Hi

I would like to know how I can construct a scale to be used for benchmarking Customer satisfaction. E.g. I construct a scale from 1-10 and if the focal company perform 5 in satisfaction for the customers and the competitor is 8 in satisfaction for the customers , the difference should be 3??? But such a scale would be an ordinal scale right:

http://www.math.sfu.ca/~cschwarz/Stat-301/Handouts/node5.html

and so by it is not allowed to calculated the difference?


Any alternatives?

Hi

I would like to know how I can construct a scale to be used for benchmarking Customer satisfaction. E.g. I construct a scale from 1-10 and if the focal company perform 5 in satisfaction for the customers and the competitor is 8 in satisfaction for the customers , the difference should be 3??? But such a scale would be an ordinal scale right:

http://www.math.sfu.ca/~cschwarz/Stat-301/Handouts/node5.html (http://www.math.sfu.ca/%7Ecschwarz/Stat-301/Handouts/node5.html)

and so by it is not allowed to calculated the difference?


Any alternatives?

When it comes to this sort of survey, I'm always reminded of the seen in the movie This is Spinal Tap (http://elsmar.com/Forums/newreply.php?do=newreply&p=234402) where one of the band members shows the Rob Reiner character a guitar amplifier and tells him that the volume control goes up to 11, instead of just 10 as is usually the case. Reiner asks why they just didn't make the "10" setting louder, and the answer, after some hesitation, is "These go to 11."

http://elsmar.com/jpg/volume_dials.jpg

In order to be able to reliably measure anything, you have to control for everything that might bias the measurements, and in the case of something as nebulous as customer satisfaction you're facing an impossible task. Even measurement of dissatisfaction is fraught with peril because dissatisfied customers don't always complain, and there might be instances of customers who are generally satisfied but have employees who complain about everything anyway. IMO, the single best indicators of customer satisfaction are repeat orders and awarding of new business. Everything else is guessing, and it doesn't matter if one customer gives you an 11 and another gives you a 10.

The site you directed us to was quite scientific-looking. In my experience, customer satisfaction surveys are simple looking forms that are fast and easy to fill out. Constructing such a quick "temperature check" must be done with care, so you can get as valuable information as possible.

Here is a toolkit (http://books.google.com/books?id=raf7cbNR-ycC&pg=PA16&lpg=PA16&dq=customer+satisfaction+survey+construction&source=web&ots=aP0r8lSN4g&sig=Mam6yzOHrZ3FiTvykwe6nNNhQ2g#PPA16,M1) that includes a section about constructing surveys --starts on page 16.

Hi

I would like to know how I can construct a scale to be used for benchmarking Customer satisfaction. E.g. I construct a scale from 1-10 and if the focal company perform 5 in satisfaction for the customers and the competitor is 8 in satisfaction for the customers , the difference should be 3??? But such a scale would be an ordinal scale right:

http://www.math.sfu.ca/~cschwarz/Stat-301/Handouts/node5.html (http://www.math.sfu.ca/%7Ecschwarz/Stat-301/Handouts/node5.html)

and so by it is not allowed to calculated the difference?


Any alternatives?

Hello, there! Whether your scale is nominal or ordinal will depend on what values you use, and how you measure the variables.

Also, you stated number regarding customer satisfaction of your competitors. How are you going to measure that? How will you know which competitors are involved in the evaluation?

Here's a pretty good thread on measuring customer satisfaction.

Measuring customer satisfaction (http://elsmar.com/Forums/showthread.php?p=208676)

There are other threads that discuss surveys.

Craig Cochran has a post here on measuring service quality:

Measuring Service Quality (http://elsmar.com/Forums/showthread.php?t=25328)


Notice the effort and time spent validating the simple survey items. That is just the start of developing a good survey.

Please know-you're not alone in your frustration of developing measures of customer satisfaction:yes:. It is difficult to do, and I hope we're providing you some information to assist in your approach.

Thank you all for the interesting thoughts and links:), I did search on the subject but I was mostly focus on the Benchmark problem.

No doubt that "satisfaction" is problematic and in the end the best thing is probably to BE the Customer yourself.
And why even measure Satisfaction - meanwhile I found this link:

http://www.marketvaluesolutions.com/customer-satisfaction-article.htm

Lets say that I don't even do a survey, that I just by my own knowledge give my company(A) and the competitor (B) a number on a scale, and thereby I want to subtract A from B, which is nonsenses unless it is at least an interval scale, but in this case the scale will always be ordinal, right? So it seems kind of impossible to do that?

Brad, yes it depends on the values and input, and thereby the kind survey, but no matter what you make use of, is seems to me impossible to construct an interval scale for CS? And thereby it is (academic) impossible to do benchmark, unless you measures and compare with sales or such, but that's not what I need?

BTW: I have anybody heart about fuzzy logic, could it solve the problem?

Lets say that I don't even do a survey, that I just by my own knowledge give my company(A) and the competitor (B) a number on a scale, and thereby I want to subtract A from B, which is nonsenses unless it is at least an interval scale, but in this case the scale will always be ordinal, right? So it seems kind of impossible to do that?
This is essentially the 10/11 problem I alluded to earlier. Not only that, but assigning arbitrary values to something as nebulous as "satisfaction" is trying to nail Jell-O to the wall. The problem isn't the measurement method, it's trying to measure something without first rigorously defining the salient components of what you're trying to measure.

Brad, yes it depends on the values and input, and thereby the kind survey, but no matter what you make use of, is seems to me impossible to construct an interval scale for CS? And thereby it is (academic) impossible to do benchmark, unless you measures and compare with sales or such, but that's not what I need?
The fact that you're struggling with measurement means that you really don't know what you need, and that's because there are too many undefined variables and dubious measurement methods and targets.

BTW: I have anybody heart about fuzzy logic, could it solve the problem?

The problem isn't solvable. You can do your best to understand customer perceptions and try to turn them in your favor, but that's about it. Customer satisfaction, in the end, depends on episodic phenomena. It's a one-interaction-at-a-time thing. When a customer is in trouble and needs for you to bend your schedule to get them out of it, you're not likely to enhance their satisfaction by telling them they're SOL, and the schedule is the schedule. Nonetheless, it might not be economically responsible for you to bail them out, so you have to weigh each transaction in risk/reward fashion. You might find it necessary on occasion to "fire" a chronically dissatisfied customer, one whose expectations are patently unreasonable.

We need to remember Dr. Deming's admonitions regarding the unknown and unknowable, the fact that there are important things that can't be rationally measured, and that relentlessly trying to measure them is wasteful.

Jim's right, and his points were well made.

The problem with measuring customer satisfaction has two parts:

1) How to know what to ask
2) How to understand the answers

Start with the objective. "We want to know customer satisfaction." From there the question must be formed: "My customer: how happy are you?" But a numerical value has no value unless it can tell you what was done well and what needs to be improved. The questions would need to be framed in specific terms:

"How happy are you with our help desk personnel's help in solving your problems?"
"How fast was our delivery?"
"How easy was our instruction manual to use?"
"How easy were our product's features to use?"

The answers to the questions would be even better if there were some means for the customer to tell you why a given factor was less than great. How else can you know how to respond to to the findings?

Customer satisfaction can be assigned a number scale, formulae to derive an average with and compare against competitors, but satisfaction is a very personal thing. Your data should be designed to help you understand why your relative position is this and you competitor's is that, so you organization can develop a strategy to narrow the gap or come up with something unique to offer instead.

All of this doesn't need a scientifically designed survey with 95% confidence levels and so on, and I wonder what makes your own knowledge thorough and sensitive enough to be actionable. Some of the best data anywhere are gathered by customer service personnel as they are doing their work with the customer. Why wait for a blanket survey to find out how well it went?

Thank you all for the interesting thoughts and links:), I did search on the subject but I was mostly focus on the Benchmark problem.

To benchmark, I ask my Customer to rate Us vs our competitor. The Customer buys from more than 1 source (remember: don't put all eggs in 1 basket) so why not ask the Customer directly? I get actual detailed data and a rating number on a scale of 1 - 4.
Just my :2cents:

To benchmark, I ask my Customer to rate Us vs our competitor. The Customer buys from more than 1 source (remember: don't put all eggs in 1 basket) so why not ask the Customer directly? I get actual detailed data and a rating number on a scale of 1 - 4.
Just my :2cents:

I do the exact same thing... a quick phone call and 4 questions. All questions are directly related to each one of my 4 quality objectives. I just don´t rate 1-4. The question is always in comparison to my competition.

Example (simple) : Are our prices competitive with the competition?

They are all yes or no answers which makes "summing it all up" easier...

IMHO there was a post that determined satisfaction in terms of sales or repeat orders, to take this a step further would be to be the preferred supplier.

To measure this would be market position to give you a start position. Then looking at strategies to increase (or maintain) that position:

These strategies are product development (innovation), cost reduction (value), reliability (Quality). All areas that are within the QMS.

Each of these has measures:

Development - What are our competitors providing... do we need to aswell... can it be improved upon (or are we just a me too), QFD (Quality Function Deployment) helps with this (plenty of threads and info here and on the net. As a simplistic example providing an amp with volume of 11 give the customer something that other providers don't but does the customer really want it? Determining what the customer wants is not easy but when you get it right.... (e.g. ipods) or when you get it wrong... (e.g. Sinclair C5) makes huge differences.

Cost Reduction - Supplier costs, factory efficiency, waste reduction (rework, scrap etc etc), efficiency all things that a good QA Eng should be involved with and driving gives you opportunity to pass on to customer.

Reliability - Effective CA's, designing with reliability in mind, monitoring customer complaints, being vigilant of competitor issues and if there is potential to happen to you. Again all things that QA should be involved with.

This brief synopsis hopefully highlights that although "focus groups", questionaires, surveys, etc etc have their place it is equally important to internally measure with the customer in mind. The real key to success in this is management commitment, I have been lucky in that 2 of the companies I have worked for have this, but there are numorous others that did not.

As a quality "proffessional" (whatever that means, I like to think more in terms of problem solver) I always try and set quality objectives that focus on business objectives in addition to meeting reg's and std's... which when you break them down are really just business models.

My suggestion would be look at what you currently measure internally and and how valuable is that information in terms of being a preffered supplier.

:2cents:

By nature is does not make sense to transfer something subjective into objective - I am not trying to do that. That satisfaction is a personal thing, seems to be a question of definition... As to the 10/11 - isn't it just about whether you measure e.g. cm vs. inch., I mean even though it is problematic you still have a scale that you can make use of, and no matter what scale your neighbour is using, you can compare and find out how who can make the most noise. The scale is continues and can be related to DB.

Concerning the last comments, I will reflect a little more upon it and be back. I thought about defining the difference, that is making categories alike worse, same, better - but I am not sure that it will make the ghost disappear...

In the end I would like to relate it to the Kano-model (and yes to QFD/HOQ).

Thank you for sticking with us! Let me suggest again... I understand your situation!

As to the 10/11 - isn't it just about whether you measure e.g. cm vs. inch., I mean even though it is problematic you still have a scale that you can make use of,

The 10/11 reference is a really good one. Good humor has an element of serious reality in it. The band member was fixated on the numbers; that the scale went to 11 gave the illusion that there was something significant about it. When Rob Reiner's character suggested that it was not an issue of how high the readings went but rather how everything is set up/standardized, the band member quickly dismisses the notion, and goes back to 11. Like Ahab with Moby Dick, things can quickly become an obsession.

..I mean even though it is problematic you still have a scale that you can make use of, and no matter what scale your neighbour is using, you can compare and find out how who can make the most noise. The scale is continues and can be related to DB.



I would have to question this statement. It makes a profound difference what scale you are making comparisons to. If my scale has an average score of 4.2, how are you going to compare to that? You have to know how my questions are set up, weighted, etc. Otherwise, comparison is meaningless. Take for example:


Were you satisfied with the service?


Were you happy with the service?


One is on a 1-5 scale, and the other is yes/no. Having one-word difference will yield significant variations, simply due to validity issues. Too, having different scale make comparisons impossible (unless you know the scale values, and you can standardize them. That's some work).


Concerning the last comments, I will reflect a little more upon it and be back. I thought about defining the difference, that is making categories alike worse, same, better - but I am not sure that it will make the ghost disappear...

In the end I would like to relate it to the Kano-model (and yes to QFD/HOQ).

I do want to hear back from you. I really am trying to help:). Useful measurements are difficult to obtain, and take a lot of work. Hurrying into something might send flags up a flagpole that aren't really an issue. Or, you're not uncovering a real problem due to a poor measuring instrument.

I mean concerning the 10/11, we have a standard decibel scale that "sound" can be to related to, and it is continous, which I think is my problem concerning satisfaction. Indeed first as stated there is a problem with language you migth ask about satifaction and the customer think of it as "happyness" - whatever. You might do kansei analysis or use speciel interview technics and so forth, but I believe that whatever you do, it cannot be fully reliable - but is doing nothing at all a better solution...

Now What I had in mind is maybe not satifaction, name it reference or loyality - it does not make it easier. First I will found out where the customers are of course, lets say to make it simple, that they allways hang out in a specific bar. So I pretend to be a costumer like everybody else, and if a person starts talked (eventhough I have not asked) abou how fantistic a product he just bought, and that he would never buy anything else. Then I assumed the loyality is kind of "high".

Of course there is the problem about time, that he is more entuastic about it, if it is new. And like mentioned in the other threads the customers reaction and wheter they tell or not can be different relate to culture etc...

Anyway, I go to another bar, asking if they have tried this and that product (the competitors), and they tell me that it does not work at all and that they would never waste their money on that again (repurchase of course depend on the type of product).

Despite all the problems and you might give me som further arguments why the above mentioned approach is a bad idea (please go ahead), but I migth have an idea that my product is doing better than the competitors.

Now, eventhought I migth not do a survey, lets say I just subjective myself do some hypothetics on a worse, same, better scale, there is still a problem.
The scales are discrete right? And I need something continues. Somebody migth say just assume that? You can't, because as with the sound I need somekind of decibel scale to relate it to. I hope you can follow me so far.

Why do I need a continuos scale, because I want to link it to the Kano-model .Hence, the Kano is about linking "objective" measures with in my case lets call it loyalty. But Kano et al. only work with a binary "objective" scale, as they work with NPD product, I assume thats the reason.

If I have an objective, like for instance Weigth (lesser the better), the scale is continous and if I link it to my at first discrete loyalty scale, it is fair to assume the continues in between the "discrete" loyality scale?

Please let me you, what you think.

BTW: .

As a quality "proffessional" (whatever that means, I like to think more in terms of problem solver) I always try and set quality objectives that focus on business objectives in addition to meeting reg's and std's... which when you break them down are really just business models.

My suggestion would be look at what you currently measure internally and and how valuable is that information in terms of being a preffered supplier.

:2cents:

good point :yes:

First step, Kano and QFD works fine to integrate ;)

hey people!

Am a first time user of this thread. I am currently doing my summer internship as part of my MBA studies in one of the reputed copper manufacturing firm in India. I am working for the acid division.

Now my project is on customer satisfaction. For this purpose I have prepared a quesstionare to find out the customer satisfaction.I am attaching the quesstionaare. I am able to find out the customer satisfaction index after assigning weights to each of the factors in the questionnare.

Now my concern is that I want to benchmark my calculated CSI with the industry average for a chemical factory or an other manufacturing firm.

Can anyone tell me from where can i get the industry average for customer Satisfaction for a chemical factory.

regards
roy

Reichheld gives several industry examples in his book about Net Promoter Score "The Ultimate Question". I strongly suggest that you read this book or at least the original HBR article as they are both very current favourites on the MBA circuit. In addition there are numerous threads concerning C-sat in the Cove which may help you.

Good luck with your MBA

Jimmy

Hello all,

I got this forum while I was searching for (CSI) measurement and scaling, I like your way of analysing, well done.:thanks:

My question is: How do you correlate the expectations of a customer with the perception (either weighted or unweighted by certain factors)?