Suvvy User Guide
На русском
  • INTRODUCTION
  • QUICK START
  • MAIN SETTINGS
    • System Instruction
      • Best Practices
        • Guidelines for writing instructions
        • Ways to improve accuracy and response quality
        • Instruction debugging
        • Cost optimization
      • Variables
      • Functions
      • Multilingualism
      • Reading URLs
      • Typical reasons for stopping a dialoguePage
    • Additional settings
    • Notifications from Suvvy
  • Knowledge Base
    • Knowledge Base
      • Direct questions
      • Large files
  • Working with Tables
    • Tables (CSV-XLS / Google)
      • Data retrieval
      • Data recording
  • ACTIONS
    • Webhooks
    • Multiagent Systems
      • Calling a subordinate bot
      • Switch active bot
  • CHANNELS
    • CRM Systems
      • amoCRM
      • Kommo
      • Bitrix24
    • Website Chats
      • Jivo
    • Messengers
      • WhatsApp
      • Telegram
    • Social Networks
      • Instagram
      • VK
    • Marketplaces
      • Wildberries
      • OZON
      • Yandex.Market
      • Avito
    • Helpdesk Systems
      • UseDesk
      • PlanFix
    • Custom Channel
      • Working with webhook
      • Sending messages
  • INTEGRATIONS
    • YCLIENTS
    • ALTEGIO
    • Google Calendar
  • Analytics
    • Chart Analysis
  • FOR PARTNERS
    • Partner Program
    • Training
  • Legal Information
    • Terms of Service
    • Privacy Policy
Powered by GitBook
On this page
  • What does it allow?
  • How to connect?
  • Additional features
  • Testing order
  • Checking responses
  • Testing on real data
  1. CHANNELS
  2. Marketplaces

Wildberries

This section provides information on the procedure for connecting the Wildberries channel.

PreviousMarketplacesNextOZON

Last updated 5 months ago

What does it allow?

This channel is designed for automating responses to questions or reviews on the Wildberries marketplace platform.

To do this, you just need to upload training information in the format of question - answer related to your products into the bot and set a key for secure messaging between the bot and the platform. The system will do everything else for you.

You will no longer need to deal with routine tasks and spend time on typical questions. The system will automatically receive customer questions from the platform linked to products, analyze the loaded training knowledge base, and generate responses based on it.

Since the bot operates on the basis of artificial intelligence, the answers will be generated dynamically, meaning that even if there is no 100% match of the question with what is in the knowledge base, the bot will be able to analyze responses to similar questions for this product and form a relevant answer based on them.

How to connect?

1

Create a bot

2

Go to the "Channels" tab and select the "Wildberries" channel

Connected channels will be highlighted in green. If you haven’t connected the channel yet, it may not be highlighted.

3

Get the Token in your Wildberries personal account

The token needs to be created with the categories "Content" and "Questions and reviews".

4

Insert the received token into the "API Key" field

5

Press "CONNECT"

6

Upload the knowledge base to Suvvy in the "Question-Answer" format

The system will automatically suggest doing this.

You need to prepare a file in the xlsx (Excel) format or Google Sheeting following a special template that can be downloaded by clicking on the "Download template" button.

The table should look like this.

In the "Article" column, you can specify multiple articles separated by a space or comma - this will bind a typical question/answer at once to several products.\If the question applies to all products, then the "Article" field should remain empty.

These simple actions are enough for the bot to start working in basic mode and responding to client questions. There are a number of additional (optional) settings for finer tuning of the bot’s work, which we will describe in the next section.

Important!

In the event that Suvvy cannot find an answer to a question in the product card or answer table, it will not send a response to the question to the marketplace but will send a notification to the notifier bot chat in Telegram (details via the link below) that a question was asked and the bot could not answer (if additional settings are not completed, the section is described below).

Additional features

Besides default operation, the bot has the ability to configure additional parameters if the basic scenario of the bot does not satisfy. Settings allow:

  • Work with product descriptions. The following work algorithm is assumed:

    • As a first step, the system checks the incoming question (review) through the "question - answer" system, if an answer cannot be found via the table, the system can retrieve the product description and respond to the question based on the product description;

    • To obtain a product description, the following switch must be enabled

    • After obtaining a description of the product from WB, the bot checks for the availability of additional description in the "Article - description" table. If it is connected. The connection is made by pressing the "Configure bot responses by products" button. The table must have 2 columns: 1 column - Article (or list of articles); 2 column - Description.

  • Work with reviews. To enable the review processing mode, you need to enable the corresponding switch:

In this case, the same table of examples with responses to reviews is used as for questions. The question column simply records the feedback, and the answer column - accordingly, the response. There's no division in the table between questions and reviews.

  • Limit the list of processed items. To do this, you need to enable selection by list of articles

  • Disable the bot’s work either with questions or with reviews. As well as changing the service instructions, which are additionally transmitted to the model, to adjust its work on the Wildberries channel. Separately for reviews and for questions

Testing order

Checking responses

You can check the operation of the bot through the "Chat" tab, where you can send test questions and receive model responses. Since questions on WB are asked by clients with a reference to a specific product, during testing in the "Chat" section, you need to specify product information. The following parameters are used:

Remember to clear the dialogue before each new test, otherwise the entire conversation history will be transferred to the model, which is not needed for working with WB.

Testing on real data

For testing the bot on real data, the following work order is proposed:

  • The bot connects to the WB working channel with a working token;

  • The bot's responses to the channel are TURNED OFF;

To stop the bot from answering in the channel, you need to disable this switch:

Thus, all questions/reviews from WB will be received by the bot, the bot will process them, but will not send responses to the channel. In the "History" tab, you can check the bot's responses to real customer questions. If shortcomings are identified - make corrections and additional settings. If the bot answers correctly - enable the bot's responses in the channel.

Notifications from Suvvy
creating a new bot
channels menu
Template