Surveys Done Right, Part 2 – Customer Satisfaction

I have been dreading writing this entry since I came up with the idea for the series (have you read part 1 yet?).  It is not that I don’t know what to say, or that I don’t want to do it.  It is simply that my fear of providing “sample” survey that will later become “real surveys” for all people without customization or personalization really, really takes hold in this arena.  I mean, who has not had to write a customer satisfaction survey in the past?  It is probably the most used, misunderstood, and poorly implemented of all surveys out there.

If we go by the surveys that I have seen, customer satisfaction surveys should only have one question since that is what most people care about anyways: “overall, how satisfied are you with us?”.  It seems that if the customer says they are overall satisfied then there are no problems.  Further, as long as we get a sufficient number of customers to say they are satisfied (somewhere in the mid-70s to mid-80s), and the number remains consistent or (gasp) even grows a little  we are perfectly set from the customer experience, loyalty, and satisfaction point of view.  After all, if 80% of people are satisfied – isn’t that good?  Well, yes and no.  I wrote about how you can ensure you will get a 90% of more in customer satisfaction surveys in the past (second most popular post in this site).  It is possible to score high – even increase your scores, and not be doing a good job.

So, before we jump into the description of the survey and the sample questions (and before I beg you again to personalize and customize them), let’s make sure we understand one thing about this survey.  This “sample” survey is not intended to tell you how many people like you.  It is intended to, historically, provide a trending line of your customer satisfaction overall – but more important it is intended to give you real answers to partial satisfaction questions (e.g. were satisfied with the speed of the answer?) in accordance to your strategy (you do have one of those right?).  Finally, consider that I used these questions for customers that have customer satisfaction as a key metric and part of the insight.  I still believe thatcustomer satisfaction is too flawed as a metric to be used reliably (also read this one), but a trending report cannot hurt – as it wouldn’t with any other metric.

Too many words to get here, so here are the the essentials of a customer satisfaction survey (part 3 of this series, best practices for writing surveys – stay tuned!):

  • First, the most important part – KISS (keep it short and simple).  I use the rule of 5+/-2 (five, plus or minus two), with five being the magic number, and three and seven the limits.
  • Second, you MUST personalize the questions to the specific items you want to measure satisfaction on – which is why there are some questions below to use, but they may not fit what you are looking for.  Tie your questions to your strategy
  • Third, send them out every business cycle (varies by industry and function), and make sure you use panel management tools so you won’t keep sending them to the same person over and over (we know what they think – no one should get more than one a year), and that you don’t send them to people who won’t respond.
  • Fourth, statistical significance is great for market research or election day polls, not so much on customer satisfaction.  Make sure you get a large enough number to make it varied (I suggest at least between 3-4% of the population) and I am talking responses – not invitations.
  • Fifth, make sure you use the same questions the next time you send them out (yes, exactly the same wording) so you can keep historical trends.  Remember, customer satisfaction is an OK metric to keep as long as you use it as a historical trending metric, not a KPI

Those are the basic rules for these surveys, but more will arise as you begin to implement them.  Of course, best practices for EFM and Surveys apply, so those are to be considered as well.  Now, without further ado, the list of questions I’d like to propose for your consideration:

1. Overall, what is your satisfaction with Company XYZ? (hate this question, but it is an easy one to answer)

Now, this question is to be answered in a numerical scale, I favor a scale of 1-4, you can use whatever works for your objectives.  A scale of 1-10 is popular with people who endorse NPS (Net Promoter Score), but I don’t buy into that.  A different way to ask this questions, which I prefer, is:

1. Overall, would you say you are satisfied with Company XYZ?

This question is answered with a simple YES-NO.  Why do I like this question? Two things, first the wording predisposes the respondent to say yes (which is a positive score).  Second, you can setup your scripts to react in the case of a negative answer – and since the question predisposes the respondent to say yes, when someone says NO they really mean it.

From the following questions, pick some of the ones that reflect your objectives, vision, strategy, and purpose for this survey.  Remember, personalization and customization are highly encouraged – so if these don’t fit your needs go ahead and change them accordingly.

2.1.  What is your satisfaction with Company XYZ web site / self-service solution?

2.2. What is your satisfaction with Company XYZ representatives on the phone?

2.3. What is your satisfaction with Product ABC from Company XYZ?

2.4. What is your satisfaction with Company XYZ return policies?

2.5. What is your satisfaction with Company XYZ use of email for customer service?

Well, you get the idea.  Keep in mind that how you ask the question is going to depend on the answer, and see Q1 above to see some examples of different ways to write the questions.  Needless to say, again, the questions you chose for the second part are related to your specific, strategic goals and are bound to change for your survey.

This is the longest post I wrote since we started this blog, but I wanted to make sure it was clear and explicit.  There is, of course, lots more to read and write about customer feedback, customer service, customer satisfaction… well, anything dealing with customers.  You will read more and we move along – but how do you like this post?  Interesting? Are you using it?  Please let me know your thoughts and experiences… and thanks for reading!

Surveys Done Right, Part 1 – Point of Delivery Surveys

Surveys Done Right – Part 1 – Point-of-Service Surveys

I am going to break some very old rules of mine to write this post.  Ever since I introduced the three-layer model for surveys while at Gartner (point-of-service, customer-satisfaction, planning) I have been getting requests for “sample questions”.  I have maintained, and continue to do, that I cannot provide sample questions since all questions need to be created according to the situation, respondent base, strategy and vision for your feedback initiative, as well as the standards and rules you set for your surveys.  Of course, they also have to be personalized to respondent and situation, and be written to match delivery and collection channel.  This is as basic as it gets when writing surveys.  My concern / problem is that when someone gets “sample questions” they become “THE questions” without further tinkering, and that is just wrong.

So, the counterpoint to that is that I have seen the concept implemented (point-of-service surveys) with some truly horrendous ideas.  I have experienced “short” surveys of 10 questions asking all sorts of things, and questions so badly written that it is almost impossible to answer.  Thus, as a public service (yes, I know I am a selfless philanthropist when it comes to surveys) I am going to break the rule and make this post about two things: a reminder of how point-of-service surveys should be done, and a set of sample questions (which I will regret for a long time, and possibly my grand-kids will as well).

First, how does this work.  Point-of-service (also called point-of-delivery) surveys are SHORT (yes, needs to be shouted), 2-3 questions surveys aimed to discover the efficacy (not the efficiency) of the service interaction.  In other words, did we do a good job delivering service and was it what you needed.  It is intended to spot any problems during delivery, and to fix them before they become customer service issues or lead to customers not being satisfied.  Simple, huh?

Now, the main point of doing this is preventing service issues from becoming problems.  Thus, the critical part is not doing the survey, but actually having processes in place to reach out to customers and fix their problems when either of these questions returns a negative answer.  This is where most companies fail, they don’t have documented, specific processes in place to take care of negative answers quickly (yes, speed matters).  The reason I am bringing this up, even if you copy the questions from the bottom of this post – please, please, please make sure you have the necessary processes in place before doing these surveys.

Final point I want to make, then we move to the actual questions.  Channel of delivery matters.  If at all possible, try to keep the survey in the same channel where the transaction took place, and to follow the interaction immediately.  If the customer called, make the survey an IVR-driven survey post-call (no, don’t have another call for follow up… it does not work that way).  Email came in? email going out (as quickly as possible, not 2-3 days later).  If you cannot maintain the channel of service be the channel of delivery (or you cannot make it immediately following the transaction) then your best bet is using email surveys.  No, not email with links to online surveys — email surveys.  The questions are within the email and they can answer simply and quickly.  OK, getting off the soap box now (yes, I am passionate about this “stuff” being done well).

CAVEAT: I know I said this before, but please, please, pleaseeeeee customize these for your situation and personalize them. Please?

Question 1 (this one should never change): Did you receive the answer you needed?

Question 2 (choose from the three below based on what else you need to measure):

Q2.1: Did we do a good job delivering the answer? (my favorite, but a little broad in meaning)

Q2.2: Was our service cordial and polite? (in other words, who needs some training or talking to)

Q2.3:Was our representative knowledgeable? (again, training or knowledge management issues)

Q3.3: Was our representative prompt to answer your questions? (do they know what they need to know?)

You get the idea, depending on what part of the interaction is critical you can change the second question.

So, please don’t let me down. Customize, personalize and (more important) let me know how it goes…

Solving the WHYDFMT Problem in Customer Service

Come on, say it out loud three times real fast.  Solving the WHYDFMT problem in Customer Service, solving the… forget it.  Quite a mouthful, ain’t it?  Alas, this is the quintessential problem of customer service and the one you need to tackle.  Forget channels, and feedback, and measurement and all that stuff.  That is secondary.  If you want to build a successful customer service and support strategy this is where you are going to focus: WHYDFMT.

Which, when spelled out, means What Have You Done For Me Today. This is the attitude of customers towards customer service – and the reason you must do things differently.  Let me explain.

I have written countless times that Loyalty does not exist (come to think of it, it is the most popular post in this site – if you have not read it yet, take some time to do so – I’ll wait right here).  It should not be measured, sought, or intended.  I have not been listened by lots of people, but lots of head-nodding ensue.  WHYDFMT is why loyalty is not a goal for your organization.  Customers, when dealing with customer service, take the attitude that each instance is a fresh, brand-new, out-of-the-box time for your to prove your love for them.  Each time they approach customer service they want something and they don’t care what you did for them last time – actually, that reinforces the feeling of entitlement (I promise, I am not bitter or resentful – just noticing things).

You gave them a credit last time they were late with their payment, well – do it again.  You refunded a fee they did not understand (although they signed up for it when they contracted the service), well – do it again.  This creates a problem within the organization where managers don’t want to continue the trend and CSRs don’t want to deal with customers.  It creates a big problem for management when customers are willing to criticize them for the performance today and forget the performance yesterday.  It makes the company look like it provides bad customer service based on, maybe, one bad instance.  That is bad for your brand and you have to change it.

So, how do you solve this problem?  Let me point you to the solution in an example: Zappos.  That’s right, the telephone / online shoe sales organization. By design, every single instance when a customer contacts the organization will be a delightful experience.  Every single time.  It is part of the corporate culture, it is part of their brand – some even would say part of their mysticism.  This is who they are, not what they do.  This is their persona, their representation to their world, their promise to their customers and to themselves.  That is why it works.

So, as usual towards the end of the post, what shall you do to partake on this?  First, recognize Loyalty is not a worthy pursuit (yes, again) and focus instead of delivering excellent experiences.  Each time and every time.  Easiest way to do that?

1. Conduct Feedback Events – use surveys, or any other tool, to ask customers what problems you have in your experiences today.

2. Implement an Experience Initiative – Zappos did not get to excellent experiences by wishing them, they planned for them.

3. Continuous Improvement – over time, even the most excellent experiences tend to become dated. Update them.

In other words, to solve the WHYDFMT problem – avoid the question.  Make it very clear, explicit, and documented (have you seen the feedback that Zappos gets from customers – almost fanatical) that you are about your experiences and customers won’t ask that question anymore… and they will become loyal (sorry, parting pun – could not resist).

What do you think?  Is this something that may work for you?  Let me know your thoughts and experiences — and as always, email me if you want more details or have any questions (ekolsky at evergance dot com).

Forget Parallel Computing…The Money is in Parallel Servicing

I know this will be hard to believe – but I am a nerd, a geek, a lover of all things technology.  Why, in the old days when Windows 3.x was just beginning I made a few extra dollars building computers from spare part for fellow students (hey, it was way cheaper and popular back then), not to mention that I am certified for Novell and Microsoft networks (yes, I was a sysadmin at some point in life).  I love technology and I use my knowledge as much as possible.

As I was cleaning one of the hard-drives in my computer last week I came across a paper I had written at some point in life contrasting the use of parallel and serial ports in computers.  Now, this is not a big deal these days with USB and Firewire and the myriad other connection options – but back then it was the only way to connect.  In this paper I looked into the technological considerations, including speed, of using one or the other.

Here is the summary you need to know for the purposes of this post.  Parallel ports work, as their name implies, by sending data at the same time through all possible pins or channels with no interruption.  This translates into a top speed of around 4 Mbps.  Serial, on the other hand, work by sending data bits one at the time and waiting for confirmation before continuing.  This reaches a top speed of 115 Kbps.  The difference is somewhere around 12 times faster for parallel ports once you count in things like hexadecimal conversion, acknowledgments and enhanced ports and accelerators.

Why am I bringing this up?  I am not trying to date or label myself, I am using this as the starting point for a new model for customer service – parallel servicing.  All the customer service we do today is in serial form.  A customer contacts us, and they wait.  We take one action, they wait.  We solve their problem, we think, and then we wait to make sure it was solved, or we wait for them to come back.  Not only that, but we cannot serve multiple customers at once unless we deploy more “ports” or resources.

The concept of parallel servicing is not having to wait, to be able to service customers at a higher speed, with certainty of delivery and better results.  How would you like to be able to service more customers with just your current resources but faster?  Three steps to get you there:

1. Proactively Service Customers.  I wrote about this many times, talked about it for a long time – and no one listened (it’s OK, I am used to it).  You have the tools, technologies, and know-how to implement proactive service today – with no major investment.  As simple as writing a few triggers for your database, then letting them run their course.  As you get more comfortable you can get more courageous — the sky is truly the limit.  A new patch released? Send an email to people who own the product you are patching.  Complementary product created for your best-selling product? Up-sell.  You can even charge for it in a subscription service!  So many ways to do this… so easy to do.

2. Change your Processes.  Yes, I am telling you to change your processes. No, I don’t know how (but would be glad to do the work for you if you want, just call me).  The secret here is to find serial processes that can be converted to parallel processes.  Why would a customer have to wait for one action to take place before a second starts?  Look, you know your processes, you know where you can find actions or functions to improve.  Once you get used to the idea of parallel servicing you will notice more and more areas where it makes sense.

3. Automate. Nothing makes processes go faster than automation – and you can run multiple instances at the same time (sorta like a parallel communication bus).  Look for areas where you can automate to improve the speed of your processes.  In lots of cases, you can turn long processes into sub-second actions on the computer – virtually parallel servicing if there is no delay!  Think about where automation makes sense, and do it.

Final words: you don’t have to do all this at once.  I am a firm believer in evolutionary changes, not revolutionary ones.  Go slow, little by little, and notice the results.  Get more courageous as you succeed.  Do more and see the speed of your service processes improve, then end up with a great example of parallel servicing.

What do you think? Interesting enough to try?  Leave me a comment, let me know how it goes…