Bag of Chips (part 3 of 3)


I have nothing to say today (The visit)

The first thing to do in order to better understand your customer is to stop talking.  I find this very hard to do.  I love to talk (and talk and talk).  While in high school and traveling home for 5 hours in a bus from a church group sitting by a girl who, I was convinced, was the most amazing creature alive, we had a lively discussion and I loved it.  She agreed to go out with me at the end.  Why?  Because she had never been with someone who could talk non-stop for 3 hours.  Did I say discussion?  It wasn’t near as charming on our first, and only, date.  Listening is a tough skill, it comes easy to some.  Unfortunately, the type of person who is good at going to a stranger’s office and asking them questions is typically the type of outgoing individual who must work at listening.

It is a subtle art to ask a question and then only talk occasionally purely for the benefit of getting your customer to talk more.  Ask the question.  Say nothing. Occasionally ask a question to clarify (why do you like having an access hole in the bottom of the machine?). During an interview, every thing you say should be a question! Every single sentence!

It is easy to fall into two traps.  In the first, the customer asks a simple question and you dutifully answer it (they are the customer, right?).  Problem is, you invariably start selling and influencing the customer.

“What do you mean by square headlamps?”

“Well, we have a unique, patented, high-output square headlamp that casts 2-3 times as much light in front of the XYX Roadstar than our nearest competitor.  So, do you want square headlamps?”

The second trap is just selling.  Usually it is important to keep the salesperson quiet during the questioning period.  They can’t help but sell, they might as well stop breathing.  Engineers are sellers as well, usually attempting to sell the brilliant idea that is not nearly described well-enough in this survey.  So, let me tell you about the new full foldable day pillow we are currently testing and how much it will do for you compared to your current pillow.

If you do have a new concept or configuration you need to elaborate on, do it in a standard way. We have found that a one-page, heavily visual handout is most effective at this.  In all fairness, visually show the alternative in an objective way and avoid any sort of marketing names or hype.  This is the kind of thing you test in the prototype survey.

Ask a question. Listen.  Use follow-on questions to get more input.  You should hear your voice about 10% of the time.  It should bother you how quiet you are.  Listen.


I hear Voices (Putting the Data Together)

You now have 40 to 100 surveys filled out, packaged together and ready to go.  What next?  You need a system to analyze this data and a context to present it.  The most important aspect of this is the data must be collated by someone who is objective and can put their emotions on the subject aside.  Let’s look at the steps.

  1. The first step is to put the objective data together. Analyze it like any other survey.   For each question, tell how many customers answered, and what their answers were.
  2. Next, schedule a meeting for everyone involved in the survey. Send them the draft data you put together to review.  Assemble them around a big table.  The goal is going to be to discuss the data, who it came from, and what it means.  Attempt to categorize each conclusion into the following categories:
  • Clear message:  A strong majority of customers said this.
  • Synthesized messages:  Using the input, you put 2 and 2 together to make 4.  No one actually said they wanted a car that went 70 mph but several said they wanted a ‘fast’ car and two said fast, to them, meant over 60.
  • Other messages: A customer who always seems to have a lot of insight said the car should go 71 mph.  This made sense to the group who thought most customers would probably agree if the question was asked that way.
  1. Finally, take this information and create a one-page summary of the conclusions arranged in clear, synthesized, and other message.
  2. OK, there is a fourth, send it around for review and comment. 


I need more data NOW! (How long should this take)

“Could you guys survey the North American market for earth moving equipment, we need to know what products are needed, by whom, and what specifications.  We need pricing information as well.  We have to have the data in 2-3 weeks or it is of no use to us.” How many times do you hear a variation of exactly that question?

This particular section is pretty simple and is going to be very short.  Here are the three points:

  • We have found we need to send 6-10 teams out, each visiting 8-12 customers, then collate the data to have a reasonable view of our market. This takes 3-4 months to answer one set of questions.
  • Our customers’ availability is what drives the schedule. We have found that we do not get objective data if we attempt to rush the survey and force our constraints on their time.
  • It is possible to do a limited survey in 3-4 weeks. It is phone based and usually involves talking to 15-25 customers. Questions need to be much more limited. You do not get the ancillary information and perspective that comes from visiting your customer, a 75% loss.

It takes time to do it right.  There are few short cuts that do not significantly affect the data quality. Somebody was going to ask the question, there it is, deal with the answer.  Ignore at your own peril, rushed surveys produce bad data and cranky customers. I hate both of those.


Ten Ways to Hear Exactly What You Want To

If you are not careful while surveying your customers, you will hear exactly what you want to hear. This is pleasant, painless, and, eventually, deadly.  I have made all of the following mistakes and each has cost significantly in data, dollars, and time.  What kills me, is I have made several of these mistakes more than once.


Ten Ways to Kill a Really Good Customer Survey:

  1. Talk a lot or ‘sell’ during the survey
  2. Ask too many questions
  3. Assign a person to lead the survey who has a bias or emotional history with the subject
  4. Attempt to make subjective questions objective (“rate our product from 1-10”)
  5. Don’t perform a prototype survey and update the questions
  6. Let one person perform all of the visits and ask all of the questions
  7. Visit only your best and most loyal customers
  8. Give each survey one vote with complete disregard to customer background and volume
  9. Assume you already have enough data and cut the research short
  10. Ignore the results or, even more commonly, figure out why the survey was wrong and your pre-conceived notions were right on target.


A Way to Get Exactly What You Need

So there it is, an idea, a way to do surveys.  The only way?  Of course not. The best way?  Probably not.  Take from this what you can and make it your own.  Get out to see your customers and listen.  Then listen some more.  Then listen. Then listen again.  And finally, thank them for their business and get out of their hair.  They have just done you a huge favor.