![Prosenjit Sen- portrait - Nov-2022](https://www.digitalcommerce360.com/wp-content/uploads/2022/11/Prosenjit-Sen-portrait-Nov-2022-266x300.jpeg)
Prosenjit Sen
In my first article in this series, I discussed how, despite the huge promise of B2B commerce, this space faces significant challenges for both pre-sales and post-sales due to the limitations of traditional technologies like “search.” In the second article, I discussed how you can skyrocket product discovery on your B2B site using the latest AI/ML technologies. In the third article, I discussed how you can provide top-tier post-sales support to keep customers happy at a low cost.
In this article I will focus on how, as a B2B commerce company, you can use AI to drive sales with a quick and frictionless path to purchase.
The purchase journey
Imagine you are a customer looking to buy a thermostat. You know of the manufacturer Warmup and have some idea of the specs (open temperature, close temperature, etc.). You need the B2B commerce site to help you make the right purchase with all necessary attributes of the thermostat.
Search is not ideal for such an experience. Consider a smart chatbot (Chatbot 2.0), which is capable of having a sophisticated conversation leveraging all your relevant reference data — in databases, HTML/PDF/Word docs, and applications like SAP and then performing mathematical computations in real time.
Here is a sample interaction:
YOU (asking the Chatbot): I am looking for a Warmup thermostat with open temperature > 60 C and price < $200.
CHATBOT: (Returns a list of 30 such thermostats, with Price and a Link to Purchase for each item.)
YOU now need to zoom in based on some of the other attributes of the thermostat but are not sure what the attributes are. So, you ask:
YOU: Give me the important attributes of Warmup thermostat.
CHATBOT: Returns the main attributes Open Temperature, Close Temperature, and Current Rating.
YOU: Help me find the thermostat with Close Temperature < 100 C and current rating < 5A.
CHATBOT: Returns 3 thermostats that meet this criteria (including previously specified Open Temperature > 60 C and price < $200). The results include the Price and a Link to Purchase each item.
YOU now want to shop around. So, you ask:
YOU: Do you have other comparable manufacturers that meet the same specs and are in the same price range?
CHATBOT: Returns 2 similar thermostats by Blodget, with Price and a Link to Purchase for each item.
YOU: give me the datasheets, and please send me the compliance and warranty information.
CHATBOT: Returns the necessary information. It also gives you a link the “Buy Now,” “Speak with an Agent” and “Buy Local.”
In this above conversation, the Chatbot is doing a lot of sophisticated AI-based processing (discussed below). Note, at every stage as it helps you find the right products, it gives you the option to close the sale.
Similar conversations are possible with a voice bot, though it would be harder, given some of the limitations of voice related technologies today.
Detect buyer signals and remove any friction
During this conversation with the Chatbot, you can use AI to continuously detect buyer signals based on the nature of the conversation and even understand the customer’s sentiment. Based on these — if at any time you sense the customer is not happy, confused or ready to buy — offer to transfer to a live agent, especially if it is a pricey product. Again, AI can help.
And, have all the necessary information that the customer may need so they don’t leave the discussion for later. Provide information on features, information on configuration/ installation, return policy, warranty, compliance, promotions, and more.
If they leave the chatbot to look up this additional information, prompt them to bring them back. If they leave, send them an email inviting them to click on the provided link and re-engage.
Post-purchase
Once a customer has made a purchase, it is important to have their loyalty. The best way to do this is again with prompt service; provide the customer any information that can help them post purchase. Provide an easy path to getting the status of an order, cancel, change order, recommend additional items that are complementary, and keep them posted with new product announcements.
Keep them informed, keep them happy.
After market
For most companies, after market is equally attractive. This is not only a big business, if the customer can buy the necessary parts and add on items easily it keeps them happy and loyal.
Technology
As discussed in previous articles, the key to providing all this functionality is to first bring together all possible reference data – structured data in relational databases, unstructured data in HTML, PDF, Word, and PPT, and data in applications like SAP.
You need a platform that can process data as follows:
- Connect and ingest data from different data sources: product specs, features, warranty, compliance, upsell information, price, inventory and more.
- Perform data cleanup: the data in different data sources often needs to be cleaned up or transformed for it to be meaningful to the AI engine. This data science work is often most time consuming and sometimes difficult.
- Provide data to the AI Engine: in real time, work with the AI engine to get different elements of data from all these different data sources, combine the various data elements, and even transform them so the relevant AI models can use them to prepare and even perform computations to arrive at the right answers.
AI technology, as discussed in previous articles, has evolved in many ways in the last 3-4 years. With large language transformers (including GPT) you can now do all sorts of sophisticated processing. In this solution outlined above, AI is needed at every step: to handle the dialog or conversation with the customer, identify the necessary data and then get results from structured data, get answers from unstructured data, detect buyer signals for purchase, detect customer sentiment, and more.
Language transformers allow you to detect patterns in large amounts of data — which can be used to now get the correct answers from your reference documents (features, compliance, warranty, etc.). Generative Pre-trained Transformer (GPT-3) can take this to the next level by providing an answer from the learnings of the reference documentation.
Conclusion
We are now at a very exciting time in the evolution of AI technology. With the availability of cheap computing power, you can bring automation to all customer support processes — pre-sales and post-sales. AI based platforms can bring you drastic efficiency and customer loyalty, at a lower cost than what you may be spending today. But it is important to be able to use the right AI technologies to accomplish the job, and most likely you will need a platform as opposed to just a tool. Watch out for platforms that require elaborate custom training of the AI models — this can become very expensive. The whole purpose of the recent AI advancements is to prevent such inefficiencies.
About the author:
Prosenjit Sen is a serial entrepreneur and currently the CEO of Quark.ai, an “autonomous support” platform that uses deep learning, natural language processing (NLP), and computer vision (CV) to bring automation to sales support and field support. He was previously employee No. 5 as part of 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 Startups Accelerator. And he is the co-author of the book “RFID for Energy & Utility Industries.” Contact him at [email protected].
Sign up
Sign up for a complimentary subscription to Digital Commerce 360 B2B News, published 4x/week. It covers technology and business trends in the growing B2B ecommerce industry. Contact editor Paul Demery at [email protected] and follow him on Twitter @pdemery.