Call Types – Root Cause Analysis

Call Types – Root Cause Analysis

I operate a small call centre for a Central Heating Firm. We have receive a wide range of calls on a daily basis, but at present have only 2 options for the customers to use and 2 VDNs to report on.

At present we receive a number of unnecessary calls that could have been avoided, but before I start doing Outbound Calling, and directing calls to other departments I need to do some root cause analysis to determine the types of call we get, the volumes of each type and the times of day we get them.

As I see it, I have 2 options:

Introduce new IVR with multiple VDNs, or
Task the Call Handlers to record the call type and report back daily.

Both processes are reliant on human input, and therefore not robust. We do not have a facility to record the call type on the system, and in any case, the system we use is not linked to the switch, so the data itself would not be particularly accurate, nor easy to report on.

Does anyone have any better ideas as to how I can record this info? I really want to avoid getting the call-handlers to record the info as this will cause greater work and the chances of it being accurate are highly unlikely….

Question asked by Ross

Not Ideal But……..

Ross.

By no means an ideal solution but…

Have you the ability to log
‘Wrap’ codes at the end of calls?

Needs careful staff management as a tendancy to
(by no means deliberate) wrap to a bucket code must
be watched for.

DaveA

With thanks to Dave

No Perfect Answer

Hello Ross,

I don’t know if there is a perfect answer to you, I’ve always preferred putting the option into the customers hand rather than the agents, after all it’s their perspective that really drives calls.

Yet, no one likes a clunky IVR. I suggest if you do decide to expand your IVR make sure to check for abandoned calls within it and see if there is an increase there.

How many root causes do you think your calls are driven by?

There is nothing to say you can’t do every suggestion here though, the wrap codes that Dave mentioned are good, but once you get too complicated (too many codes) or too general (having a “general” code) you may find that agents start selecting the same one’s over and over.

With thanks to Justin

Call Categorisation

One of the problems with call categorisation (which you need before you can do the root casue analysis) is that many calls are about more than one thing. Any manual system will rely on the human to categorise the call in detail (if they can be bothered) so any data obtained would give a general overview only.

An alternative is to use speech analytics to do a detailed breakdown of the actual calls themselves. Speech analytics works against call recordings and will deliver data not only on the context of the call but will provide information on call lengths by individual call types. There are several options now on the market, using different speech technologies, which can deliver very accurate and detailed results. Whilst it sounds very complicated and expensive, some companies will offer this as a service (rather than having to buy the kit yourself) and will also do the call capture if you do not use call recording.

Dave

With thanks to Dave

Call Analytics

Dave,

Possibly overkill for a one off data gather?

On the other hand, I can’t think of a ‘Mid range’ solution.

Seems to be Hi-tec (Analytics) or Pen and Paper / Wrap codes.

Anyone else got anything?

With thanks to Dave

Replies

Thanks all for your speedy replies. I think a combination of IVR (it needs overhauling in any case) and manual recording by the agents will have to be done. We have trialled Work Codes on the turrets with minimal sucess, but I think a call log sheet detailsing every call type wil have to be used.

Not ideal, but necessary…..

With thanks to Ross

Manual Log Sheet

Hi Ross,

Pls be honest to yourself & do not expect your employees to bother to give you the accurate data. Dave hit the nail on the head by mentioning “tendancy to (by no means deliberate) wrap to a bucket code must be watched for”

We face a similar issue …. calls are routed on basis of IVR selected by pax (this is one source of call type stats for us) & our agents also punch in codes at end of each call (which is our second source of call type stats) but i can vouch that only 70% of it is correctly selected by callers as well as agents.

Manual log sheet is similar to the transaction codes punched in by our agents but a bit more manual & hence more ‘effort’ …. coupled with a wiifm attitude & you will be having data which is pure fiction.

My suggestion is
* go ahead with the manual sheet but keep it simple (not ‘detailing every call type’ as you mentioned)
* educate the agents on the importance of getting this information
* involve the agents in the selection of top 6-7 diff call types to be used

Good Luck!

Kex

With thanks to King

King.

Quote from King:

* educate the agents on the importance of getting this information

I’d like to stress this point that King mentioned, it’s likely to be seen as needless, useless busy work by your agents which would definetly skew the data.

Sitting down with them and explaining why you need this data and how long you are asking them do to this is crucial. You need their buy-in that this is the best method to get the data.

King: Interesting to see the statistic 70% accuracy between IVR and agents, makes me wonder what percentage of that 30% is button pushers vs agents.

With thanks to Justin

Analytics

Possibly overkill for a one off data gather?

DaveA – you would think so, but I have seen this applied in very small call centre situations with amazing results and relatively low cost. It’s certainly worth looking in to at some stage, especially if you want to get into real root cause analysis to reduce your inbound call volumes and improve your agent and customer sat issues by identifying some of the internal product and process issues that you may not be aware of.

However, no matter how you collect the data, the key is what you do about it and often this is where the call centre suffers. Often the agents are seen as a “necessary evil” to deal with customers, not the eyes and ears (and mouth) of the company and there is usually no mechanism to feedback customer intelligence to the rest of the organisation. The situation is changing but slowly….

Sorry to sound pessimistic but if as much attention was paid to WHY customers are calling as there is on dealing with the call as quickly as possible, many of the call could be avoided in the first place. (Sorry, this is a personal pet complaint of mine!)

With thanks to Dave

Call Types – Root Cause Analysis

Ross:

I really enjoyed your post and the replies because it gave me a great new product enhancement idea. First I’ll paraphrase the problem…”Ross does not think his Call Handler Scripts will accurately categorize the call. Asking the agents to type in one wrap code after the call is limiting because more than one call type may be involved. There has to be an easier way.

Proposed solution: Write a small web app that provides a list of call codes as push buttons. Each agent logs into the app and pushes the buttons as the caller touches on various call drivers. Each time a button is pushed, the Web writes a record to the database including the agent name , time of day and call driver . A simple database join between the call driver log and your ACD data easily categorizes each call with multiple call drivers. If you want to get into detailed analytics, you can even do path analysis to find the most common sequences that callers run through when callers raise multiple issues during a call.

Its a good enough idea that I think I’ll have our developers build this into the agent punch clock component of our software. We already have the web based business intelligence so its “easy as pie”.

As Quid Pro Quo for the good idea, we’ll set you up for free if you think the solution would address your needs.

With thanks to Paul

Proposed Solution

Paul,

I realise this post has been on a while however I have just joined the forum with a very similar problem & would be interested to know if your proposed solution was effective ??

With thanks to Simon

Your Question is A Theme, Across Industries and Contact Centers

In looking at your question:

1. In introducing a new IVR that gives you more VDNs (how many agents?), you assume that your VDNs will give you a view of your customer needs. You need to measure whether customer activities with the IVR actually reflects their needs. Many folks don’t choose IVR menus rationally, or the IVR solution doesn’t let them.

2. Re: speech analytics – you may learn more from analyzing the transactions by linking to your CRM system – what transactions were done, looking at your call history – or by analyzing call detail history (agent talk time, skills, holds, transfers etc,) and linking this with select agent reason code data entry.

3. I have implemented about 20 speech analytics solutions – they don’t give you good data except for select info you can’t get any other way.

4. Presenting the data via good dashboards and reports may also be critical. Too many reason code views are horrible extracts into Excel that present very difficult data. I would be glad to share with you some more effective views if this is of interest.

With thanks to Greg

Listen to Calls

In the original question it was assumed that there were two options:

1) Introduce new IVR with multiple VDNs, or
2) Task the Call Handlers to record the call type and report back daily.

There is a third and that is having people sitting and listening to calls. Of course this is manually intensive but it is robust and has the advantage that they are only listening to calls and not taking them (as agents are). It can take a few days to get the data you need and the time to stop is when you can predict the types of calls coming in.

With thanks to Rob

Talk to the Agents

Of course, we are all possibly missing the obvious by looking for technology (hi or lo tech) solutions. One superb source of information is the agents themselves – they know what frustrates both them and the customers. I have been involved in a nuber of successful agent surveys that have given good insight into root causes of the reasons for customers calling. They will be able to highlight internal process issues, problems with the call centre technology they are using, and your products and services because they are dealing with the customers every day. They often have the answer, it’s just that no-one asks them and they often dont feel empower to speak up.

Greg – sorry to hear that you have had bad experiences with speech analytics (or should your “don’t” be “do”?). My experiences so far is that a good speech analytics engine can deliver highly detailed insights. On it’s own, it’s just a tool but combined with a good business analyst to interpret the results and the management willingness to fix the issues, it’s a very powerful one. Using agent selected reason codes can work, but they are open to interpretation and, if in doubt, the agent will often use the “other” category or the first one on the list because it’s quick and they can get on with the next call. Speech analytics should capture the customers intent from what they say rather than what the agent thinks they want.

With thanks to Dave

Definitely Talk to the Agents, But After You Have Data.

I would do this. Talk to the agents and tell them you are going to listen to the calls, not the agents. Thus they don’t think they are under review. Then listen to the calls and get good data. Then do what Dave said, ask the agents what they think the causes of the data you have gathered. Like the man says, they will have excellent insights into the true causes. It is just that I would not ask them without the data, since then it is just chat and impressions. You need good data, however it is gathered, to ground the conversation, otherwise you get into a lot of conjecture and opinion and you have to go and verify it anyway.

With thanks to Rob

Simple Solution

Hi,

While was going through all the suggestion , I really got so many options .
I had same problem in my previous organization .

I agree with Dave “Needs careful staff management as a tendency to
(by no means deliberate) wrap to a bucket code must
be watched for.”

To make sure that wrap up is done in every call , we had introduced wrap up% in Agents KRA with good weightage and also rigorously in pre-shift and post shift briefing we spread awareness and Every wraup up codes were pasted in every workstations .

And we counted as fatal error if anyone misses wraup which means that it effects wraup up % and Quality Score.

Which Helps agents to get good incentives as well . And it was entirely Team Leads responsibility to get it done .

It was simple !!

And it really worked !!

With thanks to Suraj

Author: Jonty Pearce