Home / Case Studies / Responsive Keywords Chatbots

Responsive Keywords Chatbots

Industry: Recruitment

The Client

MaxAroma is a New York City based online beauty and fragrance supplier, founded in 2010. Dedicated to supplying only the highest quality of cosmetics and fragrances. MaxAroma has gained a superior international reputation for providing exceptional products and customer service.

Challenge

Manually managing products, real-time retrieval of all product information, managing the existing database is challenging.

Solution

Implemented a synchronization between the recruitment agencies' products names and our client's Salesforce CRM.

Results

By using a single platform, the recruitment team closed job profiles faster, reducing update time.

The Objective

The client wanted Technology Mindz to automate the process of giving services to their customers through salesforce so that all of the customers get responses to their questions & updates well in time.

The Challenge

Clients were facing multiple challenges, including:-

  • Customers should receive an auto response email of their order status in below scenarios
  • If they ask for status of their order via email or
  • If they inquire from the contact us page available on the Maxaroma website.
  • The response will be sent according to the current status of their order.
    (Example- Shipped,Pending,Canceled,Refund, Pending-Phone order, Cancellation Approved)
  • Customers should receive an auto response email if they ask for below questions via email or from the contact us page on Max Aroma website.
    (Cancellation request, How to return my item, Shipping Fees, I have not received my order,Wrong and Damaged product, Offers and Promotion by email or contact us form of Maxaroma website.)
  • The chatbot should be keyword responsive. If a customer types any keyword in the Chat window, the bot should respond back by picking those keywords.

The Solution

Technology Mindz was engaged by the client to review and analyze their current Business logic implementations.

  • Instead of doing manual coding, we did most of the part by automation provided by salesforce.
  • Create custom fields
  • Create new email templates and email alerts.
  • Create flows for the different business requirements with different conditions.
  • Create an intent set to collect the keywords for auto response on chatbot.
  • Add utterances in dialogs manually to auto response on chatbot.
  • Create a group in Einstein bot.
  • Create dialogs in Einstein bot.
  • Assign an intent to the particular dialog so that this dialog will display when the customer types anything which is in the intent set.

TrainBot feature

  • Enable the Einstein of the dialog which has at least 20 utterances so that Intent Matching Quality of the bot gets increased.
  • Create an intent model.
  • Intent model will show the percentage of how much our bot gets trained.
  • When customers type anything from the chatbot which is not the utterances in the intent set or any dialogue then these keywords/phrases will get stored in the Bot training tab of Model management builder of Einstein bot After 24 hours of typing.
  • Bot training section also shows suggestions of dialogue and the number of occurrences of that keyword. Based on those occurrences we can reclassify this keyword or can ignore it.
  • Clients can reclassify these keywords/phrases to the related dialogue by clicking on drop down button of that keyword and select the reclassify.After clicking on reclassify one popup will appear from that we can select the intent source and intent set and click on classify button or can click on ignore button to ignore it.
  • After reclassifying, this keyword will get added to the utterance of that selected dialogue directly and when the customer types the same keyword next time selected dialogue will get displayed.
  • In this way the number of utterances will increase. When the number of utterances will increase then again go to the bot builder, select the modele management and build model and now our bot will become more trained.This will show the percentage of our trained model.
  • Here is the loom video for refrence.

Technical Approach

We applied the following approach to satisfy client’s requirements:-

  • Setup Trello as a user story and project management tool, with Kanban board.
  • Organize once-a-week meetings to gather new requirements and groom the backlog.
  • These meetings were also used for user story prioritization.
  • These meetings were also used for Demo to the scope item which are completed in a week.

The Benefits

Clients got the following benefits after implementing the solutions.

  • Status of order will be sent to the customers automatically by selecting the ‘Status of shipment’ element from the ‘Automated Response type’ picklist field.
  • Cancellation request, How to return my item, Shipping Fees, I have not received my order, Wrong and Damaged product, Offers and Promotion will be sent to the customer automatically by selecting the Cancellation request, How to return my item,Shipping Fees, I have not received my order, Wrong and Damaged product, Offers and Promotion element from the ‘Automated Response type’ picklist field.
  • When customers type anything in the typing box of the chatbot they will get the response according to the keywords.
  • If they don’t find what they are looking for they can start chatting with a virtual agent on mouse click or transfer to an agent and this keyword will be added to the bot training model for future use and client can reclassify these keywords to any relevant dialog or can ignore it. Occurence of keywords also count in the bot training model.

The Results

The client is now able to auto respond with the auto email functionality for most of the cases, customers will get the reply to their query just on mouse click and customer will get the auto response to their queries sent by contact-us form page and customer can find their solution by typing keywords on typing box.