Speech analytics and recording

You can listen, but now you can learn what your callers are really saying

Trawling through hours of calls takes time and will not give you a true picture of how your call centre and your company are serving customers.  Speech analytics is the answer, say our experts – and they have some innovative ideas.

You know what they’re saying, but what do they mean?

THE potentials for speech analytics (SpA) are enormous -- but only for those who accept it as an enabler for a solution rather than a product which will deliver on its own.  Current costs mean that RoI for SpA is limited to medium-large contact centres. 

Call centres already recognise the need to listen to what’s going on.  Voice and screen recording and quality management have been available for years and continue to evolve and deliver good results and RoI.

But the more savvy contact centres are also questioning the content of their calls.  The cost of communication is high, therefore unnecessary or inefficient communication should be reduced.  Competition is high, so knowing how customers react to offers and service is important.  If the worst happens and customers complain then the ability to analyse and rectify the causes will ensure customer satisfaction.

For companies with high call volumes SpA offers the ability to analyse the content of their calls efficiently and consistently.

Current solutions utilise three technologies to categorise calls:

Keyword spotting, probably the most common and most cost-effective, enables calls to be categorised automatically by looking for those containing certain words or phrasing.  These calls can then be manually evaluated, used for coaching or used as part of root cause analysis.  Set-up and configuration costs can be in the range of £10,000s, depending on the scale of the call centre.  Currently this should be considered the entry level SpA and where most companies are likely to see a RoI.

Phonetic indexing is the next level of sophistication.  This requires more detailed processing to translate calls into a database of phonemes.  It is slower than keyword spotting and uses a lot more disk space, but it does allow faster repeat searches once the phoneme translation has been completed.

Speech-to-text transcription is the most expensive and the most powerful.  Calls are automatically transcribed into text files which can be analysed quickly and as many times as you like.  It relies on a high level of accuracy which can only be delivered by training the system to recognise words in the specific call centre domain.  This training requires a lot of manual interventions, so the time and cost of implementing speech to text can be high, typically £100,000-plus.

The speed of SpA results is one of the areas which are likely to improve.  Current solutions are post-process, which means the results are available some time after the event depending on the resource available to analyse the calls.  In future this is likely to speed up to the point where near real-time is available.

In the future we’re likely to see the emergence of more sophisticated SpA which will allow age or gender spotting.  Imagine being able to determine the age of your caller, in real time, and route them to customised services.

Mike Murley is sales director of ASC Telecom UK; m.murley@asctelecomuk.com

 

Let them listen the easy way

  

MOST organisations which sell over the phone are required to record calls – for very good reasons.  Customer disputes are all too common and call centre liability is substantial.  Specific disclosures must be read as a part of the transaction and the proof of compliance with this process must be available in a form of verbal receipt.

For example, a specific marketing agency will allow listening to recordings if the client asks for it.

In one actual example, a credit card holder disputed a charge for extra protection which had appeared on his bill (he must have agreed in an earlier call).

 

The agent said she could cancel the protection plan and refund one month, but the client wanted all of the months refunded.  He was told this was not possible; he had agreed to the extra protection.

The customer then insisted on proof -- he wanted to hear the recording where he had agreed – and the agent said she could request a copy of the recording from the call centre’s system.  She could not immediately provide the refund and resolve the dispute.

Common practice in situations like this is to burn the recording onto a CD and send it to the customer.  Some have the capability to email it to the customer as well. However, sending out an actual audio file and leaving it in the hands of a customer requires a series of approvals within a call centre – as a minimum, by a legal department, supervisor, and a QA manager.  Inevitably, this runs up the cost of call resolution and stretches the wait time for the customer, which impacts customer satisfaction (and provides opportunities for negative word of mouth until the issue is finally resolved).

Now, there is a new technology, in the form of a “perishable PIN number”, which provides a practical a practical solution.  This is how it works:  the agent sets up access to the recording (in accordance with the call centre’s policy) and assigns to it a PIN number that expires after a period of time determined by the agent.

The customer can then listen to his/her original call simply by dialling into the call centre recording system via any phone line.  The secure-access PIN will then automatically trigger playback of that one recording.  All of this can be taken care of within seconds of the customer request.  Handled once and dealt with.

 

Dominic Martin is marketing manager of Data Track Technology; dominic.martin@dtrack.com

 

On premise vs. hosting: now there’s a third choice

 

THERE’S traditionally been a big difference between the dialler you buy and the hosted version.  With the first you would expect to have a more integrated and effective dialler but the cost would need to be swallowed upfront.  With hosted you would pay very little for the dialler but pay as you go for the minutes. 

In call centres the biggest cost is people and the minutes.  With a hosted dialler the up-front technology cost is lowered, but to access the system you’ll need to be logged in all day (that’s a telephone call you pay for) with call minutes on top, charged at a higher rate by the supplier. What that means is that hosted diallers are often more expensive, but the additional cost isn’t always visible, as call minutes tend to come from someone else’s budget.

 

But it’s no longer a binary choice.

What’s changed is the ubiquity of the data centre. Customers no longer need to worry about how and where they host their data if they have an on-premise system, as a new breed of hybrid suppliers have sprung up to seek to offer the best of both worlds.

My company, for example, provides the dialling equipment that customers can install an "on premise" system in our own data centre, linked securely to the customer’s own network.  We host the calls and manage the system, which means customers gain from the far greater integration benefits, enhanced features and ease of management of on-premise systems, but paid for through Opex rather than Capex -- a huge advantage for companies wanting to be agile with IT.

 

Compliance: my top three tips when buying a dialler

 

Is it fit for purpose?  It’s easy to comply if you want to.  For example, I could run a Formula One car on UK roads and easily comply with CO2 emission regulations.  All I need do is turn the engine off and get it towed!  But what call centre managers need is something that is fit for purpose -- and can perform while complying rather than merely complying.

 

Consider its provenance  If you choose a system from a US vendor you’ll need to check its compliance with UK rules.  In our experience these vendors can make their diallers compliant, but this may be to the detriment of its performance, as my F1 example demonstrates.

Ofcom regularly alters regulations and we, as a UK based supplier, react to ensure our customers are compliant within months.  US vendors will always prioritise their home (and likely biggest) market which means UK customers often wait up to year to fall into line.

 

Does it self-regulate?  With diallers, as with most things in life, you get what you pay for. So my advice is to buy a decent dialler that is able to self-regulate and manage itself to avoid breaching compliance.  Otherwise you have to pay someone to do it manually, with the associated employment costs, which also leaves you open to user error.

 

David Groves is product marketing manager of Azzurri Communications’ contact centre practice; david.groves@azzu.co.uk

 

There’s little point in recording calls if you can’t find them

 

CALL recording offers little value if calls can’t be quickly accessed when you need them.  Technology advancements that greatly streamline search and retrieval, generating tremendous benefits in increased productivity and time savings, are a fantastic asset and should most definitely be on your wish list.

It’s far more efficient to conduct a search when there is additional information about the calls.  Technology is available that captures the call audio and, optionally, synchronised screen activity, with other data about the interaction all encapsulated together.  This includes information from the communications network, the telephone system, text-based annotations and call tagging.

This additional metadata also provides greater insight into the interaction — annotations provide supplementary context around the call content and call tagging allows users to append important notes with just a click.

 

If you choose the right recording system, all of this embedded information within the media file can become searchable metadata.   And fast and reliable retrieval can be essential in resolving a dispute or demonstrating that certain rules have been followed.  In addition, others in your company -- including marketing, sales and operations teams – can find and listen to recordings to gain business insight that previously went untapped. 

Securely and easily sharing calls is especially useful for companies looking to better leverage the business intelligence from their call centre interactions.  Typically, this would involve emailing an attachment of an MP3 or WAV file of the call.  Drawbacks commonly encountered with that include the inability to share only a specific segment of the call and a lack of control over what happens to the recording after it is sent.

With the right technology, those stumbling blocks can be overcome.  Select portions of a call can be exclusively shared and access to the encrypted PVD file can be granted via a secure link.

Security concerns are addressed as the file itself never leaves the central repository on which it is stored and access permissions can be set to expire after a set time. Permissions are granted or restricted to determine whether recipients can further annotate or share the access.  These advances enable call centre agents to collaborate with colleagues to facilitate accurate, secure and efficient information sharing.

 

Kevin Burns is managing director of Oaisys; kevin_burns@oaisys.com

 

How to make speech analytics work for you

AGENT skills and abilities vary dramatically, so it’s important to monitor them to target training and mentoring to achieve improvements in sales conversions, customer service and productivity.

This is where call recording combined with speech analytics comes in.

There are systems available that give each call with a unique identification number making it easy to search for call recordings by agent, line, date, time or spoken phrase. 

Once in place each call can be automatically graded with a score – such as high, medium or low --according to the presence of a phrase, for example: “Welcome to Your Company”.

 

Each call can also be scored according to the number of times a certain phrase is heard, e.g. “Is there anything else I can help you with?”

Mentoring can then either focus on a playback of the targeted segment of the call recording or the complete call recording. Playback of targeted segments can save significant training time and improve the effectiveness of the overall training schedule.

Speech analytics can also be used to form a view on call topics or market intelligence.  And, with outcomes associated with a customer record, callers can be played a reminder recording the next time they call.

Bear in mind that it is imperative that callers are warned that they are going to be recorded, given a menu or IVR selection to acknowledge this before call recording is switched on and delivered to an ACD skills group.  It is also essential that a comprehensive set of user permissions be put in place to ensure that listening to recordings can be restricted or delegated to mentoring staff. 

 

Joseph Moussalli, product manager, NewVoiceMedia; joseph.moussalli@newvoicemedia.com

 

IP is making recording tricky…and it could be serious

 

NOW that more call centres are adopting IP, recording calls can be significantly trickier.  Unlike traditional TDM networks, there are no dedicated trunks or extensions to record.

Nearly every call centre’s IP system has some variation; each has particular quirks and processes that can make setting up voice/data recording less than straightforward.  Although suppliers have standard ways of implementing their systems, they do not always work flawlessly.

Previous experience of testing and operating IP networks can help call centres ensure that their recording systems operate effectively. Identifying and resolving problems can be very challenging.  It can mean changes to the network software and how this is configured.  For smaller call centres without a dedicated IT manager, it can be worth enlisting an external expert.

 

Even once a recording system has been installed, it can still be difficult for call centres using IP networks to tell whether all of the required calls are being made recorded or whether some are being “lost”.  The implications of missed calls can be serious: as well as preventing effective performance monitoring, it can expose you to legal issues and even fines.

My advice to ensure that all calls are being recorded is to establish a link between your recording system and the data held on your CTI system.  This can help validate both the number and length of the recordings that are made, ensuring that this matches the total volume of calls processed.

In addition, have in mind that one of the most common complaints call centres receive from consumers about their technology concerns poor voice quality.  Sub‑standard recording quality can impact on customer service levels and can hinder your ability to accurately monitor performance.

In order to ensure recordings are of the right standard, I suggest that your people run tests to validate voice quality.  These work by automatically playing a reference sound clip to the call recording system from an external system and then comparing the quality of the subsequent recording to that of the original clip.  The closer the recording is to the original, the higher the quality of your IP system.

 

Trevor Richer is marketing director, EMEA, of Empirix; tricher@empirix.com

 

A wise choice will take patience and hard work

 

TIMES have changed for call centres looking for voice recorders.  Gone are the days when the cheapest system which runs on a chosen technology was deemed good enough to do the job.

To choose wisely there’s a lot more to consider and a lot more to gain, especially if your call centre has 50-plus seats.  Now quality of service and business efficiency have to go hand in hand if you’re to achieve and maintain profitability.  What can voice recording do to help?

In my view, a lot. We just need to think of issues such as quality of customer-agent interaction, customer engagement and caller feedback to start appreciating the contribution it can make.

From a technical point of view, an open standard recorder allows you to integrate to any telephony infrastructure – including probably the one you already have.  It brings the cost right down, because there’s no need to buy proprietary hardware.

Recorders of this nature are much more “lightweight” for the owner, as they are easy to implement, simple and quick to deploy, and bring the added benefit of low cost overhead.

Beware though: lightweight does not mean less powerful.

Open standard recorders are usually rich in functionality, so the agents’ task is facilitated and enhanced by things such as fast call retrieval, easy marking and grouping of the calls as well as tagging recordings with relevant call data. The end result is a better service to the customer and more informative reports for the management.

That’s not all.  Open standard technology recorders link easily into complementary applications such as speech analytics, workforce management and call surveying.  To get the best out of such applications though, involves a thorough understanding of what the technology can truly deliver.

Take speech analytics: if you want your spend to bring a return, you need to invest a lot of time and effort to identify relevant data for analysis and to set the system up accordingly.  In order to do so you need to understand fully the degree of integration of data between your speech analytics solution and your voice recorder.

In this case patience and hard work pay results: well planned and pro-active data mining based on the right technology enables contact centres to understand key elements that make their business a success or a failure.

 

Paul Manyweathers is the MD of CyberTech UK; paul.manyweathers@cybertech-int.co.uk   

 

Six ways to make ‘why’ spell savings

WHILE you may already record calls, the sheer number can quickly exceed the ability to review and analyse them.  We have really started to see – particularly over the last year – the market responding to the benefits of speech-enabled analytics.

And this is reflected in the latest research. Ovum, for example, reports significantly more interest as companies start to understand the business case. It estimates that spending on speech analytics will increase at a compound rate of 13.5 per cent through to 2014 – and that’s despite hard pressed IT budgets.
Uncovering the “why” behind each call is important.  Speech analytics can be used to discover trends about what customers are saying about your offer, your competitor’s offer, your processes, in fact anything that you want to know about.
Our product comprises a series of proprietary speech mining algorithms that allows data to be refined to a greater degree. This allows you to categorise calls and provide a root-cause analysis for each category. You can compare subsets of calls against other calls, you can even start looking for frequency and correlation together – so that the word “cancel” in proximity to a competitor’s name could be flagged to provide immediate market feedback.
Here are my top six benefits of speech analytics:


Increasing first contact resolution  Discover the root cause of unresolved calls by identifying terms that appear more often than in calls that are dealt with first time
Reducing customer attrition  Identify calls which lead to account closure so that you can focus on techniques that provide customers with better reasons to stay
Identifying effective sales strategies  Analysing agent behaviour on the most successful calls highlights the most successful approaches, and examples of where agents have gone off script with better results
Spot hidden call drivers  You can make significant savings by examining common reasons for calls and working to eliminate the factors that cause unwanted calls

Quality monitoring Speech analytics can accelerate quality programmes by automatically segmenting calls and delivering them directly to the quality monitoring workflow

Managing vendors and partners more effectively Track calls, isolate and identify relevant customer feedback

 

Robert Wint is senior vice president marketing, EMEA, Verint Systems; robert.wint@verint.com