8 Tricks for Speech Analytics Success

Kveta Vostra offers best practice advice for dealing with all aspects of speech analytics, from staffing to quality monitoring.

1. Staffing

The best technology and best systems in the world mean nothing without the right – not the best – people. The more advanced your system, the fewer people you’ll need, so you can take the time to get it right and train them well.

Be careful how you set their motivation – make sure they’re incentivised to value what you want them to. It’s important to get stakeholders from different departments on your speech analytics team, don’t think of it just as “an IT thing.”

Find representatives from different departments who can help identify business needs, objectives and requirements from various company areas. Naturally, you need staff who own the project as well – and people who can collect the important terms and search objectives from their departments.

2. Get top management buy-in

Design the project in a way that ensures C-level management sees the need for it and will use a little muscle to support it.

If you can show how the project will give the company an integrated, holistic approach to customer service and improve customer satisfaction through a greatly improved experience, you’re well on your way.

3. Don’t skimp on staff

This is important enough to merit its own point.

Consultant Donna Fluss recommends setting up a speech analytics team with “quantitatively oriented business analysts”, highly experienced in performing searches and analysis on mountains of data, and having all your speech analysts spend “at least one week” with calls before even starting formal training, to give them “context” when they get busy interpreting phone conversations.

4. Remember the whole customer

Speech analytics is one way to hear the voice of the customer, but remember it’s only one part of all your customer interactions.

Speech analytics isn’t a magic potion, it’s not a silver bullet – it’s an essential, integral part of an overall customer experience. People call you for different reasons than they email or chat – they usually have more complex problems if they’re calling.

Speech analytics adds part of the overall picture. “Voice of the customer solutions combine multiple, traditionally siloed technologies associated with the capture, storage and analysis of direct, indirect and inferred customer feedback,” according to analyst firm Gartner, identifying “Technologies such as social media monitoring, enterprise feedback management, speech analytics, text mining and Web analytics” as necessary for “a holistic view of the customer’s voice”.

5. Start modest

A project that could bring a perfectly acceptable return with the correct goals will be judged a failure if too much is expected too soon.

Be satisfied with incremental gains, tweaking the system and going deliberately enough to allow for adjustments along the way.

6. Set the right goals

What do you want your speech analytics solution to do? Be specific. Do you want to identify processes for improvements? Improve agent training and monitoring? Improve compliance? Whatever goal you choose, have you established the baseline before implementing the system, to benchmark where you are now?

It’s very difficult to know if the speech analytics system is helping you reach your goal if you don’t know where you were when you got it.

7. Make results actionable

Do you have a specific plan on paper – or Google Doc or Microsoft Word doc, come on, you get the point – to take actions based on something you discover using your speech analytics tools?

Does it get written up and tossed into the ether, or is there somebody with ownership? Do you need to make organisational changes to allow your company to act rapidly on issues?

8. Plug speech analytics into your quality monitoring

Much of the value of speech analytics is in how it can save time and effort for quality monitoring. In the past, supervisors had to listen to random calls and hope that there were identifiable issues to deal with.

With speech analytics, they can identify the optimal calls to review, sort out all calls that match certain criteria, such as “sales calls that failed”, or “agents with highest conversion rates”, or “calls with compliance issues”.

With thanks to Kveta Vostra at ZOOM International