the exact location of the response in the source document.

If you have this capability, you can offer it to the customer via self-service chat or voice.

It’s also essential that the self-service tool can pull answers directly from your original datasheets, user guides, and other reference materials. With such volumes of reference documents and products, there should be no need to mark up or personalize a document. Otherwise, this project will become too expensive and impractical.

AI/ML systems can deliver such capabilities with high accuracy.

3—Click to buy: provide a frictionless path to purchase

Finally, you need to close the sale. It is essential that the customer can buy the products very easily. You need to provide a purchase journey with minimal friction and personalize it so that customers don’t need to spend a lot of time completing the order. The last thing you want is to go through all the steps above and then make it difficult for the customer to buy the product.

You need to observe customer interactions on your site and be able to escalate them to a live agent if you feel they need help (or are unhappy) – or if you think they are likely to. to be a very attractive buyer. There are AI models that can help.

4—Post-Purchase Support Issues: After placing an order, the customer should be able to easily get the order status, modify an order or even return an order received at any time. Existing voice bots and chatbots have several limitations, which often require expensive and often undesirable live support for customers. Email support is generally slow.

You should also have an ongoing communication channel with customers to provide reminders about maintenance, add-on products, promotions, new product announcements, etc. Make sure you are always in the lead.

blog-ProsenjitSen-B2BBuyerReport-Picture2

Percentage of B2B buyers citing the B2C online features they most want on a B2B website. Source: Sapio Research.

Keeping the promise of B2B

The promise of Amazon-inspired B2B commerce is beginning to materialize, despite its complexity. The table below shows that the risks of inadequate support not only increase the cost of sale, but can also result in the permanent loss of the customer. Fortunately, there have been significant advancements in deep learning, NLP, and other AI technologies in recent years. It is now possible to overcome these challenges and take care of your customers, just like Amazon does.

blog-ProsenjitSen-CXchallenges-Picture5

Percentage of manufacturers citing customer experience challenges as top 3 for impacting customer relationships. Source: Sapio Research.

In my next article, I will discuss how advances in AI (Deep Learning, NLP, and CV) can be used to address these pre-sales and post-sales issues discussed above.

About the Author:

Prosenjit Sen is a serial entrepreneur and currently CEO of Quark.ai, an “autonomous support” platform that uses Deep Learning, NLP, and Computer Vision to automate sales and field support. He was previously employee #5 on the founding team of Informatica, a pioneer in online data integration technology. Prosenjit is a mentor for the Alchemist Accelerator and the Bay Area IIT startup accelerator. And he is the co-author of the book “RFID for Energy & Utility Industries”.

Favorite

Share.

About Author

Comments are closed.