Speech analytics success factors: How to get the most from your solution

As anyone with an iota of business nous will tell you, getting the right technology in place is only part of the puzzle. The technology is important, yes – but you also need the skills, processes and structures to support the tools. And so it is with speech analytics.

While speech analytics primarily sits within the call centre, it has the potential to be an agent for change across the entire enterprise, delivering insights and driving innovation and improvement for multiple departments.

Unsurprisingly, for it to realise this potential and become more than merely a quality monitoring tool, it requires support, cooperation and resources from teams throughout the organisation.

So what steps can your business take to ensure that you’re maximising the benefits of speech analytics? Here are some best practices that you can follow to give your speech analytics project the best chance of success.

Build a team and build support

When it comes to laying the groundwork, Donna Fluss, president of DMG Consulting, has suggested the following initial steps:

• Find a senior sponsor for the speech analytics initiative – ideally, choose an influential leader who does not work in the contact centre.
• Emphasise the enterprise-wide benefits of speech analytics, and put together a cross-functional team of influencers to participate in a speech analytics steering committee. This may require significant up-front “politicking” to convince non-contact-centre leaders to participate.
• Communicate the uses of speech analytics to all internal departments that can benefit from its findings, to get their buy-in and support for the new solution.
• Assemble a dedicated team that is not part of the contact centre to manage the speech analytics process. This will help to make it an enterprise application.
• Staff the speech analytics team with resources who understand the business, not just quality assurance people or contact centre supervisors.

In terms of building the team, Samantha Richardson, solutions consultant, multichannel at Webhelp UK, adds: “Have a great set of inquisitive analysts who can use the application and create meaningful insight reports. And have a healthy relationship between your insight teams and your operation and those people who are responsible for making any of the changes you need to from the insight generated.”

Build trust

Speech analytics will require a degree of tuning and configuration to optimise its handling of the unstructured data that it is analysing. Because of this, Art Schoeller, principal analyst at Forrester, recommends that businesses take pre-emptive measures to ensure that confidence in the technology does not falter.

“Contact centre operational folks are used to black and white, operational data and metrics,” he explains. “For instance, they know their current average speed of answer, and their current average handling time. It’s pretty exacting. But when you’re dealing with speech analytics, it is more unstructured data and so you’re dealing more in probabilities. For example, the likelihood that we saw this phrase coming out of speech analytics 5,000 times is 88%. It’s not exacting.

“So this raises the question of accuracy. And speech is always going to take some tuning and modification and improvement. Therefore, you need the staff in place who understand the shades of grey language and will investigate something a little bit more if they think they see a result - they will sample some calls, take a look, and better understand what the aggregate information is saying so that they can confirm or deny what the dashboard reports are saying.”

This is important, says Schoeller, because this is where he has seen a number of systems fail.

“I’ve seen companies do the proof of concept, implement the technology, get some results that they think are great, and then they get a little more into production and have a couple of instances where they’re doubting what the system is saying, or making changes based on findings and then find out that it isn’t really happening. So where I have seen systems fail is when they end up being viewed as being untrustworthy.”

Have processes in place to ensure that action can be taken

Collecting data and producing insights is only part of the process. Unless the information is disseminated to the relevant parts of the business, the organisation is wasting its potential. Therefore, it’s advised that the company establishes a formal speech analytics reporting process that allows the findings to be shared with other departments and managers in a timely fashion.

Fluss says: “Speech analytics is one of the most exciting, transformational tools that I’ve seen in the last 30 years when accompanied by the appropriate practices and resources, and supported by management. But even if you only want it for informational purposes – rather than transformational – you still need to use it, and so you need to have a process set up to apply the findings.”

She continues: “This can be on a daily, weekly or monthly basis, but on a continuous basis. You need to identify trends, and then those trends need to be communicated to the appropriate parties. And then those parties have to be held accountable for applying the findings.” 

Fluss recommends creating a closed-loop process that measures improvements on an ongoing basis, with managers recognised and rewarded for tackling and correcting the issues identified by speech analytics.

“Producing pretty charts is not what speech analytics is all about, although it’s nice to get those,” she adds. “Disseminating the information and addressing the underlying issues and/or identifying sales opportunities is number one.”

Ensure that teams work together

For the insights generated by speech analytics to become actionable, it will require the relevant departments to collaborate – something that can often demand a degree of coordination.

“It’s useful to know that you’ve got a lot of complaints coming in about credit card applications being denied, but once you know this, you need to be able to turn to the credit card department and find out what’s going on,” says Fluss.

“So when it comes to being able to apply the findings throughout the organisation, everyone needs to work together. Ideally you should have a committee of representatives from different departments that can meet regularly to discuss some of the findings, although this is not what we see too often. But as the primary organisation running this programme, the contact centre must be able to reach out and share information with other departments, so they must work together.”

To monitor and facilitate cooperation and overall effectiveness of the speech analytics project, Fluss also advises that organisations create a mechanism for tracking and reporting the progress of each issue, KPI and department on a regular basis. Furthermore, this reporting should be shared with all relevant departments and managers.

Maintain focus by using KPIs and baseline numbers

By determining key performance indicators in advance, you will bring more focus to the project, with a clearer idea of what it is you are hoping to achieve/realise with speech analytics.

ContactBabel’s Inner Circle Guide to Speech Analytics recommends that businesses put baseline measurements in place before any implementation takes place, such as how many calls are tagged with a particular issue. The team can then monitor and suggest changes to processes and approaches based on findings of the initial analysis, while measurement post-action will quantify the cost savings or alteration to other key metric.

The report notes: “The ability to see trends - to know that the instances of the words 'website' and 'password' have increased by 2,000% this week compared to the norms of the past six months - quickly identify likely pain points for the customer and potential broken processes. The continual tracking and analysis of similar information or categories over time also allows a business to see whether the remedial action that they put into place has actually worked.”

Sean Murphy, director of product marketing at Genesys, explains: “I would say that the main success factor to maximising the benefits of speech analytics is to be very focused when pursuing speech analytics projects.  I use the word ‘focus’ in reference to many different areas of foci:  focus on one area of the business, or even a single KPI at a time - don’t try to boil the ocean all at once.

“Focus each speech analytics project even further by focusing on a small group of agents at first in order to be able to test different hypotheses quickly and make changes as needed in order to perfect your approach before rolling-out any new processes to a wider group.”

Don’t waste speech analytics recordings

“First and foremost, make sure that you maximise your use of call records,” says Omer Minkara, senior research analyst in the customer management technology practice at the Aberdeen Group. “When we did our speech analytics study last year, we found that only about 30-35% of the call recordings are being processed or analysed. So that means that two-thirds of the call recording data that is collected is not utilised. It’s sitting there, and there’s a lot of information that can be captured to improve business results. So I would recommend companies to deploy speech analytics more regularly to analyse those recordings.”