Genesys Cloud - Main

 View Only

Discussion Thread View
Expand all | Collapse all

Overall Customer Sentiment score calculation

  • 1.  Overall Customer Sentiment score calculation

    Posted 07-15-2021 12:16
    Hello Community,

    Hope all are doing good! 

    Can anyone please throw some light on how exactly is the overall Customer sentiment score is calculated?
    I am unable to get how exactly each of the positive and negative sentiment scores are weighted to calculate the overall score.

    Appreciate your help on this! :) 

    Thanks,
    Nivedha
    #QualityManagement

    ------------------------------
    Nivedha Kalathi
    Conn3ct (Connect Managed Services (UK) Limited)
    ------------------------------


  • 2.  RE: Overall Customer Sentiment score calculation

    Posted 07-15-2021 23:22
    Behind the scenes, that is actually done by Google CCAI, but Genesys will be getting some controls exposed to users to allow for manipulation of the variables used.  In the meantime, if you want to know how Google does it, click here:  Natural Language API Basics  |  Google Cloud.  Basically, it looks at the overall emotion of the text and determines if it is negative, positive, or neutral and the trend upward or downward from the 0 start point.  There is a range of -100 to +100 and the score is the overall average of the conversation.  If you can get through the link above and fully understand it on the first pass, get me you resume and I'll find a job for you!!

    ------------------------------
    Robert Wakefield-Carl
    Avtex Solutions, LLC
    Contact Center Innovation Architect
    robertwc@avtex.com
    https://www.Avtex.com
    https://RobertWC.Blogspot.com
    ------------------------------



  • 3.  RE: Overall Customer Sentiment score calculation

    Posted 07-16-2021 09:03
    ​Hi,

    Is there any API available in Genesys cloud to capture  the overall Customer sentiment score .It should include positive and negative sentiment analysis score.

    Appreciate your help on this! :)

    ------------------------------
    Anitha ME
    Infosys Limited
    ------------------------------



  • 4.  RE: Overall Customer Sentiment score calculation

    GENESYS
    Posted 07-16-2021 09:32
    Hi Anitha, 

    The transcription URL itself contains the information you may be looking for, overall summary information is also part of the Recording data set and can accessed directly from the Speech and Text API endpoint. 

    Within the URL:

    "sentiment": [{
    					"participant": "external",
    					"phrase": "you are treating me well",
    					"startTimeMs": 1619441069316,
    					"duration": {
    						"milliseconds": 1284
    					},
    					"sentiment": 0.7257699966430664,
    					"phraseIndex": 4
    				},​


    Per Interaction:

    {
      "conversation": {
        "id": "",
        "selfUri": ""
      },
      "sentimentScore": 0,
      "sentimentTrend": 0
    }


    Depending on which level of data you require may change which endpoint you need to use. 



    ------------------------------
    Cameron Smith
    VP, Product Management - Workforce Engagement Management

    cameron.smith@genesys.com
    ------------------------------



  • 5.  RE: Overall Customer Sentiment score calculation

    Posted 07-16-2021 10:32
    ​Thanks for the information.


    ------------------------------
    Anitha ME
    Infosys Limited
    ------------------------------



  • 6.  RE: Overall Customer Sentiment score calculation

    GENESYS
    Posted 07-16-2021 09:17
    Hi Robert, thanks for responding. 

    Just to add a small correction here on which service (Google or Native Genesys Cloud) is used depends on the organizations configuration. 

    For customers NOT using Agent Assist
    Transcription is provided by our native transcription services which is developed by Genesys. As well as topic detection, sentiment analysis and so forth. 

    For customers who ARE using Agent Assist
    In this setting there is no need to transcribe the interaction twice, so Genesys consumes the transcript created by Google during the Agent Assist workflow and then pushes said transcript back through our existing transcription workflow. This retains a consistent transcript across both use cases as there are subtle differences in the output of each transcription engine. Topics are then processed using the native topic classification

    Hope that clears up any confusion. 


    ------------------------------
    Cameron Smith
    VP, Product Management - Workforce Engagement Management

    cameron.smith@genesys.com
    ------------------------------



  • 7.  RE: Overall Customer Sentiment score calculation

    GENESYS
    Posted 07-19-2021 08:51
    Actually, the Sentiment Analysis engine in Genesys Cloud is not Google CCAI.  It is Genesys Native and is trained from open corpuses as well as out own Contact Center specific tagged data.  As a results, our Sentiment Analysis is tuned towards Contact Center conversations, unlike other Generic providers.

    ------------------------------
    Rakesh Tailor
    Senior Director, Product Management - Workforce Engagement
    Genesys Cloud
    ------------------------------



  • 8.  RE: Overall Customer Sentiment score calculation

    Posted 01-31-2022 20:13
    Hi Rakesh,

    Thank you for clarifying this.  Has there been any work done on benchmarking the Sentiment Analysis score, -100 to +100, that Genesys Cloud produces?

    Best regards,

    Paul Shaw

    ------------------------------
    Paul Shaw
    CALLSCAN AUSTRALIA PTY. LTD.
    ------------------------------



  • 9.  RE: Overall Customer Sentiment score calculation

    GENESYS
    Posted 02-01-2022 16:21
    Hi Paul,

    We do see that the score generated is not well distributed and we do have an update coming out in March (March 23rd), that will use all sentiment events across the interaction that better represents the customer sentiment.

    It would be good to benchmark after that update.

    Thanks,

    ------------------------------
    Rakesh Tailor
    Senior Director, Product Management - Workforce Engagement
    Genesys Cloud
    ------------------------------



  • 10.  RE: Overall Customer Sentiment score calculation

    GENESYS
    Posted 07-16-2021 10:12
    Hello Nivedha,

    The customer sentiment score represents the customer's overall sentiment from -100 to +100.

    The customer sentiment score weighs all of the positive and negative markers that took place at the end of the interaction (last 15% of the call) to provide an indication about the customer' experience with the contact center as the interaction with the agent is ending. The goal of the score is to provide a visual indication on how satisfied or dissatisfied the customer was at the end of the interaction. This metric can help in assessing customer experience, potential repeat calls, as well as indication of possible customer cancellation (churn). 

    Regards

    Jose Ruiz
    WEM Product Manager
    jose.ruiz@genesys.com



    ------------------------------
    Jose Ruiz
    Genesys - Employees
    ------------------------------



  • 11.  RE: Overall Customer Sentiment score calculation

    Posted 07-17-2021 12:14
    Hi Cameron, Jose,
    are you planning to release a sentiment analysis analytics aggregate API that allows to query the aggregate sentiment score for conversations in a queue to return the max, min and average sentiment score and filter that ideally with some predicates. An interesting result would be to check the sentiment score avg, max, min per wrap up code or for a specific user for example.
    The current APIs are limited if we can't have analytics queries.
    Thanks for clarifying

    ------------------------------
    Hichem Agrebi
    hichem.agrebi@cc-expertise.com
    CC-Expertise Ltd
    ------------------------------



  • 12.  RE: Overall Customer Sentiment score calculation

    GENESYS
    Posted 07-19-2021 08:57
    Actually, you can now Search by Sentiment, so you can find the interactions with a min or max Sentiment Score today.  This is available as a Beta.

    We have an Aggregate API for number of Positive or Negative Sentiment Events.  Not Sentiment Score.

    An Aggregate API for Sentiment Score is in the works, we are planning on having that API ready for mid October (currently set to Oct 6th).


    ------------------------------
    Rakesh Tailor
    Senior Director, Product Management - Workforce Engagement
    Genesys Cloud
    ------------------------------



  • 13.  RE: Overall Customer Sentiment score calculation

    Posted 07-19-2021 10:37
    Edited by Hichem Agrebi 07-19-2021 10:43
    Thanks Rakesh for confirming the roadmap plans,
    The current /api/v2/speechandtextanalytics/transcripts/search is more restrictive (pagination limited to first 1000)  than other search APIs.
    Having an aggregate API is definitely more performant than scanning through all conversation transcripts.

    Here's what I am using the search (with limitations) for currently and waiting on the aggregate API to hopefully better report on score as well as trends to compare user's scores and trends with queue scores.


    ------------------------------
    Hichem Agrebi
    hichem.agrebi@cc-expertise.com
    CC-Expertise Ltd
    ------------------------------



  • 14.  RE: Overall Customer Sentiment score calculation

    Posted 07-16-2021 11:56
    Hi Nivedha,
    Based on GC Topics, Noralogix GCEvents, Azure Logic Apps and Azure Cognitive Service we implemented real time email transcript sentiment analysis.
    Also can be done key phrases extraction. Here you can find demo.


    ------------------------------
    Taras Buha
    taras@noralogix.com
    www.noralogix.com
    ------------------------------



  • 15.  RE: Overall Customer Sentiment score calculation

    GENESYS
    Posted 07-19-2021 10:15
    Hey Taras,

    Thanks for sharing this demo! I liked seeing how sentiment was analyzed for the individual emails. What does the final customer score look like? I assume it's just an average of the sentiment scores that are calculated, but wanted to check. 

    Cheers!

    Matt

    ------------------------------
    Matt Lawson
    Genesys - Employees
    Online Community Manager
    ------------------------------



Need Help finding something?

Check out the Genesys Knowledge Network - your all-in-one access point for Genesys resources