How can you measure speech analytics' ROI?

Implementing and running a speech analytics programme is no mean feat, requiring considerable time, resources and – of course – investment. Indeed, the costs of running such a project can stack up over time, and businesses need to consider not only the IT costs of implementation and license fees, but also the likes of bandwidth (if hosted offsite), maintenance and support agreements, additional servers to support the audio processing and analysis, and ongoing training.

For this reason, it is imperative that you are able to demonstrate to the bean counters that speech analytics can and does deliver a worthy return on this investment. It’s little surprise that when ContactBabel surveyed thousands of contact centre professionals for their views as part of ‘The Inner Circle Guide to Speech Analytics’, by far the most important issue they highlighted was how to build a strong enough ROI case to get the required corporate buy-in.

So where do you start?

Broadly speaking, speech analytics projects can deliver return on investment via a variety of contributions that reduce costs and increase revenue. ContactBabel has listed the following:

Cost reduction:

• Reduction in headcount from automation of call monitoring and compliance checking.
• Avoidance of fines and damages for non-compliance.
• Reduction in call volumes after understanding why customers are calling, and acting to optimise any broken processes elsewhere in the organisation (e.g. website, marketing, distribution, etc.) that are causing these calls.
• Reduction in cost of unnecessary callbacks after improving first-call resolution rates.
• Avoidance of live calls that can be handled by better IVR or website self-service.
• Reduced cost of quality assurance and monitoring.
• Lower cost per call through shortened handle times and fewer transfers.
• Lower new staff attrition rates and recruitment costs through early identification of specific training requirements.

Revenue increase:

• Increase in sales conversion rates and values based on dissemination of best practice.
• Increase in promise-to-pay ratios (debt collection).
• Optimised marketing messages through instant customer evaluation.
• Reduced customer churn through dynamic screen-pop and real-time analytics tailoring calls to the customer.
• Quicker response to new competitor and pricing information.

Art Schoeller, of Forrester Research, elaborates on where hard dollar measurable ROI can be realised by speech analytics.

“In terms of quality assurance, some of the ROI is that you can analyse more calls,” he explains. “But in more significant-sized contact centres, where you might have a quality assurance team of three or four people whose only job is to listen to calls, you might be able to downsize the team by a body. That’s an ROI lever that’s hard dollar.

“In terms of process analysis, a lot of speech analytics case studies demonstrate that an enterprise was able to identify some process that could be improved, and at the end of that was a reduction in cost. Certainly it’s harder to tie customer satisfaction and Net Promoter Score to hard dollar ROI. You can look at a process where you can improve call handling or solve a problem and build some automation to better handle that problem, or identify there’s a product problem. But process analysis is always tied to whatever ROI you can get out of process optimisation.

“And in the risk and compliance category, fraud is a big issue now, and all you’ve got to do is detect two or three fraudulent calls and that can be worth a lot of money!”

A further recommendation is to ask the vendor to help you create an ROI to justify the project in terms they understand. Many vendors have tools which can be used to estimate ROI, sometimes based on case studies of customers with similar operations, and they will of course be keen to share these with potential clients.

Best practice

When it comes to monitoring and measuring the ongoing impact of a speech analytics programme, as a course of best practice it is advised that businesses put baseline measurements in place before any implementation even takes place. This will enable you to quantify the benefits of any steps that are taken as a result of speech analytics, measuring the new metric against the pre-installation baseline figure.

“It is vital to have benchmark data before a solution is deployed,” emphasises Mark Pritchard, customer experience specialist at Kcom. “A combination of agent performance and customer satisfaction metrics as well as customer loss vs customer retention should be obtained through analytic alerts.”

Donna Fluss, president of DMG Consulting, is in agreement. “Measuring ROI is actually really simple to do, as long as you remember to baseline what was going on before you started the process. Using average handle time (AHT) as an example – if you are a large busy shop, and you know your AHT was 180 seconds before you put it in, and by applying it on a consistent basis for three months it goes down to 175, that’s a huge difference.”

She continues: “If you’re using it to improve compliance to collections rules, and you know from the monitoring and auditing that you’re doing that you initially started at a 75% compliance level, and then because of speech analytics and being able to identify issues on a timely basis, you get compliance up to 85%, you don’t have an obvious payback, but your risk of being fined goes down.”

Metrics to measure

Elaborating on the types of metrics that could be baselined, Sean Murphy, director of product marketing at Genesys, adds: “The metrics to measure the ROI of a speech analytics solution must be the KPIs that the organisation is most focused on improving, whether those are cost reduction type metrics such as AHT or call volume, revenue improvement metrics such as sales or collections conversion rates, or customer experience metrics such as first call resolution rate, customer satisfaction rate or Net Promoter Score (NPS).”

Emphasising the importance of best practice, he adds: “Regardless of the KPI in question, in order to measure the success of any speech analytics project (and therefore the ROI), it’s important to follow proper statistical procedures for the measurements to be valid. For example, I’d advise always creating both test (or ‘treatment’) groups on which the speech analytics findings are applied (through training or changed processes) as well as control groups which continue ‘business as usual’, so that the true impact of the changes can be properly measured.” 

So, speech analytics is perfectly measurable, and its ROI is clearly demonstrable. And by following the best practices as outlined above, organisations can get a very accurate picture of the financial contribution that the technology is making to the business.

And for those still having a hard time persuading their organisation that there will be a return on investment in the first place, Omer Minkara, senior research analyst in the customer management technology practice at the Aberdeen Group suggests an alternative approach to quantifying ROI.

“Many companies are still confused about the business value that it would bring, even though they know it’s a valuable tool. The reason why adoption is still where it is today – which is about 30% overall – is because many companies are still struggling to build a business case,” he says.

“What I would recommend is that you understand and identify the opportunity cost of not deploying a speech analytics system - what happens if you don't gain the insights from running your post-call analytics or real-time analytics through speech?  How many of your customers do you lose, and how many of those lost customers could you have saved if you had the speech analytics in place? How much could you improve agent productivity if you had a speech analytics system, versus how much do you lose related to agent productivity because you don't have one?”

He adds: “Once you run those math in the back-end, you can quickly understand what it can do for your business from a quantified perspective.”

Once implemented, ContactBabel research estimates that overall the time taken for a speech analytics solution to pay for itself can be as little as six months (varying up to 18 months). If you have followed best practice and can therefore demonstrate this fact, it will go a long way to saving you another battle with the bean counters by justifying the expansion of the speech analytics project into other areas.