Creating a realtime reporting tool for Genesys flows

[Lucas1]

I wrote a blog article about a tool I created for work to allow us to track custom metrics in our Architect flows using the Event Bridge Integration and Participant Attributes.

It only costs ~$40 a month and allows us (in near-realtime) to alert on issues, do anomaly detection and track customer behaviour - all in the same reporting platform we use for the backend services our flows depend on.

Creating a realtime reporting tool for Genesys flows

I'd love to hear people's thoughts on my approach and any improvements/considerations.

[JDunn]

Hi Lucas,

This is very interesting. Could you say a little more about how you're using these metrics?

So for example, you're setting this metric when the customer's identity isn't known (e.g. not in your CRM)?

CustomerRoutingMetric_Unknown(2023-03-05T08:18:04.568Z)

What are you doing to then shape your conclusion? I guess if this value is normally set 20 times per day and then today it's at 500, something is wrong with your API call to CRM but...

Just a little more on the types of metrics you are setting and what that information is allowing you to do would be very interesting.

[Lucas1]

We use the metrics for both the run of the mill questions:

• How are people responding to a question (with quick responses) and what is the impact of changing the wording?

• How many people who had their intent recognised, was served information but still responded to our follow up question with quick response X?

• What is the breakdown of successful/failed X self-serve journeys across WM, Voice, WhatsApp etc?

• How many people entered X (Acc # etc) successfully first/second/third time, and what is the impact be of changing the wording?

• We're seeing random failures, track how many times DataAction X responds with X over the next few days to diagnose whether it is this

• etc

And leverage them using DataDog to:

• Monitor the amount of successful/failed submissions of X and alert me (via phone/SMS/Slack) if the number unexpectedly changes

• If flow metric X increases then what were X metrics/logs of the backend service it depended on at that time?

  • We are building ever more complex journeys/routing that depend on more services across the business, so this type of correlation becomes more and more useful

• During a live incident create a document (DD Notebook) to curate your investigations alongside up to date metrics

  • This was useful when looking at, and providing updates on a recent incident where customers were submitting X across different platforms (Genesys being just one) and we wanted one place to show the sudden increase in submissions and their respective success/failure rates

• X team rely on this metric (see below) so alert me if this metric drops to zero, indicating it may have been accidentally removed

• etc

We also extract a lot of conversational data (inc these metric Participant Attributes) into a Data Warehouse that contains data for other systems. Meaning Data Analysts are able to join conversations by metric(s) X to data from other systems to get a much richer understanding of a customers journey through all the various systems. I'm no Data Analyst but some questions I've seen asked using this data are:

• If a customer did X or answered X in a flow/bot, and where transferred to an agent then how long did this take the agent to resolve

• For customers who went through flow/bot X how many resulted in a 'case' being raised in X, could we update the flow/bot to reduce this

• etc

Sorry that response was quite wordy.