Skip to main content
TalkWiz

CallMarker Call Analysis with AI: Direct Connection and Real Outcomes

TalkWiz connects to CallMarker in two layers: an automation that sends every transcribed call for analysis when it ends, and a read-only API connection that brings real names, missed calls and real outcomes. You see not only what was said in the call, but also what happened after it.

Updated

What the CallMarker connection includes

CallMarker already records and transcribes your call center's calls. TalkWiz connects to it in two layers that work together:

  • An automation that sends every call (webhook): you set it up once in CallMarker, and from then on every transcribed call reaches TalkWiz when it ends, with the agent name, customer, campaign and the status the agent marked. Calls arrive already transcribed, so there is no transcription cost.
  • A connection to the CallMarker API (read-only): you paste a CallMarker Access Token in settings, and TalkWiz brings real names, links every call to its customer, imports missed calls and shows real outcomes: what was really sold, and by whom.

You can start with the automation alone and add the API later. Disconnect the API? The automation keeps working as usual.

Setting up the automation, its parameters and status mapping are covered in the CallMarker call analysis guide. This page explains what the connection gives you, and above all what the API connection adds.

What you get from every call. Within a minute of the call ending, every call gets a full analysis:

  • A score for the agent by criteria, a summary, objections, pain points, buying signals and a next step.
  • The call outcome in CallMarker's own terms: Interested, Not interested, Follow-up or Completed (and for service calls, Unresolved).
  • A comparison between the status the agent marked in CallMarker and what actually happened in the call.
  • A ready follow-up message for the customer, a smart daily report and alerts on calls that need attention.

Want to see what it looks like? Take a look at a sample analysis.

What the API connection adds

The API connection fills in what the automation cannot send:

  • Real agent and campaign names. Campaigns that arrived as an ID number get their name from CallMarker, on existing calls too.
  • Every call is linked to its call and customer in CallMarker, including the call's real duration and direction. Links are updated every hour.
  • Import of missed calls (your choice). Calls with an agent that did not arrive through the automation are imported every hour with their transcript. You can limit this to specific campaigns.
  • Analysis of past calls. Choose up to 31 days back and up to 1,000 calls, imported newest first.
  • Real outcomes. Opportunities (purchased or not purchased, with the amount) and appointments of customers you have calls for come from CallMarker and are shown next to the AI's analysis.
  • Context from earlier calls. When the customer has already spoken with the call center, up to 5 earlier calls with them (from the last 180 days) and their outcomes go into the analysis. The AI knows whether this is a follow-up call and what was already said.

Real outcomes on the dashboard and the call page

The AI identifies what happened in the call. Real outcomes show what happened after it. The dashboard has a "Real outcomes from CallMarker" card with:

  • Purchases and revenue per agent, next to opportunities not purchased and appointments scheduled and held.
  • How many of the calls the AI marked "Interested" really ended in a purchase.
  • The average call score of customers who bought versus customers who did not. A large gap means the criteria in your scorecard really predict a sale.

On the call page, the customer card shows "What actually happened": that customer's purchases and appointments. Quality and coaching managers see which behaviors in a call lead to a deal. Read more about call quality assurance and sales coaching.

Setup, step by step

Setup is done by an admin of your organization, in settings under "Integrations".

The automation:

  1. Create an automation address in TalkWiz. It is shown only once.
  2. In CallMarker, create an automation with the Call AI Analysis Finished trigger that sends a POST to that address with the transcript and the call details. The full details are in the guide.
  3. Check on the integrations screen that calls are arriving.

The API connection:

  1. In CallMarker, on the Integration information page, turn on AI assistants (MCP). CRM must be enabled on the account too.
  2. On the same page, under Access Tokens, click Create new token and copy the token (64 characters). It is not the API token at the top of the page.
  3. Paste the token in TalkWiz and click "Connect and test". The token is tested with CallMarker, campaigns get their names and the first sync starts.
  4. Choose what to sync: import of missed calls (from all campaigns or only some) and real outcomes.
  5. If you like, start an analysis of past calls.

Sending the analysis back to CallMarker (optional): in settings under "Send data back" you can add CallMarker as a destination. The analysis is added as a note on the customer card, by the customer's phone number, and existing notes are kept. Choose what to send (summary, outcome, score, next steps and more) and whether to send automatically after every analysis or only with a click on the call page. This needs the CallMarker account's API token, the one at the top of the Integration information page.

What it costs in tokens

A token is one minute of analysis:

  • A call from the automation: one token per started minute of call. The call arrives transcribed, so there is no transcription surcharge.
  • Missed calls and past calls: analyzed like any call, at the same price. During the trial, importing stops at the end of the trial calls.
  • Syncing names, links and real outcomes: no tokens.
  • A call with no real conversation, such as a hang-up right after answering: not analyzed and not charged.

Plans and prices are on the pricing page.

Privacy and security

  • The Access Token is read-only. It is stored encrypted, and only its last 4 characters are shown on screen. You can revoke it at any time in CallMarker or in TalkWiz.
  • The token is never sent to the AI. TalkWiz reads the data from its own server through the CallMarker API, not through an AI assistant.
  • Real outcomes only for customers you have calls for, visible to your organization's admins. When you disconnect the API, the token and the outcomes are deleted.
  • The automation address can only send calls, not read data. If it leaks, create a new one and the old one stops working.
  • Transcripts are stored encrypted, and sensitive details such as credit card numbers are hidden before the analysis.

Looking for a different connection? All the ways to bring calls into TalkWiz are on the integrations page.

Frequently asked questions

Do we have to connect the API, or is the automation enough?

The automation (webhook) is enough for every transcribed call to be analyzed. The API connection adds real agent and campaign names, a link to the call and customer in CallMarker, import of calls that did not arrive, analysis of past calls and real outcomes. You can start with the automation and add the API whenever you like.

Which token do we paste: the API token at the top of the page or an Access Token?

For the API connection you paste a 64-character Access Token, created under Access Tokens on the Integration information page in CallMarker. The API token at the top of that page is used only if you choose to send the analysis back to CallMarker as a customer note.

The connection failed with a message about CRM and AI assistants. What do we do?

The CallMarker API works only on an account that has CRM enabled and AI assistants (MCP) turned on in Integration information. Turn both on in CallMarker and paste the token again.

Does TalkWiz change data in CallMarker?

The API connection is read-only. The only thing TalkWiz can write to CallMarker is a note on the customer card, and only if you turned on sending data back. The note is added to the existing notes; statuses, opportunities and appointments are not changed.

How many tokens does a CallMarker call cost?

One token per started minute of call, because calls arrive already transcribed and there is no transcription cost. Syncing names, links and real outcomes costs no tokens, and a call with no real conversation is not analyzed and not charged.