You can now easily and flexibly rearrange Signals dashboards. Previously, most cards displayed at full width in a single column, and changing their order meant editing each card’s order setting individually. Now you can place reports side by side and size each one to suit its content, which pairs well with chart filtering when you want to compare reports at a glance.

To get started, open the dashboard menu and select Arrange cards. A grid appears behind the cards while you’re arranging. Drag a card by its top-left corner to move it, or drag its bottom-right corner to resize it. Cards snap to the grid, have minimum and maximum sizes based on their content, and shift automatically to fill empty space where possible. Filters are unavailable while you arrange, and nothing changes until you click Save layout (or Cancel to discard).

A few related changes:

You can now search the audio clips grid in the Montage Library by Eval Number or Conversation ID, and Conversation ID now appears as a column in the grid.

We’ve made several enhancements for building a montage in Montage Builder:

When KBo Flow lands on a research question, it then forms a cohort by filtering down the data to ensure that it’s analyzing only conversations that directly relate to the question at hand. Now, this cohort will be clearly shown in the email, so you can see exactly what was analyzed/in scope for any given KBo Flow email.

In addition to searching by Evaluation ID, you can now also search the Authenticx platform by Conversation ID!

Insights cards are able to be renamed, reordered, and filtered. This has made it more important than ever for users to be able to view the underlying data for insights cards. Now, clicking on any Signals insights card will open the drilldown menu with options to “View Data”, “View in Conversations”, or “Ask KBo”, so you can go deeper in the platform and easily explore underlying data.

Previously, the Average Score by Skill graph in Signals showed both individual questions and question groups as separate bars on the same view. We’ve added the option for users to choose whether they see only individual questions or question groups via a dropdown menu on the chart!

Previously, the question results filter did not support questions that are part of a parent group, which meant that many options for question results that users would expect to see were not available. Now, when a user accesses the question results filter, these questions will be visible for selection.

Chart PNG exports previously captured only the chart itself, so the model or question selected on the dashboard was lost, leaving exports ambiguous for anyone viewing them outside the platform. Exports for charts with a model or question selector on the chart (e.g., Model Results, Classifiers, and Question charts) now include the current selection.

You can now filter by Agent Managers to easily see how each Agent Manager’s team is performing overall. This new Agent Managers filter is available in both Conversations and Signals, and can be used for chart filtering, grouping, and in KBo.

Now, you can define a score range for any selected module in the module score filter. For example, if you would like to view conversations where the quality module score was greater than or equal to 80% and less than 100%, this is now possible! To do so, select a module, and click + Condition to add up to one additional “AND” condition for that module. As part of this update, we have also changed the operators to plain text (i.e., “greater than” and “less than” instead of > and <) to make it easier to understand what conditions you are applying.

In Signals, you can now group by module score! All group by capabilities are available on standard bar/line and donut charts, which includes the following reports: Contact Types, Conversations Over Time (line, bar, and donut), Eddy Effect by Classifier, Eddy Effect by Question Response, Eddy Effect Over Time, Model Results Count, Patient Journey Stages, Primary Themes, and Question Response Count.

To access this configuration option on these reports, go to that report’s menu, select “Configure”, and select Module Score in the “Group By” section. As always, grouping a report will only persist when applied on a custom dashboard.

When Module Score is selected for grouping, you can:

How the range builder works:

Examples of how to use this:

As an additional enhancement, there is a new chip on grouped charts to show what grouping is applied. This grouped by chip comes before other chart filter chips, and will show up in emailed dashboards as well.

When this report is filtered such that there are fewer than 20 Themes available, this report shows fewer than 20. To address this, we’ve renamed this report simply “Top Themes” so it is always accurate.

The full names of the Drivers in the filter dropdown were cut off with “…”. Now, users can see the full name of each Driver when setting up their filters.

Previously, users were limited to a max of 5 in-chart selections on line graphs, and 7 on bar charts. Now, users can select up to 100 selections on chart dropdown picklists, and the graph will paginate as usual (showing 7 per page on bar graphs, and 5 per page on line graphs).

Once a user renames a report in Signals, it is difficult to determine what the original report was (particularly with all the options for grouping and chart filtering that are now available!). Now, the original report name lives in the “Configure” menu so that it is always clear what the original report is.

Module Results and Model Results were too similarly named, causing confusion for many! Often, users would (understandably) mix these up when adding reports to Signals dashboards. We’ve changed the name of the “Module Results Count” report to “Evaluation Results”, and “Average Score by Module Over Time” to “Average Score by Evaluation Module Over Time.” Helpfully, this report will still show up when you search for “module” in the Add Cards modal.

There was an issue where users could not remove a dynamic date (e.g., this week, this month, etc.) from a saved filter. This has been fixed!

We’ve continued to invest in making the experience between KBo Flow and KBo Chat seamless. Click “Continue in KBo” via any KBo Flow email to chat with KBo about the emailed report. When you land in KBo chat, you’ll see an unread indicator on any unread chats or Flow reports, can provide feedback on the KBo Flow report via the thumbs up/down icons, and can chat with KBo about the report to get clear examples and go even deeper.

Previously, if you were having a conversation with KBo and you expanded the chat or closed the chat, the message would disappear. Now, in-progress messages continue to populate if you switch KBo views mid-message or if you close the chat. This ensures you never lose your progress and KBo continues to answer you while you work!

KBo Flow is your proactive, always-on research assistant in Authenticx. Instead of you having to log in and ask a question, KBo Flow automatically reviews any context you provide, analyzes your conversation data on its own, looks for patterns worth knowing about, and sends you research findings by email, automatically, either weekly or monthly (you decide!).

Think of it as your Authenticx research analyst that’s always working in the background on your behalf.

To set up your KBo Flow subscription, first click on the KBo icon to open KBo chat in the platform.

Open KBo in fullscreen mode by clicking on the “chat history” icon in the top right, or on the arrow on the left side of the chat panel. Then, click on “Settings” in the bottom lefthand corner.

Add in your user context to provide KBo with information about your focus areas, goals, and priorities. This will help KBo focus on research questions that will be most relevant to you. Better context in = better outputs.

Finally, select a weekly or monthly KBo Flow cadence, and click “Turn on” to activate your subscription! Weekly emails will land in your inbox on Mondays, and monthly emails will land on the first business day of each month.

Note: You can work with your Authenticx point of contact to add in org-wide client context, such as overarching goals and priorities, competitors you’re tracking, program events and dates, your glossary of terms/acronyms, descriptions of hierarchies/metadata fields. Your CSM will do this on your behalf!

What’s Changed: Smarter Answers to Layered Questions

KBo can now answer multi-part questions in a single pass instead of defaulting to whichever single technique it knows best. Ask something with several layers to it, and KBo combines the right techniques automatically and hands back one synthesized answer with its reasoning shown.

How it works:

Example — “What are the common challenges or frustrations that aren’t addressed and result in a transfer or further action?”

Now, KBo chat will also expose its “thought process” to users as it is working to generate an answer to a user’s prompt. This helps to build trust in KBo’s process, and gives visibility into what KBo is doing when it takes longer to generate more comprehensive answers.

We're excited to share a set of updates to our Standard Quality and Insights models. All of the changes below are live at the start of the day on Thursday, October 1. We timed the release to line up with the start of Q4, so your Q3 reporting reflects the current model versions and your Q4 reporting starts cleanly on the updated ones.

Three of our Standard Quality models are receiving updates: Negative Language, Active Listening, and Agent Preparedness. The other 16 models are unchanged. We expect scores on these three models to improve because their criteria are now more clearly defined. If you see a change in results starting in Q4, it most likely comes from these clearer definitions and not from a change in agent performance.

Negative Language: Did the agent avoid using negative or inappropriate language?

Many of you asked us to remove slang from this model, so we did. Slang and its example list are no longer part of the criteria, and "slang" has been removed from the rubric question. We also defined criticism and negativity toward the customer more clearly, so the model can tell real negative language more consistently.

Active Listening: Does the agent use active listening skills?

The criteria now account for times when asking for information again is appropriate. For example, an agent may need to confirm demographic details more than once, or may be handling a call that involves several patients or accounts. In these cases, agents will no longer be penalized for asking again.

Agent Preparedness: Does the agent demonstrate preparedness for the call?

We defined more fully what preparedness is and what it sounds like on a call, and added more examples to guide scoring.

With this release, the Positive Feedback and Financial Assistance models are out of beta and generally available. Please reach out to your Authenticx contact if you’d like to enable either of these models in your account.

Positive Feedback

The updated criteria define positive feedback more fully. They draw a clearer line between standout "wow" moments and everyday courtesy or routine praise. You should see fewer routine polite remarks flagged as positive feedback, which makes the moments that do get flagged more meaningful.

Financial Assistance

We defined the different types of financial assistance programs more fully, added guidance on when an interaction should be labeled N/A, and clarified how to handle interactions where more than one program is mentioned. You should see fewer false positives and clearer identification of the main financial assistance option discussed.

Interactions analyzed from October 1 onward will use the updated models. If you compare Q3 and Q4 results, keep in mind that some differences on the updated models will come from the updated criteria. Your Customer Success team can help you interpret trends across the transition or answer questions about how the changes apply to your programs.

Thank you, as always, for the feedback that shaped these updates!