Introduction
Strawberry House — manufacturer of chocolate-covered strawberries, gift bouquets and sets. The company operates in several cities of Uzbekistan: Tashkent, Samarkand, Fergana, Namangan and Andijan.
The main requests are received through Instagram, website and Telegram, and applications are maintained in amoCRM. The business is characterized by pronounced seasonality. On normal days, the flow of requests remains predictable, but before the holidays, demand grows sharply and in a few days creates a load comparable to several ordinary weeks.
The busiest period is at the beginning of March, when customers en masse choose gifts for March 8
Main task
The client needed more than just a bot to answer typical questions. The main goal of the project is to prepare the sales department for a sharp jump in traffic and prevent the loss of applications in the period from March 6 to March 10.
During the season, the number of incoming messages increased approximately threefold. During March the company received more than 70,000 messages.
With such a workload, managers were physically unable to simultaneously:
— quickly respond to new clients;
— view sent Reels, photos and links;
— determine whether a person has a real intention to buy;
— specify the recipient and the reason for the gift;
- communicate in different languages;
— filter out spam and advertising offers;
- collect phone numbers;
— place ready-made orders
Tripling the staff for a few peak days was expensive and logistically difficult. We needed a system that automatically scales with the number of requests
What have we implemented?
For Strawberry House, we set up a multilingual AI bot ChatRex with integration into amoCRM. The bot became the first line of processing the incoming stream. It accepts a new request, determines the language, analyzes the content of the message, separates potential buyers from spam and conducts initial qualification
ChatRex does not replace the manager when selecting a specific set, calculating the cost and placing an order. It takes away the most massive and load-sensitive stage - the first touch
How a bot copes with a sharp increase in traffic
Unlike an employee, a bot can simultaneously conduct a large number of conversations.
When the flow of messages triples, ChatRex does not need to expand the shift, redistribute the queue or postpone some requests until the next day
Each new message undergoes the same processing:
- The bot records the request.
- Defines language and writing.
- Analyzes text, link, post or Reels.
- Separates spam and untargeted messages.
- Determines whether there is purchasing interest.
- Clarifies to whom the gift is being chosen and for what reason.
- Finds out if a specific kit, photos or shipping is needed.
- Requests a personal phone number.
- Checks the correctness of the number.
- Submits the prepared application to the manager.
The bot identified that the request was not related to a purchase, but to an advertising collaboration through barter. It continued the conversation in the client's language and requested details of the proposal without conveying an unstructured message to the manager
The client sent Reels without a detailed description. The bot recognized the publication, clarified whether such a set was needed and for what reason, and then received a phone number to transfer the application to the manager
What did the client receive?
Strawberry House received not just a chatbot, but a tool for managing seasonal workload.
After implementation:
— sudden surges in traffic no longer require proportional expansion of the first line;
— clients received a quick initial response;
— almost half of the stream from Instagram was automatically parsed by a bot;
— requests in different languages were processed according to the same logic;
— spam did not overload the sales department;
— managers received already qualified applications;
— night calls did not go unanswered;
— the team could focus on placing real orders.
The main result of the project is ChatRex helped the business go through a period of multiple growth in demand without a multiple increase in staff and without having to manually parse each new message. For seasonal businesses, this means that high demand is no longer a problem for the sales team and becomes an opportunity to handle more potential orders.
