# Call Log, Smart Notes & Call Quality Management

What if a phone call could help you write a client note and plan the next step? CMA connects that work. In this lesson, we will follow a call into the Call Log, through transcripts and Smart Notes, and into quality scores and supervisor review. You can watch the full lesson or jump to any chapter.

All people and calls in this video are made up. The names, phone numbers, transcripts, and AI results are examples for training. The screens show how CMA works. The phone dashboard drawing shows sample activity. It is not connected to a live call center.

Start with this map. The phone system sends call details to CMA. CMA stores those details in the Call Log. A transcript is a written record of the call. That text can support three tasks. Manual analysis answers a staff member's question. Smart Notes saves a note in the client record. Call Quality checks the call against a scorecard, or set of service standards.

PBX means your organization's phone system. For Asterisk and FreePBX, there are two parts to the link with CMA. Call records bring in past calls for review and reports. Live phone tools can show who is on a call, who is waiting, and which client is calling. Staff may also click a number to dial. Your setup and access rights control which tools you can use.

This drawing shows the main parts of the PBX Dashboard. From Common, open PBX Info, then Go to PBX Dashboard, if that choice is available. Extensions show staff phone status. Queues show callers waiting for help. Active calls show calls in progress. Other charts show call volume and talk time for the dates you choose.

The two dashboards answer different questions. The PBX Dashboard shows phone activity. The Quality Dashboard shows call scores and review progress. A large number of calls does not tell you how well staff handled them. We will come back to the Quality Dashboard after we review a sample call.

First, bring the call into CMA. The Asterisk importer runs on a schedule. It adds phone records to the Call Log. It may link a call to a staff member by extension and to a client by phone number. Check those links before using the record. More than one person may share a phone number.

Other phone systems can also send records to CMA. Five9 can send recorded call details, recording links, and any transcripts it provides. You may need to sign in to Five9 to play a recording. Other options include a recording-file import or a webhook, which is a protected connection that sends call data. An administrator must set up the right link for your phone system.

In Common, choose Call Log Detail. Pick a date range and use the agency or staff filters you need. This report shows only our twelve made-up calls. Each row shows who handled the call, its phone number, length, and links. Disposition tells you how the call ended, such as answered. Open the call ID to see its details.

On the Call Log form, check the date, call length, staff member, and client link. Use the recording to check what was said when one is available and you have access. This sample has a transcript but no audio recording. That is why its Recording field says N.

Transcription turns a recording into written text. CMA retrieves a recording that can be processed, sends it to the chosen service, and saves the text. Some imports already include a transcript. Check the status and read the text. Completed means the work finished. Names, numbers, and speaker labels may still contain mistakes.

Analyze Transcription lets you ask a set question about a call. Choose a prompt, which is an instruction for the AI. For example, ask for the client's request, agreed steps, and any missing details. CMA saves the answer with the call. Check it against the transcript. You can edit the answer and choose to add it as a Client Note.

If a transcript is missing, check that the call has a recording or imported text. If processing failed, ask an administrator to check the error and scheduled jobs. Clear Transcription starts a new processing cycle. Use it with care. The old quality result may leave current reports while its earlier evidence stays in history.

Smart Notes helps with client records. When the feature and its rules are turned on, CMA can create and save notes from call transcripts. This can reduce repeated listening and typing. The note is already saved, but it still needs a person to check key facts and decide what to do next.

An administrator opens System Management, then Call Log Smart Notes under Communications and Workflow. Each rule says which calls to use, which prompt to run, and what type of note to save. For example, a rule could create an appointment summary or a counseling note. Ask for facts from the call and clear next steps.

A call needs a ready transcript and must be linked to a client before this process can save a client note. Voicemail is not part of the standard Smart Notes process. The feature must be on, and a matching rule must be active. More than one rule can create more than one note. Call Quality runs on its own; it does not need a Smart Note first.

Open the client record and choose Notes. Look for the Smart Note subject and note type set by your administrator. Here, our made-up client, Jordan Example, has a note about preparing for an appointment. The Call Log link leads back to the call. Use that link to check the original words behind the summary.

Check details that could change the next step. What did the client ask for? Who agreed to help? Was a due date given? Here, the client asked for a document checklist. The staff member agreed to send it but gave no deadline. A useful note keeps that gap clear. It should not make up a date.

If a note seems wrong, compare it with the transcript or recording. Then add a correction through your agency's normal process. Keep the Call Log link. Assign the next task to a person. A saved summary does not send the checklist, finish a task, change a payment, or make a decision for your agency.

If no note appears, keep up with the notes your work requires. Ask an administrator to check the client link, transcript, and matching rules. A new rule does not create notes for all old calls by itself. An administrator must plan any work on older calls. This helps avoid unwanted changes to client history.

Call Quality checks whether staff showed the skills your agency expects. A scorecard lists those skills and how to judge them. In CMA, use Call Quality to review a call. Use the Quality Dashboard to see scores and review progress across calls. These are the tools we mean by the Quality Manager scoreboard.

An administrator opens System Management, then Call Quality Scorecards. Start with a draft or the sample scorecard, and change it to fit your agency. Calls about enrollment, client service, and payment changes may need different rules. The sample is a starting point. It does not set policy for every agency.

Routing means matching a call to the right scorecard. Staff agencies and companies help select the group. You can also choose staff, phone systems, campaigns, and call direction. Within one field, any selected choice can match. Across different fields, every field you filled must match. An empty field adds no limit. The lowest priority number wins. If two match at that number, the lower scorecard ID wins.

Staff company means the staff member's company, not the client's company. A quality review can run even if the call has no client link. Use Preview Matching Order with sample call IDs to check which scorecard would match. This checks the matching rules only. It does not ask the AI to score the call or check every rule needed to process it.

A criterion is one skill or rule on the scorecard. Give each one a stable key, or short label, that stays the same. Add a clear name and instructions. Set a weight above zero; this is the number of points it is worth. Explain when the rule does not apply. Mark Critical failure only for rules your agency treats as critical. Say when the reviewer must be able to identify the staff speaker.

Our sample has four rules. Introduce the purpose. Confirm the client's need. Get approval before a payment change. Name an owner and deadline for next steps. The payment-change rule is critical. But it does not apply when no payment change was asked for or made. Clear rules about when to use a criterion matter as much as its points.

Save Draft keeps your edits without changing the rules in use. Publish Version saves a new set of rules for future calls. That saved version cannot be edited. Older scores stay the same unless you choose to run them again. Clone Published Version replaces this scorecard's draft with its published rules. Archive stops new calls from matching, but keeps past results. Check your draft before choosing either action.

Let's work through a score. Each rule has a weight, or number of possible points. The first two rules pass, earning twenty and thirty points. The next-steps rule fails, so it earns zero out of twenty. No payment change took place, so that thirty-point rule is not applicable. We leave its points out of the total possible score.

The call earned fifty points out of seventy possible points. Divide fifty by seventy, then multiply by one hundred. The score is seventy-one point four three percent. It is not fifty percent, because the rule that did not apply was left out. CMA does this math from the result for each rule.

The review screen shows the same score. You can also see the original AI result for each rule. We made up these results to teach the process. They do not show how accurate the AI will be on your calls. Check each finding against the call and the scorecard's saved instructions.

Critical findings need their own review. In this made-up call, a payment was changed without the client's approval. The call scored seventy percent, but it also has one critical failure. A critical failure does not set the score to zero. Check it even if the total score looks high enough.

Unable to assess means there is not enough clear evidence to decide. It is not the same as Fail. Part of the call may be missing, or you may not know which person was the staff member. If any rule has this result, CMA leaves the overall score blank. The score is also blank if no rules apply. A blank score is neither zero nor a pass.

Read three parts of the result. Completed means the scoring process finished; the call may still have no score. Reviewed means a supervisor saved a review. Findings show what passed, failed, or could not be judged. A call can be completed and reviewed and still have a failed rule.

From Common, open Call Quality, then choose Review Quality for a call. The list shows calls waiting for review, calls that need attention, and work that is pending or failed. A critical finding or an Unable to assess result sets Needs review. A call with another type of failure may still say Unreviewed. Do not use the attention flag as your only guide.

Check the call, staff member, date, and scorecard version before you start. The version ID at the top is an internal record number. It may differ from the version number in the results list. Keep track of which saved version was used when you compare calls or discuss a finding.

Open Scoring guidance. It shows the instructions and rules saved with this review, including when a rule does not apply. Use these saved instructions to judge the result. Do not judge an older call by a new draft that was never used to score it.

Choose Highlight excerpt to find the words that support a finding. The transcript has been masked to help hide personal details. Excerpts are quotes, not reliable times in the recording. Listen to the recording when you need to check the words and have access. If a required action was not heard, the finding should say so. It should not invent a quote.

Look at our missing-deadline example. The staff member said, I will send the checklist. That gives an owner and an action, but no deadline. The Fail result fits the saved rule. We keep Fail and explain that no deadline was given. A good review can confirm an AI result as well as correct it.

If you disagree, choose the result the evidence supports and explain why. Choose Unable to assess when there is not enough clear evidence. Save Review records your decisions and keeps the original AI findings. Here, the saved review still scores seventy-one point four three percent. We agreed with the original results, so the score did not change.

Reevaluate Call asks for a new AI review. An administrator with access can queue the call using the published scorecard that matches now. Save your review work first. The old result moves into history right away. The new one may show Pending with no score until processing finishes. This action does not save your open review. It also does not erase the earlier evidence.

From Common, open Quality Dashboard. Choose the dates of the calls you want to see. You can narrow the list by agency, staff, campaign, scorecard, or version. Other filters cover scores, processing, review status, and critical findings. Choose Apply after changing filters. Clear the score limits when you want to include calls with no score.

Start with Coverage and review. Our sample has twelve eligible calls, meaning calls that meet the rules for quality processing. Ten evaluations are complete, meaning the AI has finished checking those calls. Nine have numeric scores. Five have saved supervisor reviews. Review completion is five divided by ten, or fifty percent. The completed call without a score still counts in that group of ten.

Each call counts once, even if its staff member belongs to more than one agency. Needs supervisor attention counts calls still marked Needs review. Saving a review clears that status. It does not remove a confirmed critical failure or guarantee that the call now has a score. Keep review progress separate from the findings.

Check for gaps before judging the scores. Pending work is still waiting or running. Failed work needs someone to check the problem. No scorecard match means an eligible call did not match a scorecard. Backlog shows unfinished work. Token measures show AI usage for the current results you selected, not all AI use in CMA. Ask an administrator to help with these gaps.

Scores by version keeps saved scorecards and review states separate. It averages only calls with number scores. Compare calls that used the same version. Check how many scored calls are behind each average. A new set of rules can change what the score means. A higher average under new rules does not prove that service improved.

Criterion failure rates show which rules failed most often. The rate uses only Pass and Fail results. It leaves out Not applicable and Unable to assess. Our payment-change rule shows a one-hundred-percent failure rate, but just one call was checked under that rule. Review that call. One result cannot tell you how the whole team is doing.

Override rate looks at calls with a saved supervisor review. It shows the share where at least one rule's result differs from the original AI result. Our review did not count as an override because we kept the original results. A change in this rate may point to unclear rules or poor transcripts. Review examples together so supervisors agree on how to apply the rules. Use the dashboard to plan coaching through your agency's usual process.

Start with a small group of sample calls before using this process more widely. Check staff extensions, client links, recording access, transcripts, and scheduled jobs. Use different call types, including calls that are hard to judge. Make sure supervisors can see the calls they need within their agency or company.

An administrator chooses a start date, called the activation date, and an AI model. They publish a scorecard, then turn on quality processing. A call must have been answered by a human staff member whom CMA can identify. It needs a ready transcript with text, a call date on or after the start date, and a matching scorecard. Calls handled by AI voice agents are not part of this staff-quality process.

Have supervisors review examples of Pass, Fail, Not applicable, and Unable to assess. Discuss where they disagree, especially on critical findings. Improve the draft rules before publishing the next version. An administrator must plan any review of older calls. Publishing new rules does not change old scores by itself.

Remember the full path. Import the call and check its links. Read the transcript and any analysis. Check the saved Smart Note and give follow-up work an owner. Use a clear scorecard, check what was said, and save the supervisor's review. Then read the dashboard with the call count and scorecard version in mind. CMA brings this work together so your team can spend less time piecing together calls and more time helping clients.