How to connect Skeepers to semantic analysis tools

Using classification plans created with Skeepers, Erdil, Proxem, Viavoo.

Through semantic analysis via classification plan you can:

  • Facilitate and enrich your Customer Listening.
  • Gain customer insights by leveraging your customer verbatims complementarily to your survey results.
  • Determine which areas contribute most to satisfaction or dissatisfaction and identify improvement areas.

Learn how to collect classification plans from semantic analysis by following the steps of this article.

Upstream 

  • Classification projects require contracting with one of our partners or ourselves. Depending on the case, these partners will be co-contractors or subcontractors.
  • Obtain necessary elements for connection from the partner or your Skeepers Customer Success Manager.

Partner Connection

Next, you need to connect the semantic analysis solution to Skeepers.

Go to Settings. Then on Available Applications. 

2024-06-21_16h25_44.png

Click the + Integrate button of your chosen solution.

Then Add a new authentication.

The elements to enter differ depending on the solution; here's an explanation for each solution. For Skeepers' semi-automatic solution, configuration is done by our teams during project setup.

 

Erdil

Enter the elements below that will be provided by Erdil except the name, which is at your discretion.

 

Proxem

Enter the elements below that will be provided by Proxem except the name, which is at your discretion.

 

Skeepers (auto)

Enter the elements below that will be provided by your Skeepers Project Manager except the name, which is at your discretion.

 

Viavoo

Enter the elements below that will be provided by Proxem except the name, which is at your discretion.

Remember to Save.

 

File Classification Plan

Once logged in, you can design your classification plan.
Go to the Classification Plans tab and click on Add a Classification Plan.

Creating the Classification Plan

Your classification plan is now ready to be configured:

  • Name it so that it can be easily identified.
  • Save.

The classification plan is initially empty. Concepts (and their hierarchy) will appear as they are identified in verbatims. For example, as long as no verbatim mentions "Quality / Price Ratio", this concept will not appear in the classification plan.

Additional Data

Through the Additional Data tab, determine which data will be sent to the semantic analysis solution. This data serves 2 purposes:

  • Assisting semantic analysis by associating useful information with it. For example, if your questionnaire starts with "Would you recommend our brand to your friends, family, or colleagues?" and your respondent rates it 10, then in an open-ended question asks "Why?" and the respondent says "The quality / price ratio", by associating the NPS score with this open-ended question, the semantic analysis solution can identify that here, the Quality / Price Ratio is positive as if the customer had indicated "The brand offers good value for money."
  • Qualifying verbatims in the third-party solution. If you have access to a third-party solution, this allows you to cross-reference concepts with ratings, averages, or attributes. For example, I want to cross-reference my concepts by store to identify actions to be taken based on the store.

Tips: Seek advice from your partner or Skeepers teams to assist you in choosing additional data to add.

Proceed with mapping fields:

  1. The Internal Entity corresponds to the information to be retrieved from Skeepers (here, a label called NPS on a question and the store).
  2. The Distant Field corresponds to what will be added/modified in the third-party solution.
  3. The Transformation Field applies only to very specific cases where the data needs to be modified in a very specific way (e.g., transforming a phone number starting with +336XXXXXXXX into 06XXXXXXXX).

2024-06-21_16h36_25.png

Choose your internal entity.

2024-06-21_16h38_06.png

 

Enter your Distant Field.

Don't forget to Save your changes.

 

Actions

 

Once your mapping is ready, you can use it within a Campaign action to make it work. 

In the Actions tab of your Campaign, create a scenario that fits your needs.

What is an action?

  • Then select Classification Plan
  • Select your classification plan from the dropdown list.
  • Select all verbatims that need to be sent to the semantic analysis solution.

In this case, we selected When a page is validated and If the Page is "main form" Then I trigger connector a

Don't forget to save.

All that's left is to launch your Campaign and verify in the responses that the verbatim analysis is working correctly.

Analysis retrieval time:

  • Automatic analysis: The analysis is "real-time", expect a few minutes to get the feedback. If it takes longer than 30 minutes, check the steps above.
  • Semi-automatic analysis: The analysis is completed within 2 business days. If it takes longer than 3 days, check the steps above.

 

Frequently Asked Questions

Can Skeepers connect to another partner besides those mentioned above?

 

Yes, through webhooks, our API, or raw data sharing, you can share your responses with another semantic analysis partner. This partner can then retrieve verbatims in their solution and exploit them. However, since there is no return flow, Skeepers cannot leverage the partner's analysis in the solution and use their classification plan. 

 

What is the difference compared to the verbatim analysis option?

 

The verbatim analysis option is based on a subject analysis rather than a concept analysis defined in a classification plan.

There are 2 main approaches: subject analysis and concept analysis​. Both are now available in the Skeepers Feedback Management platform. 

  • Subject analysis​
    It focuses on subjects (almost words) – e.g., delivery, customer service, price. This analysis is exploratory (i.e., the subjects to be identified are not predetermined). ​
    In this context, sentiment analysis (assigning a positive, neutral, or negative tone) is performed at the message level, not specifically at the subject level.​
  • Concept analysis: It focuses on broader concepts that combine multiple subjects – e.g., delivery delay, delivery to wrong address, unpleasant customer service. This analysis categorizes based on a pre-defined concept plan.​
    Sentiment analysis in this context is generally performed at the concept level (a message with 3 concepts potentially has 3 different tones), but sometimes at the message level globally: it depends on the provider.

We recommend subject analysis for product review analysis (and generally analyzing messages covering a wide range of subjects) or a very specific question "Which word best defines our brand?".
We recommend concept analysis for feedback on experiences (purchases, customer service...).

To learn more about the verbatim analysis option, here's an article on the subject

_________________________________________

Want to try it out? Click here

Was this article helpful?
0 out of 0 found this helpful