The Essential Guide to Conversation Analytics

Are you hearing more and more about conversation analytics, but don’t really know what it is, how it could make a difference to your contact centre operation, or where to start with finding the solution that’s right for you?

Then you’re in the right place! As our Editor Megan Jones recently interviewed Matthew Yates, VP Engineering at MaxContact, about the ins and outs of conversation analytics to help you make a more informed choice.

What Exactly Is Conversation Analytics?

Conversation analytics, otherwise known as speech analytics, provides both post-contact and real-time capture and analysis of services and support in a contact centre.

These insights can then be used for quality assurance (QA), as well as driving wider operational improvements across the contact centre.

How Does It Work?

Conversation analytics works by drawing on all interaction data, including:

The solution then takes all that data in and analyses it – using both traditional analytics methods and, more recently, artificial intelligence (AI)-enabled methods – to create insights.

Where Can Conversational Analytics Make a Difference in My Contact Centre?

Here is an overview of the top 5 time-saving capabilities of conversational analytics:

1. Targeting Customers With Relevant Follow-Up Actions

‘Call to Action’ analytics can help you find out which products or services your customers like and establish a follow-up action off the back of this insight.

For example, sending further information via an email, or scheduling a follow-up call.

This insight can also help you really start to understand how effective each agent is and identify training opportunities for the various elements of the calls too, such as objection handling.

Thereby going above and beyond the traditional QA checklist of “did the agent introduce themselves correctly?”, “did they name the company?”, etc.

2. Demonstrating Compliance to the FCA

If you’re taking payments and need to make sure you can stand up to the scrutiny of the Financial Conduct Authority, conversation analytics can help – by demonstrating your agents’ compliance in reading out a direct debit guarantee (for example).

Analytics can also help with vulnerable customers. Not only by identifying them, but also highlighting appropriate (and inappropriate) responses and training needs.

3. Providing Feedback to Agents in Real Time

Traditionally, quality assurance (QA) would require a team of people performing that role.

However, with analytics-driven agent assist, feedback can be given to each agent in real time, as opposed to a post-call coaching session with a human being.

Artificial intelligence (AI) is taking this one step further now too, helping contact centre managers to understand how effective the agents are in a call.

This can even drill down into nuanced aspects of the conversation. For example, monitoring an agent’s ability to handle objections, ask open questions, and listen effectively.

If you are looking for best practice advice for giving feedback, read our article: How to Provide Closed-Loop Feedback With Employees and Customers

4. Autogenerating Summaries and Suggested Disposition Codes

Analytics can also save time in after-call work (ACW) too, by generating an auto-summary of the call and any outcomes.

When this is auto-populated into the case notes, agents can skim-read and make any tweaks, instead of writing all the notes from scratch – saving around 90 seconds per call.

Not only that, but analytics can suggest disposition codes – removing the chance of human error.

5. Taking Your IVR to the Next Level

However, analytics is not just about quality assurance (QA) and time savings on after-call work (ACW), it can support the end-to-end customer journey too. This includes the very first point at which a customer attempts to contact you.

For example, you can now use Generative AI to capture a much more informed conversation at the IVR stage, where there’s an opportunity to address the reason for the call BEFORE a customer even gets to an agent.

Armed with the ability to automatically transcribe and understand customer intent in real time, the system can come back with relevant suggestions from the company knowledge base and frequently asked questions.

A far cry from the static decision trees and structured IVR systems seen in previous years.

Not only that, but analytics can also support intelligent routing.

So when a human-to-human conversation is required, the system can automatically understand the intent of “why I’m calling” and route the call to the most appropriate agent – all helping customers to get speedier resolutions to their queries.

What Else Should I Be Looking Out for When Choosing a Conversation Analytics Solution?

So, you know where it can help, but what are the other details to remember so you don’t get caught out with any unexpected surprises?

Custom Quality Assurance (QA) Checks

Will you need the ability to create custom QA checks within the solution?

If you do, then you’ll need to double-check that the system is flexible enough to accommodate this, as well as support accurate analysis and reporting on these custom checks.

Not only that, but always check with your vendor that you can self-serve and go in and customize the quality checks yourself – or that it’s possible to go back to the vendor to take care of this without having to pay out more money in professional services.

Languages

Are you operating in the English language only, or do you need support in multiple languages?

If it’s the latter, then make sure you ask your vendor about which other languages the system supports. Don’t make assumptions!

Support

As with any technology, it’s important not to forget about the human support aspect, so you don’t get any surprises later on!

Are you confident in the vendor’s ability to respond to any issues that you might have? Is the account manager somebody that you can trust?

Data Security

It can also give you peace of mind to investigate the vendor’s data security – just to make sure that the data you’ll be feeding into the system is being handled securely.

Vendors should be able to provide relevant evidence of their data handling security. For example, if they are ISO 27001 and GDPR compliant, and operating securely in Azure, Google cloud, or AWS (if they are a cloud-hosted solution).

For more information on contact centre security challenges, read our article: Top Call Centre Security Challenges and How to Fix Them

How Much Budget Do I Need to Secure?

It’s useful to know that there are tiers of solutions available in conversation analytics.

These vary in cost (per month) as the available features and functionality scale up, as follows:

---

1. Post Call Analytics

Approx. £25 per user / per month

This can be summed up as entry-level conversation analytics – providing insight about what happened in your contact centre yesterday and the quality of those interactions across individual teams.

---

2. QA Scorecards

Approx. £40-£50 per user / per month

This adds to the above with another layer of insight, offering auto-QA with the introduction of QA scorecards.

---

3. Real Time

Approx. £70+ per user / per month

The top end of the scale is then all about real time, giving users the ability to analyse what’s happening live on a call, and provide next-steps suggestions or auto-alerting to vulnerability.

---

And How Do I Decide Which Tier to Start At?

Don’t feel you need to jump in at the deep end! It can be better in the long run to start with a smaller step, to realize the value, and then scale up your features and capabilities from there.

Also be sure to ask your chosen vendor to see the road map of what they’re working on and what they’ve got planned.

With thanks to Matthew Yates, VP Engineering at MaxContact, for contributing to this article.

If you are looking for more great insights on using analytics in the contact centres, you should read these articles next:

• An Introduction to… Contact Centre Analytics
• Top 10 Use Cases for Speech Analytics
• Interaction Analytics in Contact Centres – An Executive Briefing
• Customer Data Analysis – How to Analyse Data in 7 Steps

Reviewed by: Xander Freeman