Latest News

TMLabs Advanced Feedback Analysis Application for ServiceNow – by Tobias Schwartz

TMLabs’ latest innovation combines ServiceNow’s enterprise-grade Survey Management capability with Google’s Cloud Natural Language API to obtain real-time insights from unstructured text using Google machine learning.

Now more than ever, with companies having to completely transform how they serve their customers, ServiceNow surveys are becoming a critical tool for understanding customer satisfaction. However, the following issues remind us why it’s a challenge to get maximum value from surveys:

  • 15% is the average customer survey response rate1
  • 50% or higher survey response rate should be considered excellent in most circumstances2
  • 30% – 40% response rate is required for an effective NPS campaign3
  • Metrics with incomplete survey data can lead to:
    • Incorrect inferences
    • Generalisations
    • Sampling bias
    • Lower than expected survey scores
  • Historical ServiceNow Cases and Tasks often contain valuable sentiment data that’s not easily harvested to assess service outcomes.

With the release of Feedback Intelligence Pro, TMLabs is extending the already powerful features in the standard version of the Feedback Intelligence application, providing organisations with cutting edge sentiment analysis capabilities using machine learning. Available now to download from the ServiceNow Store: https://bit.ly/3b249PJ.

Key features and benefits include:

  • Complements actual survey responses with sentiment scores to get a complete picture of how a service is performing.
  • Helps to make more targeted service improvement decisions and measure the outcome of those decisions straight away, ultimately increasing customer satisfaction and prevent customer attrition.
  • Uses Google Analytics Machine Learning to calculate sentiment scores in real time using user generated comment data on ServiceNow Cases, Incidents, or Requests.
  • Sensitive info (such as contact details and credit card numbers) is masked before sent to Google’s API.
  • All calculated scores are mapped to the ServiceNow configured survey scale for consistency.
  • Prevents surveys being cancelled due to lack of response via intelligent flows.
  • Complete survey data can be used to make important business decisions in real time.
  • Harvest anonymised sentiment data from historical survey records.

The solution has been architected in a modular fashion, and can be tailored to any industry and customer context by utilising ServiceNow Now Platform Flows and Notification engine. TMLabs are able to assist with implementation of Feedback Intelligence Pro with any customer globally.

The next section describes the various features in more detail to provide transparency as to the exact logic implemented in the app.

Process flow

The following steps illustrate the process and data flow for a given ticket (e.g. Incident, Case):

  1. Ticket is created
  2. Customer adds comments to the ticket
  3. Ticket is resolved
  4. Survey is sent to customer
  5. Customer comments are sent to Google’s Sentiment API (sensitive data is masked before sending)
  6. Google returns a sentiment score
  7. The sentiment score is mapped to the survey score

Now all surveys for tickets with customer comments will have a score based on the customer sentiment.

Data masking

Below steps illustrate how the controls for sensitive data masking has been implemented based on regex patterns:

  1. If system property for number (i.e. phone number, credit card number, social security number) filter is active then replace numbers with provided character in system property (default is 0).
  2. If system property for email filter is active then replace emails with provided default value for email in system property.
  3. If system hyperlink filter is active then replace hyperlinks with provided value in system property.
  4. If system property for name filter is working then the user table is scanned to check whether any name matches can be found and replaced with value provided in system property (default is 0)
    1. Firstly, the entire comment string is split into a list of unique words
    2. Then the user table is queried whether any of the words in the list match.
    3. For each match the respective word in the comment will be replaced with the sys property value.

Custom table count

No custom table is created by installing the app from the ServiceNow store.

Survey results controls

The app contains a number of properties to control how the Google API sentiment score will be translated into the survey score.

For records that don’t have any comments a system property can be checked to automatically set scoring for those records as neutral, i.e. Google score = 0.

For records that return a neutral Google API sentiment score there are properties to set a value to increase the sentiment score to indicate that even a neutral score is likely to be a positive and satisfactory service experience for the user.

Historical sentiment analysis

Below steps illustrate how the analysis of existing (closed) records is performed:

The historical analysis is triggered by a UI action on the survey definition form.

  • All survey metrics within the survey are assessed. Only numeric metrics are suitable.
  • Survey trigger conditions are assessed and notifications are turned off (so that users won’t be notified about a record survey from long time ago.)
  • Existing survey records that had no response are updated with the Google API sentiment score.
  • For records that meet trigger conditions but did not have a survey record created previously then a new survey record will be created and updated with the Google API sentiment score and survey trigger conditions are assessed.
  • The survey scores are calculated taking into account the system properties highlighted in previous sections of this article.
  • Notifications for the survey are turned on again to allow for future surveys to be sent.

Now all existing records that met the survey definition trigger conditions have a corresponding survey record and a calculated score based on Google’s sentiment API providing a comprehensive picture of the service performance over the respective time period.

12017, https://www.customermonitor.com/blog/increase-your-net-promoter-score-survey-response-rate-checklist

22019 https://www.customerthermometer.com/customer-surveys/average-survey-response-rate/

32018, https://www.retently.com/blog/survey-response-rate/#:~:text=NPS%20surveys%20score%20much%20over,insights)%20from%20every%20delivered%20survey

Leave a Reply

Your email address will not be published. Required fields are marked *