Speech Topics Matched Report
The primary purpose of the Speech Topics Matched report is to determine how many calls, or calls with evaluations, have at least one match to a particular Topic (as organized by the report’s layout grouping, row, and column). The report takes the form of a pivot table, similarly to the Average Score report.
- By using the column and row selections, users can sort the report by different data sets, such as calls that were evaluated; this gives more precise results for customers also purchased the Quality Management module.
- Users are encouraged to use filters to obtain results. In some implementations, requesting all call records can result in a very large data set. This can result in a long waiting time, as the execution time of the report is directly related to the amount of data in the report. Using filters ensures that the relevant data is returned quickly and efficiently.
- Users can take advantage of a flexible environment to process data, discover trends, and make meaningful business decisions.
Note: This report is not designed to show how many times per interaction a match occurred; the report shows users whether a match occurred or not.
Differences from the Average Score Report
Speech reports can be associated with two metrics: matched or unmatched. In these reports, the Overtalk and Clarity averages are represented as single percentages. Overtalk is a measurement of how long one party talks over the other. Supplementary data is displayed in the Additional Columns drop-down. The Clarity measurement gives a percentage to indicate how clearly the agent’s voice is perceived by the speech engine.
Without a Topic, no one is interested in a yes-or-no result (all Topics that matched, and all that didn’t). For this reason, Topics must be carefully validated to ensure that either a row or column contains the Topic.
Criteria Specific to Speech Reports
In addition to evaluation criteria such as “Evaluator”, the following Speech Data Elements are added for the Quality Management module:
- Non-Integer elements:
- Topics (Variable)
- Sentiment (Very Positive, Positive, Neutral, Negative, Very Negative)
- Integer elements (0-100%):
- Topic Match Percentage
- Topic Not Matched Percentage
- Sample size (x out of y) for Topics
- Sample size, overall for other metrics
Accordions allow users to group results. By default, the first record is expanded, and all other records are collapsed.
Speech Core Reporting
The user can refresh the reports to pull new or changed data based on selections. The Speech reports allow the current screen view to be printed. Date picker controls allow the user to change the date range, or use predefined date ranges. The user can access filtering details as well. Any information can be highlighted by using basic search, but advanced search is also available for more comprehensive and precise filtering.
These reports offer users multiple benefits. Users can:
- Save custom reports.
- Print the report as formatted by the browser.
- Run reports on specific date ranges.
- Filter, search, and highlight any data.
Speech Report Main Navigation and Fields to Filter
Users can now filter based on interaction, Topic, and evaluation data. Users can select whether the agent or customer produced the Topic, and which evaluation question might have been answered in a particular way. At this point, users can filter the results and analyze the data based on those criteria.
Filtering Use Cases
The following use cases apply to the current filtering capabilities of the application:
- A user can filter all calls with evaluations in which the question “Was the caller satisfied?” was marked as “No,” and can use this in combination with “Sentiment” to generate a deeper analysis.
- A user can filter “Goal Met” = “No” to see all agents who have not met their goals. They can then review all calls for those agents based on the Topics selected.
All Speech Analytics reports give users the ability to export all the information they choose to display in the report to Excel, in the same format as shown in the report.
Customers need to be able to export the information as it is shown on their screens. The ability to export to Excel provides this capability. You can:
- Export all data as calculated and shown on the screen.
- After exporting, modify data that includes additional sums and Excel functions.
- View exported data in a rich format.
- Export conditional formatting along with primary data.
Conditional formatting is accessible from the main navigation bar. Conditional formatting allows users to apply a pre-selected color to highlight data in any field, based on a defined threshold set point. For example, users can highlight any results under a certain value in red. This applies to any of the available metrics in this report.
Conditional Formatting Use Cases
Conditional formatting is highly flexible, and can be used to enhance your data searches in many ways. However, a few examples are as follows:
- A user wants to check whether a Topic match is less than the desired amount, or more.
- A user wants to check whether a percentage value is less than required by the business, and whether they need to take action.
- A user sees exceptional performance in one particular metric, and wants to highlight that metric in green.
Edit Panel Main Features
Users have the ability to name a report which is saved only for their use. Users can select which metrics apply to this report. In addition, users can hide and show results related to deactivated agents, subtotals and grand total, and goal subtotals and grand total on evaluated calls. Users can also sort the report alphabetically, and see color differences where applicable to call out these totals for more efficient viewing.
Edit Panel Use Cases
A few of the many use cases for customized reports are as follows:
- A user wants the ability to access parameters regarding setup of report data in a simple way that is clear and easily accessible.
- A user wants to exclude agents who no longer work for the company.
- A user might not want to see the totals or grand totals, and just see the basic results instead.
- A user might not want to see the goal totals or grand totals if they are running the report on matched calls and not scored calls.
Edit Panel Metrics
Users have the ability to report on four different metrics. They can see how many calls were matched to a topic, or not matched. Users can also see statistical results on the amount of time agents spent talking over a caller, and vice versa. In addition, users can measure the clarity of the voice of the agent, as determined by the speech engine. For a more comprehensive look at your quality of service in this report, all of these metrics are now available to be reported against a rich set of dimensions.
Panel Metrics Use Cases
Users might want to see a different metric, and have an additional way to identify improvements which can be made to their call center. The Overtalk and Agent Clarity values can help show additional statistics in this regard. Some examples of this are:
- A supervisor wants to identify where agents and callers disagree, and uses Overtalk as a metric and overall sentiment as a dimension to determine this.
- An auditor wants to see how Overtalk affects customer satisfaction, filter results based on template data involving customer satisfaction, and compare to the Overtalk statistics.
- Less-than-optimal results on speech-to-text transcription occur regularly for certain users, and the business needs to analyze the clarity of those agents. Perhaps these agents have issues with their phone equipment, which also results in customers having difficulty hearing and understanding the agents.
Users have the ability to create reports on the following data fields: Agent, Topic, Main Group, Labels, Overall Sentiment, Supervisor, and Trainer. Reports can also be created using Evaluator and Template Name, and automatically narrow the results of the report to interactions that have evaluations attached.
Grouping Use Cases
A few of the many use cases for report groupings are as follows:
- A user wants to determine the overall sentiment of calls and compare them to groups of users, as well as to particular Topics.
- A user wants to pinpoint campaigns and analyze ways to improve them, so they label calls that have a desirable outcome for those campaigns. The users then create a report on overall emotion and Topics. Finally, the users listen to those calls and label them.
- A user wants to look at all calls that have been evaluated, organize them by Topic, generate a search for evaluations by Topic, and include additional labels or notes.
Users can now include additional data on the left side of any report, provided that the data meets validation criteria. The following fields can now be included as overall statistical information: Agent Clarity Percentage, Average Call Duration, Hire Date, Overtalk Percentage, Percentage of Records with Evaluations, Evaluation Average Score, and Evaluation Goal Average.
Column Use Cases
A few of the many use cases for additional data columns are as follows:
- A supervisor wants to see how many calls were evaluated based on a particular Topic, and chooses to increase the number of evaluations performed.
- A user wants to target cells of an average duration, which were possibly defined by a label.
- A user wants to organize all calls and Topics for a group of newly-hired agents, so the user sorts each agent’s hire date from newest to oldest, and looks at the most recently-hired agents.
Users now have the ability to drill down to data at the individual cell level, and provide a circle or bar chart depicting the data that is in the selected cell. A user can hover over the value to drill deeper into the relevant data, launching a new browser tab that contains the data records from the report selection.
Customers are looking to find root cause analysis on a number of issues. The slide-out allows for deeper data analysis to get to the root cause of issues or unusual data. Some examples of this are:
- A user wants to see why a particular agent’s use of a particular matched Topic is unusually high or low.
- A user wants to expose all results from the slide-out in a search window, so that they can review and evaluate calls that might have an unusual number of matches or non-matches.
- A user wants to determine where in a call a particular Topic is mentioned, as this report does not go to the interaction level, and the overall values sets are too large. In this case, the user can quickly filter to narrow down to the relevant results.
- A user can use the single-Topic Overall slide-out for multidimensional selections.
- A user can see overall subtotals of a single dimension (organized by a pie chart or graph), allowing them to filter results at a glance and drill into these filtered results for deeper analysis.
Slide-Out Use Cases
Users must be able to sort and filter by different dimensions. Using the Overall slide-out, the user can see data sliced individually. For example:
- A user wants to see all Topics associated with one user, and they would like to see how this selection is divided by that user.
- A user wants to be able to see all applicable labels for a specific Topic, and proceeds to drill down even deeper by label.
- A user wants to see how supervisors compare to a specific Topic, and proceeds to drill down to data outliers for deeper analysis.
- A user wants to analyze evaluations on Topics, and needs a quick way to see Topic data as it relates to the evaluator.
Single-Topic Slide-Out with Subtotals and Grand Total
Users can now see subtotals on rows and columns, presented as a circle or bar chart. They can hover over these selections to launch a second browser tab, revealing those individual interactions in a search window.
Slide-Out Total Use Cases
A user is looking at sums of data for different types of dimensions. Some examples of this are:
- A user wants to compare the Topics of one supervisor to those of another, in order to determine whether an issue is related to an agent or their management.
- A user might want to analyze why a particular group of users consistently performs differently from another. The user labels some interactions which are interesting, and in order to analyze those interactions deeper, the user then runs the report based on a label grouping.
- A user might want to drill down to the summaries of the labeled and grouped calls, for even deeper analysis, and needs to access this data from a report directly, instead of needing to create a search filter.
Open topic with navigation