Додо пицца доставщик
Додо пиццы
Додо пицца лого

Telephony bot

Pilot for automation of the contact center «DODO Pizza»

«Dodo Pizza» is a pizzeria chain, one of the largest catering chains in Russia.
Automate the processing of some calls to the «Dodo Pizza» contact center.
Додо пицца лого
Client
Task
Solution
A bot that understands what customers are saying and helps them whenever possible.
Service presentation for «Dodo» at Conversations conference in november 2019.
ABOUT
ABOUT
ABOUT
ABOUT
ABOUT
ABOUT

What our bot can do

01/
Answer questions from customers who have already made an order.
Make changes to the order, if possible.
Switch to operator if it cannot help.
The bot warns the operator and transmits information to him about the call
If a client asks to switch to a pizzeria manager, the bot first repeats the client's first phrase to the manager.
Basic skills
Additional skills

Call topics

02/
15%
20%
35%
30%
questions about the delivery area
questions about current orders
other questions
questions about pizzerias

The appeals that the bot has learned to handle

03/
Courier late complaint
Want to change your payment method
Change of delivery method
Order cancellation
Product quality complaint

How does the voice bot work

The client calls «Dodo Pizza» and gets on the robot
The bot transcribes the client's response and examines the received text
Dodo IS transfers customer data to the bot (if he has already made an order)
Based on this data, the bot determines the topic of the appeal.
The system answers the person with a voice or communicates with the operator
The client is satisfied
04/

Interesting fact

«Connect with the emperor»
A phrase that «broke» the Dodo bot system due to an error in speech recognition.

General scheme of automation of CC

05/
Telephony
Сhatbot engine
Telephony
Voice menu
Speech2Text
Dodo IS API
Intent classifier
Bot script
Text2Speech
Operator
Client
Client
We launched a bot for «Dodo» as a pilot, a little «raw» . We improve the service in the process by analyzing the dialogs between the bot and real customers.

Bot development process

06/
Reviewed contact center agent guides
We listened to the recordings of over 300 «operator-client» conversations
Structuring data, collecting statistics
Compiled a classifier of intents - call topics
We worked out the scripts of the «client-bot» dialogues
Integrated with Dodo IS * and customer telephony
Step 01
Step 02
Step 03
Step 05
Step 06
Step 07
Highlighted topics of conversation that needed to be automated
Step 04
preparatory stage
main stage
Transferred to development
*Dodo IS - "Dodo Pizza" information system which supports Dodo website and application. It is used by franchisers and operators of the@nbsp;contact center.

Economics

07/
2
7
15%
250 000
minutes
rubles per minute
calls per month
average duration of a conversation between a client and an operator
average cost of a contact center operator
of the work of a contact center can really be automated
getting calls to the Dodo Pizza contact center
> 500 000
rubles per month
Total
saving «Dodo Pizza» within the framework of the pilot project.

Tools

08/
We used the following services in our work:
A simple and convenient platform for developing chatbots, voice assistants and contact center automation systems.
The AI service for chatbots analytics provides a detailed assessment of their effectiveness.

zDialog

OneDash

OneDash сервис
zDialog фреймворк

Difficulties of the project

09/
Problem
Problem
Problem
Decision
Decision
Decision
We knew nothing about the work of the contact center.
There was no data to train the chatbot. We did not transcribe the recorded conversations not to waste time and money.
While we were working on the project, some of the topics of appeals became irrelevant.
The bot made speech recognition errors in 46% of cases.
Problem
Decision
We have updated the bot knowledge base.
We replaced the voice recognition platform and settled on Yandex SpeechKit.
We used typical phrases that customers say. They were written by the operators of the contact center.
To reduce the risk of error, we asked «Dodo» as much information as possible.

Quotes

10/
Alexey Medvedovsky
leader of the IT direction of the contact center "Dodo Pizza"
«We had a lot of competence in the development of chatbots, but we had no idea how a contact center works».
«Before we met EORA, we had no idea what chatbots and artificial intelligence were».
Roman Doronin
CEO «EORA»

Mass media wrote about us

11/
RUSBASE лого
ICT.moscow лого

Results

12/
As part of a pilot project, the bot learned
successfully conduct and end a call
recognize the subject of the appeal in
67%
97%
of cases
of cases
among them
The pilot's launch was successful, and the bot's accuracy increased as more data was collected. The service started working in full-fledged mode.
Команда Eora
We are satisfied! And the pizza is really tasty :)
See the case
«How «Dodo Pizza» benefited from contact center automation»
Contact us