Automated scoring options in evaluation forms

AI-generated summary

Genesys Cloud provides automated scoring options for evaluation forms that enable supervisors to streamline quality management by reducing manual interaction reviews while maintaining consistent, unbiased assessments. The platform offers two primary automated scoring capabilities: AI Scoring and Evaluation Assistance. AI Scoring uses artificial intelligence models trained on predefined quality criteria to automatically evaluate customer interactions at scale across voice and digital channels. This approach delivers faster, more consistent evaluations, reduces human bias, and ensures broader interaction sampling. The capability provides detailed answer explanations that supervisors can use for targeted feedback and agent training. When language auto-detection is enabled, reasoning text appears in the evaluation form's configured language; users can override this behavior through API settings to generate reasoning text in the conversation language. AI scoring employs prompt filters to display answers only when the AI model achieves high confidence levels, with historical accuracy monitoring tracking AI-generated answer reliability using a rolling window of the 25 most recent submissions per question. When accuracy reaches 75% or higher, Genesys Cloud automatically displays AI answers for future evaluations. If accuracy falls below 75%, the system flags the question with a low-confidence warning and hides the AI-generated answer while continuing to monitor performance. Evaluation Assistance uses topic-based conditions representing business-level intent through collections of related phrases to identify matching interactions and automatically generate answers to evaluation questions. Topics improve recognition of organization-specific language by adapting the underlying language model to focus on relevant terms and phrases. Users can create evaluation assistance conditions by selecting whether conversations should include or exclude specific topics from a predefined topics list, with multiple conditions supporting multiple topics and requiring all conditions to evaluate as true for automatic answer entry. Setup requires appropriate licensing (Genesys Cloud CX 1 WEM Add-on II or higher) and specific permissions including Quality > Evaluation Form – Add, Quality > Evaluation Form – Edit AI scoring settings, and related analytics permissions. When creating evaluation questions, users can select from three automated scoring options: No Automation, AI Scoring, or Evaluation Assistance, and specify the Evaluation Focus setting as either Full Interaction or Evaluated Agent. The system constructs prompts automatically from question text, answer options, and help text; clarity in these fields improves AI effectiveness. Questions must be specific and contextual with supporting help text that defines evaluation criteria. The system limits questions to a maximum of five answer options and requires a minimum of in conversation transcripts to execute scoring. Character limits apply to form fields: for questions and answer options, and 6, for help text. Technical considerations include exact text matching between model responses and configured answers, requiring consistent use of accents and special characters across all form fields to prevent matching failures. Form republication resets question history tracking, while the system retains history for up to 30 days of inactivity before deletion

PrerequisitesClick to expand

Genesys provides automated scoring options for evaluation questions that allow supervisors to focus more on coaching and less on manual interaction reviews. These capabilities streamline quality management processes while helping agents continuously improve. By turning conversation data into actionable feedback, Genesys AI Scoring and Evaluation Assistance help drive measurable improvements in service quality and operational efficiency. Both capabilities support voice and digital interaction media types.

AI scoring

Genesys AI Scoring is an artificial intelligence (AI)–powered capability that automates the evaluation of customer interactions. Instead of relying solely on manual reviews, it uses AI models trained on predefined quality criteria to assess interactions at scale.

This approach delivers faster, more consistent evaluations, reduces human bias, and ensures a broader and more representative sample of interactions is reviewed—giving supervisors a more complete view of agent performance. In addition to scoring, AI Scoring provides detailed answer explanations that supervisors can use to deliver targeted feedback and support agent training.

Note:

When language auto-detection is enabled, AI scoring determines the language for reasoning text using the first available value in the following order:

The dialect configured for the evaluation form
The language detected for the evaluation form
The conversation language

Therefore, if the evaluation form and conversation use different languages, the reasoning text appears in the evaluation form’s language. This is expected behavior.

To preserve the previous behavior and generate reasoning text in the conversation language, explicitly set the form’s dialect through the API to enforce a specific language for the reasoning text.

For more information about AI Scoring guidelines and best practices. see AI scoring best practices.

Evaluation assistance

Evaluation Assistance uses conditions made up of topics, which represent business-level intent through collections of related phrases. For example, to identify interactions where a customer wants to cancel a service, a topic such as Cancellation can include phrases like “close out my account” or “I want to cancel.” Topics also improve recognition of organization-specific language by adapting the underlying language model to focus on relevant terms and phrases.

When an Evaluation Assistance condition is added to an evaluation form question, speech and text analytics identify interactions that match the associated topics. If a matching interaction is found, an answer to the evaluation question is automatically generated, helping streamline evaluations while maintaining alignment with quality criteria.

For more information, see: Understand programs, topics, and phrases, and Work with a topic.

Set up automated scoringClick to expand

Create a question group. For more information, see Create and publish an evaluation form.
When you add a new question, you can set automated scoring for each question.

The options are:
- No Automation
- AI Scoring
- Evaluation Assistance
Select your desired Evaluation Focus setting:
- Full Interaction
- Evaluated Agent
Save the question and finish the evaluation form setup. For more information, see Set up evaluation focus.

AI scoringClick to expand

You can enable AI scoring per question within the evaluation form editor to have Genesys Cloud use AI to prefill the answer on an evaluation.

Create question prompts

The prompt is constructed automatically from the text of the question group, the question, the answers, and the help text. If you rephrase unclear questions and add context to these fields, you can improve the effectiveness of your questions for AI scoring.

Examples

Example unclear question 1: “Was the customer satisfied?”

Clear Question for AI Scoring: “Did the agent resolve the customer’s issue to their satisfaction by the end of the conversation?”
Help Text: “Focus on whether the agent addressed the customer’s main concerns and if the customer expressed positive feedback or indicated satisfaction at the end.

Example unclear question 2: “Did the agent greet the customer properly?”

Clear Question for AI Scoring: “Did the agent greet the customer with a friendly tone and mention their name at the start of the call?”
Help Text: “Check if the agent greeted in a warm, polite manner and used the customer’s name within the first 30 seconds of the call, setting a friendly tone.”

Prompt filters

Configured questions for AI scoring use a prompt filter, so Genesys Cloud only provides answers to questions where the AI model’s answer has a high confidence level.

Answer limit: You cannot save an AI scoring configuration for questions that have more than five answer options.
N/A or no evidence: In this case, the AI model has no sufficient information to answer a question. Genesys Cloud displays the context when the AI model cannot provide an answer.
Historical Accuracy: Genesys Cloud continuously monitors the accuracy of AI-generated scoring answers using historical evaluation data. Accuracy is measured by how often human evaluators modify or override an AI-provided answer. When an AI-generated answer reaches a sufficient level of historical accuracy, Genesys Cloud automatically displays that answer for future evaluations that include the same question.
Historical accuracy is calculated using a rolling window of the 25 most recent submissions for each question. If accuracy within this window falls below 75%, the question is flagged with a low-confidence warning and the AI-generated answer is hidden from evaluators. Even when an answer is temporarily hidden due to low historical accuracy, Genesys Cloud continues to save the AI-generated response and monitor its accuracy as new evaluations are submitted. Once accuracy rises above 75% again—based on the same rolling 25-submission window—the low-confidence flag is removed and the AI-generated answer becomes visible to users.

Notes:

AI scoring runs only when a conversation meets at least one of the following criteria. If neither condition is met, the system does not perform an evaluation.
- The transcript contains at least 300 characters
AI scoring enforces the following character limits:
- Question field: 500 characters
- Help Text field: 6,000 characters
- Each answer field: 500 characters
AI scoring performs an exact text match between the model’s response and each configured answer option. Inconsistent use of accents or special characters (for example, tildes in Spanish) across the question, help text, and answer options can prevent a valid response from matching an answer option. This may cause the Default Answer to be selected or result in a low-confidence error. To avoid this issue, use accents and special characters consistently throughout the form.
You can configure up to five answer options for each AI scoring question.
If you change and republish a form, Genesys Cloud resets the tracked history of questions with configured AI scoring.
Genesys Cloud tracks the history for a particular question for up to 30 days. If there is no activity for the question after 30 days, the history is deleted.
For more information about AI Scoring guidelines and best practices. see AI scoring best practices.
For more information about how to create Agent Auto complete evaluation forms, see Agent auto complete form type.

Evaluation assistanceClick to expand

Add an evaluation assistance condition to a form question

Create an evaluation form question.
Click Add Evaluation Assistance Condition that is associated with the answer for which you want an answer to be automatically generated.
From the select if conversation list, select if you want the conversation to include or exclude one or more topics.
From the these topics list, select one or more topics.
(Optional) Click the Add Evaluation Assistance Condition option again to add another condition to the same answer.

Notes:

A topic cannot be used in more than one evaluation assistance condition.
When you delete a topic, the condition that includes the deleted topic is no longer valid.
An evaluation assistance condition can only be configured for one answer per question.
When more than one topic is selected for a single Includes condition, only one of the topics must be found for the condition to be true.
When more than one topic is selected for a single Excludes condition, only one of the topics must be found for the condition to be false. For the condition to be true, all topics must be absent from the interaction.
When more than one condition is created for a single answer, all the conditions must be true for the answer to be automatically entered.

[NEXT] Was this article helpful?

Get user feedback about articles.

Loading...

Automated scoring options in evaluation forms

AI scoring

Evaluation assistance

[NEXT] Was this article helpful?