A profile is a customizable set of parameters, conditions, and filters that allows you to automatically select calls for analysis and set individual processing settings. Using profiles, you can adapt the analysis to different scenarios, client types, or tasks, ensuring more accurate and effective use of data.
Now, it is possible to configure up to 10 profiles.
When creating multiple profiles for analysis might be necessary
Let’s consider a few common scenarios:
International company:
If your business operates in different countries, it is important to account for language specifics. For example, conversations in Ukrainian are better transcribed without translation. Meanwhile, German-language dialogues can be automatically translated into English so that a manager in Ukraine can easily understand the content of the call.Different levels of employee experience:
Your team may consist of both experienced employees and newcomers. New managers should be evaluated based on a larger number of criteria since they are still learning the processes and may make more mistakes. This helps identify areas that require training and adjustments.Different departments with different tasks:
For example, you might have a sales department and a customer support department. For salespeople, it is crucial to monitor how well they follow scripts and present the product. Meanwhile, for support specialists, the key criteria are the speed and quality of customer query resolution.
Creating separate profiles for these cases will allow you to fine-tune the analysis and make it as effective as possible.
Creating and Configuring a Profile
To create a profile, go to the General Settings section and select AI Analytics.
Click Add Profile and name the profile:
Then you need to customize the profile you created.
Click Edit:
You will see a window with various settings.
Each section will be explained in detail below.
Transcribing and analyzing language
The call transcription language is the textual transcript of the call. When transcribing in the original language, artificial intelligence (abbreviated as AI) detects the interlocutors' language and outputs the transcript in the original language. If you select English, the AI translates the call content into English.
Analysis Language — the language in which AI analysis will be displayed in the reports within the Call Log section.
Conditions for Analysis
In this section, you can configure parameters or rules for calls to be analyzed by a specific profile.
For example, in one profile, you might want to set rules so that only calls from the Sales department and those lasting more than 30 seconds are analyzed.
To do this, simply apply the appropriate filter:
There are also preset filters for Callback, Answered Calls and Proper calls:
There are a lot of parameters for selecting the analysis conditions. It is important to compose the conditions correctly, otherwise the call will not pass through the filters of the specified profile settings.
If a single call is analyzed by multiple profiles, the analysis result will reflect the profile that was processed first on the server.
At the same time, the analysis of one call by two profiles will not be displayed.
Analytics Settings (and a description of each metric)
This section includes various metrics by which AI will analyze the calls.
The base cost includes up to 5 metrics.
Each additional metric increases the cost of analysis. You can find more details about the pricing here.
You can select at least 3 metrics by checking the checkboxes.
These metrics have standard evaluations and criteria for AI analytics.
The criteria and scores for each metric are described below in the section Description of each metric and its standard criteria/conditions.
To set your own conditions for the existing metrics, select the checkboxes of the desired metrics and activate the Configure selected metrics.
After activating the checkbox, a drop-down list will appear in the section with a list of selected metrics and an opportunity to set a description for each metric for your business processes (you can use the default descriptions described in the blue blocks below as an example):
Attention! The limit of each field to be filled in is 255 characters.
Description of each metric and its standard criteria/conditions (click to open section)
Description of each metric and its standard criteria/conditions (click to open section)
Summary - short summary of a conversation.
Example of standard conditions/requirements for AI:
Give a summary of the dialogue.
Dialogue quality - consists of 4 parts (opening, anamnesis, presentation, closing), and also includes conversation evaluation and details.
Example of standard conditions/requirements for AI:
Opening - At this stage, the representative should greet the customer, tell them his name and company name, name the customer by name or ask for the customer's name.
If the representative initiates the call, ask whether the customer is comfortable speaking now.
Check whether all of this happened and whether the representative was polite.
History(Anamnesis) - At this point, the representative should get details about his business and the aim of the call from the customer.
After asking those questions, the representative has to summarise the information and ask whether everything is correct.
Presentation - At this stage, representatives should provide information about the company and the product.
Provide enough details and be focused on details about customers\' business to make this presentation more personal.
Closing - At this stage, the representative should summarise the dialogue and get confirmation from the customer that everything is clear.
Schedule the exact time and form of contact or provide the exact next steps.
Score - Average score.
Review - What were the mistakes for each step that made an impact on the provided score
Title - brief one-sentence summary of the dialogue.
Example of standard conditions/requirements for AI:
How would you summarize the dialogue in one sentence?
Parasite words - whether the manager used unnecessary/meaningless words.
Example of standard conditions/requirements for AI:
Did the manager use parasite words?
Yes = 1
No = 0
Obscene language - whether the manager used profanity.
Example of standard conditions/requirements for AI:
Did the manager use obscene words?
Yes = 1
No = 0
Keywords - list of keywords.
Example of standard conditions/requirements for AI:
Set of keywords that have meaning and can be key or descriptive for the conversation.
Sentiment - analysis of client mood, manager mood, and overall dialogue.
Example of standard conditions/requirements for AI:
Sentiment of the conversation in 1 word;
Sentiment of the client in 1 word;
Sentiment of the representative in 1 word;
The descriptive sentiment of the conversation overall and if it was changing during the conversation provides that in a descriptive manner.
The determination of mood depends on the selected "type of service" (e.g., Claude, Sonnet, GPT-4, etc.), as these models are trained and configured using different algorithms. Each AI system has a predefined set of possible moods it can recognize and display. The selection of a specific mood is based on the internal algorithms and configurations of the chosen type of service.
Key points - key moments of the conversation in list form.
Example of standard conditions/requirements for AI:
An array of keywords that you can find in the conversation that are meaningful and can be key or descriptive to the conversation
Mistakes - manager's mistakes when communicating with the client.
Example of standard conditions/requirements for AI:
What are the mistakes made by the representatives? Can you outline them? With detail.
Good points - emphasis on the positive in the dialogue.
Example of standard conditions/requirements for AI:
What representative made the best? What can be an example for others to outline
Next steps - next steps discussed during the conversation and further actions.
Example of standard conditions/requirements for AI:
What are the suggested next steps after a call also, focus on the next steps that were discussed in the call
Review required - rating from 1 to 10 whether the dialogue needs to be checked manually, and an explanation of the rating.
Example of standard conditions/requirements for AI:
Score - On a scale of 1 to 10, do you suggest reviewing this call manually?
Details - Why do you give such a score?
Follow-up - main conclusions of the call and recommendations for further actions.
Example of standard conditions/requirements for AI:
Main points - What are the main points of the call? Who and what said? What should be the key takeaways of the call?
Next steps - Who, what and when should do after the call?
Own parameters - your custom parameters for AI analytics.
Simply enter a task that is not covered by our standard AI analytics parameters offered on the analytics page.
The text input limit is 255 characters.
You can create only three parameters.
Example of your assignment:
1. You want the AI to check that the manager verifies that the customer's email is clarified by the manager.
To do this, you can form the following request: “Check that the manager doesn't forget to clarify the customer's mail”
2. Or you don't want the manager to give his/her Viber to the customer. You can give an example request:
“Make sure that the manager does not use the phrases “my Viber”, or “his/her Viber” and does not pass his Viber to the client”.
Stop-words - in this field, you can specify one or more words that are prohibited for managers of your company to use during the call.
The limit of one selection with words is 90 characters.
Phrases are not counted, only individual words.
Necessary phrases- in this field, you can specify one or more phrases that MUST BE used by managers of your company during the call. If necessary phrases were not used in the dialogue, AI will draw attention to this, highlighting them separately in the analytics.
The limit of one phrase (select) is 90 characters.
Add context for transcription - this is a field where the user can add short phrases or terms that will help AI better understand the subject of analytics, the scope of your company's work, etc.
With a correctly filled "transcription context", AI will be able to provide more accurate and correct transcription of the conversation.
Field limit - 244 characters.
Context example:
Ringostat is a SaaS platform. Telephony, SIP, Callback, tariff, etc.
Important: context for transcription should be given in the language in which you want to receive the transcription. That is, you cannot specify the transcription language, for example, English, and write context phrases in Ukrainian.
Add context for analytics - a field where you need to describe in detail the structure of the conversation and specify specific terms.
Field limit - 2000 characters.
Context example:
As a sales department manager, I expect 'a dialogue between a Ringostat sales department manager and a client who is interested in implementing a call analytics system.' Ringostat company provides analytics tools, as well as convenient business telephony.
Important: context for analytics should be given in the language in which you want to receive analytics. That is, you cannot specify the analytics language, for example, English, and write context phrases in Ukrainian.
Other settings
Service type
In this block, you can select the model that will analyze the calls.
Each AI model has its own cost:
Detailed cost information is described here.
Based on our experience and analysis of the effectiveness of various AI models for call processing, we recommend the Сlaude Sonnet model.
It demonstrates an optimal balance of analysis quality and cost, providing high accuracy and reliability of results. Additionally, this model features a more "human-like" approach — it better understands conversation context and captures client mood and emotions, which allows for more accurate call interpretation.
But the choice is yours 🙂
Duplication
You can quickly create new profiles by copying existing ones. This helps avoid reconfiguring the entire profile — you just need to change the necessary parameters in the copy:
Comparison of analysis conditions
If you have several profiles, you can compare their settings in one window and quickly determine if changes are needed in any of them:
