How Chatbots can be useful for marketing digital agencies?

Online chat on the website gives an opportunity to communicate with a company in writing, which is pleasant thing for many people. However, in order to gain more customers by chat, there should be a special employee, responsible for this activity, which, of course, costs money. We are now telling a story about the case, when one of our clients managed to decrease the workload on Chatbot operator by 80% and increase the attendance of the website by 12%.

About the client

The company, which turned to us is a marketing internet agency. It's clients are owners of developing business companies, in most cases people at the age of 27-40 years coming from big cities. They appreciate own time and like to fix arrangements, that's why they prefer writing to discussions: the correspondence can be done in parallel with other processes and details of negotiations never will be lost or forgotten.

This was a reason for receiving requests from the potential clients not only by calls, but in many cases by writing: e-mails, chatting and messengers.

People having online dialogue do not like waiting and the agency manager was monitoring chat and social networks all the time in order to catch all the requests and react on them quickly. However, it was not an easy task and some talks have lost. Requests did not converted into deals, and the agency started to lose clients.

It was getting obvious, that the whole process should be automated.

Our solution

We suggested to replace "living" online-consultant with a Chatbot. It collects all messages from Telegram, Viber etc. and as an online-consultant from JivoSite and communicates with the potential client, switching to the "living" operator occasionally.

By this way we reduced the workload on online-chat manager and we defended the agency from the human factor. Apart from the manager the Chatbot:

  • notices and reacts on all messages;
  • do not lose dialogues and never confuses them;
  • does not forget the important information about the client;
  • answers call within a second;
  • operates 24 hours a day and without rest days;
  • it is never angry, nor catches up provocation and does not step into conflict.

We prepared scenarios for the clients dialogue with chatbot.

As a result Chatbot allowed us to solve different tasks: it was telling about the company and its services, answered typical questions connected with a specific business area and the skills of the experts of this area, it could develop commercial offer and make an appointment, advised promotions and discounts.

Dialogue 1:

  • Hallo! Can I help you?
  • Order call/Order commercial offer/Tell about cases/Fix an appointment
  • Tell about cases
  • Medicine/Finance/Car industry/HoReCa
  • Finance
  • We elaborated marketing strategy for "Rocketbank" and "Renessaince bank". Details you can find on our website.

Robot reacted on specific words ( in our case "cases" and "finance"), and answered them according to the its specific program. If he could not find answer or the clients asked for an operator it switched on the operator immediately.

Dialogue 2:

Switch on the operator

  • One minute, your request is being processed
  • Operator Andrew on chat
  • Can I help you?

Chatbot is integrated with the client's Bitrix24 CRM. All the requests were transferred through it into the database and were recorded together with the history of dialogues. If needed, the operator can use these records in communication with the client in future. At the same time Chatbot used the information from CRM in its dialogue, and took into account the previous correspondence with the client.


Moreover, Yandex remarked, that the specialists in this company started to react more quickly on messages in online-chat and it was an additional benefit for the web-site. Its reputation increased, the audience increased by 12%, and as a consequence the number of incoming requests increased.

Back to Chatbot builder page


info sign

Can't find your country?
Visit for EU (€); or for any other country ($)