Voice of Vovici

New Year, New Job: Helping Employees Break that Particular Resolution

One New Year’s resolution that is perennially popular is the resolution to find a new job.  A significant number of employees evaluate their careers at the end of the calendar year.  As a result, expect to see many job fairs and more help-wanted listings (yep, even in a recession). 

If you are an employer or HR professional, you want to make sure that your best employees stay with you.  You want to help them seek those new career opportunities within your organization.  As I’ve written before, 47% of top-performing employees are actively looking for jobs, while only 18% of low performers are (Source: Leadership IQ).

All of which makes this a great time to conduct an employee loyalty study.  Because this is one New Year’s resolution you do not want your best employees to keep!

Apostle Model Best Practices and Survey Template

If a recession is a good time to look at employee loyalty, it’s an excellent time to look at customer loyalty.  Selling an existing customer those of your products and services that you aren’t currently selling them is the most cost-effective approach.  Doing it well requires a strong understanding of what drives your customers’ loyalty.

Segmenting the results of your customer satisfaction survey by the attitude and intended behavior of current customers can help you identify what separates those customers you are likely to lose from those that are loyal.  It’s a great foundation for a successful plan of attack.  The Apostle Model is a good approach for such work.

As I’ve written before, the Apostle Model segments your customers into four groups:

• Loyalist/Apostle - high loyalty, high satisfaction - "staying and supportive"
• Mercenary - low to medium loyalty, high satisfaction - "coming and going; low commitment"
• Defector/Terrorist - low to medium loyalty, low to medium satisfaction - "leaving or having left and unhappy"
• Hostage - high loyalty, low to medium satisfaction - "unable to switch; trapped"

These segments are graphically represented as:

Here are the best practices for using the Apostle Model in your own research.

• Use a 1-10 or 0-10 scale, defining 10 as best.  This scale is easily understood by respondents and provides you greater dispersion of results than a shorter scale.  This greater dispersion  is helpful when you are trying to contrast loyal customers against at-risk customers.
• Use 9 and 10 to indicate high loyalty and high satisfaction. You want to understand the outliers who are exceptionally satisfied and exceptionally loyal.  Jones and Sasser reported in their original Harvard Business Review article that, for Xerox, completely satisfied customers (rating of 5 on a 5-point scale) were six times more likely to repurchase over the next year and a half than somewhat satisfied customers (ratings of 3-4).  Resist the urge to broaden the top box to ratings of 6 through 10.
• For each dimension, use an index (two to three questions) rather than a single question. Normally I preach that fewer questions are better, since shorter questionnaires certainly are better from a respondent’s point of view.  That said, the advantage of using two to three questions is that by averaging them you have a more stable measure than a single question alone provides.  This reduces the volatility of your results.
• Conduct the analysis regularly, not just once. The most successful businesses thrive and grow by expanding their base of loyal customers. Don’t do this analysis as a one-off project, but approach it as an ongoing benchmark to track your organization’s performance.

Here are six questions that you could use in your Apostle Model:

1. Taking into account all of your experiences with [product], please rate your overall satisfaction with it. 
2. Think about all of your expectations for [product] before you purchased it. Please rate whether [product] falls short of your expectations, meets or exceeds your expectations.
3. Now consider your ideal [product category]. Please rate [product] on a 1 to 10 scale for how closely it comes to your ideal.
4. When you next purchase a [product category], how likely is that you will purchase from [vendor]?
5. Assuming you communicated your experiences with this [vendor] to others, how favorable would your comments be?
6. How likely is it that you would recommend [vendor] to a friend or colleague?

Use the first three for the Customer Satisfaction Loyalty index, and the bottom three for the Customer Loyalty Index.  These are simply suggestions. Depending on your industry and business model, other forms of these questions might be better for your organization.

Whatever the economy, those companies that focus on understanding customer loyalty will be the most successful, and the Apostle Model is a great method of achieving that understanding.

Correlation between Employee Loyalty & Customer Loyalty

Walker Information, Vovici’s partner for employee loyalty benchmarking, has done detailed annual studies looking at customer loyalty and employee loyalty.   When Walker compiles its U.S. benchmarks, it finds that both types of loyalty move in parallel to one another, tracking each other for increases and declines:

While correlation alone does not necessarily imply causation, Walker has found that high-risk employees provide poorer customer service than loyal employees: a sure way for employee disloyalty to lead to customer disloyalty.  As I wrote last week:

If you don’t take care of your employees, they won’t take care of your customers.  Loyal employees have a positive impact on customer loyalty and retention:  where 92% of loyal employees do tasks for customers “above and beyond the call of duty”, only 54% of trapped and high risk employees do so, according to Walker.  Where 89% of loyal employees help coworkers who have heavy workloads, only 60% of trapped or high-risk employees do.  In a recession, of course, it is more important than ever to keep existing customers loyal, because the cost of acquiring new customers is so high.

This is Reason #1 of the top five reasons to measure employee loyalty during a recession.  Make sure to review them all.

When Survey Incentives Run Amuck

Sometimes you have to offer incentives to guarantee a sufficient response rate to your survey.  Good times to use incentives include when you are surveying a group that your organization does not have a direct relationship with, when the survey is long, when it is on a dull subject or when similar past surveys have had a low response rate.

On such occasions, you will get the highest response rate if you offer each respondent a small gift, rather than offering them the chance at a larger prize.

Here are some best practices for when you offer an incentive for each completed survey:

• Make sure that you state this incentive is only good for the original recipient.
• Set the survey to be closed or private, so that each potential respondent receives a unique link by email.
• Do not let respondents save the survey to return to it later (sometimes called “save and resume”).  By setting up one-time use, subsequent clicks on the link will show a page indicating that the survey has already been completed.  Otherwise, if the recipient distributes the link, people clicking on it will see and overwrite the last respondent’s answers.
• Set a quota, thereby capping the number of incentives that you must provide.  Make sure to refer to the limit in the email invitation (“the first 300 respondents will receive…”).

What happens if you don’t follow these guidelines?  Well, one Vovici customer, Plow & Hearth, recently offered a $25 gift certificate to each respondent.  Recognizing that this was a larger than average incentive, a recipient published the survey invitation to FreeStuffTimes.com, one of many web sites that link to special offers and freebies.  The result was a sudden avalanche of responses.  Plow & Hearth quickly realized the situation and shut down the survey.

The company generously and graciously handled the situation, by sending all uninvited respondents the following email:

Thank you for your participation in our recent Lifestyle survey.

Unfortunately, the survey you responded to was only intended for a very limited, pre-selected audience of our customers whom we contacted via e-mail. Without our knowledge or consent, the survey was subsequently posted on several coupon sites by one of the original recipients. Once we became aware of the issue on Monday morning, the survey was disabled immediately.

We do however value your feedback and appreciate the time you spent completing the survey. To thank you for your unsolicited participation, we would like to offer you a $10 appreciation reward for use with your next Plow & Hearth purchase. Please reference the reward code below to redeem this offer….

So, when offering incentives, make sure to follow the guidelines above.  If you forget, and do have a deluge of responses, you can usually find the date the link was made public.  Contrast the results before that time with the later results, to see if the freebie seekers differ significantly from the original respondents.  If they do differ, discard all of the late results.

And, finally, to forestall any negative publicity, make sure to offer later respondents some reward, as Plow & Hearth so gracefully did.

How to Ask Respondents Their Age

Yesterday’s Pluggers cartoon made fun of the age-range demographic  question (please go read it; I don’t want to violate their copyright by embedding it here). 

You wouldn’t think this was a good idea for a comic strip, but it actually made the Top 10 Plugger comic strips for 2008.  Clearly it hit a chord with many readers.  When counseling new survey authors, I always advise them to avoid showing any bias, but showing an age range of “56 and over” reveals an interesting lack of perspective, grouping 70 million U.S. adults (almost a third of all American adults) into one demographic.

As I've written before, in general, you should avoid asking a respondent to select their age range.

• Ranges can offend the respondent. A 56-year old would not think that he had much in common with a 90-year old, yet he is grouped into the same category in the survey referenced by Pluggers.
• Ranges are arbitrary.  I could easily subdivide the “56 and over” into four segments (56-61 years old, 62-67 years old, 68-75 years old, 76+ years old), each with different attitudes and approaches to work and retirement, but you could probably come up with your own reasons for dividing it differently. 
• Ranges limit your analysis.  You can’t report the average age of respondents from such a question.  You can’t group the respondents into generational cohorts using a standard list of age ranges; for instance, Baby Boomers are part of the 36-45 age group, all of the 46-55 age group, and only some of the 56+ age group.

So what should you do?  Check out "Asking the Age Question in Mail and Online Surveys" (PDF) by Benjamin Healey and Philip Gendall of Massey University, who tested three different methods of asking the age question: 

1. Asking for the respondent’s date of birth
2. Asking for their current age in years
3. Asking for their age within a series of ranges. 

Here’s their conclusion:

The best advice for survey research practitioners is to ask respondents when they were born; either their date of birth or the year in which they were born. This format appears to work well in all survey modes, is parsimonious of questionnaire space, is easy to administer, and generally produces low non-response and high accuracy. The other piece of advice specifically for online survey researchers is to avoid drop-down response menus.

I would take this advice one step further and only ask the year of birth:

In what year were you born?  ____

If this was a web survey, I’d use a text box rather than a drop-down list and validate that the answer was between 1895 and this year.  (The world’s current oldest person, Tomoji Tanabe, was born in 1895.)

You lose a small degree of precision asking the year of birth over asking the actual birth date, but you remove a degree of risk about privacy.  Governor Palin’s email account was hacked by someone who used her birth date and zip code to gain access.  Marketers can use birth date, zip code and gender to uniquely identify individuals in many zip codes.  As a result, I expect more and more respondents to opt out of providing full birthdates, no matter what their age.

Sometimes, as in this case, the simplest way to ask the question will give you the best results.  And it will keep your survey from being lampooned in a comic strip!

Product-Customer Growth Matrix

In the recession of the early 1990s, when I worked for BIS Strategic Decisions, our CEO, Graham Cooper, led a strategy session in which he drew his own variant of the Ansoff Matrix.  Where the Ansoff Matrix contrasts markets and products, Cooper’s matrix contrasted customers and products:

The numbers indicate the relative level of effort required to acquire each type of revenue, from least effort to most effort:

1. Selling existing products to existing customers
2. Selling existing products to new customers
3. Selling new products to existing customers
4. Selling new products to new customers

A healthy company allocates its resources across all four quadrants.  For Cooper, our company’s efforts during the recession had to be invested in selling existing customers those existing products that they were not currently buying:  additional reports, retainer hours, and custom consulting projects.  This would provide the best short-term return on investment. 

For me, this “Cooper Matrix” provides a concise argument for selling more to existing customers during a recession.  Survey research can help with each quadrant, but for the first quadrant the focus is simply on customer loyalty:  what are customers’ current intentions to repurchase, and what steps can your organization take to improve that likelihood?