2010 TreasuryDirect Customer Satisfaction Survey

2010 TreasuryDirect Customer Satisfaction Survey

February 8, 2010

Background

The Bureau of the Public Debt plays an integral role in financing the operating expenses of the federal government. We sell U.S. Treasury securities to the public to help fund our country’s expenses and to pay off maturing debt. Launched in 2001, the web-based TreasuryDirect system allows customers to purchase and manage U.S. Treasury securities directly from Treasury.

To provide high quality and responsive customer service and to meet our obligations under Executive Order 12862, we need to know what our customers’ needs are and what new system features they desire. Surveying our customers gives us the opportunity to know our customers better and to get valuable feedback on the quality of our services.

Getting in touch now is important because of the rapid advances in business technology. Today’s investors are more electronically savvy than their predecessors. They can tell us whether the newest services we’ve provided in TreasuryDirect meet their investing needs, what investment features are most important to them, and how we can continue to improve and gain new investors. Again, the bottom line is customer satisfaction. Have we achieved it…and are we positioned to continue achieving it?

This is the second comprehensive survey of our TreasuryDirect customers. We’ve recently contracted the services of a research company experienced in survey methodologies. Working with the vendor, we will:

• Ensure the survey is conducted with statistical accuracy;

• Ensure the data collected will be accurately analyzed;

• Ensure the data we benchmark will reliably support comparisons we may want to make when conducting future online surveys; and

• Learn and build our own in-house knowledge for online data sampling.

Survey Goals

The survey will gather basic demographics, general information about our customers’ investment experiences with Treasury securities, and a few key facts about their Internet usage. Most importantly, it will measure our customers’ satisfaction with TreasuryDirect. Through survey responses, we will:

• Measure (quantitative and qualitative) customer satisfaction with our TreasuryDirect services and with our customer service representatives.

• Gather attitudes and opinions about:

  • the information and transactions available through the TreasuryDirect web application, and

  • other financial services sites frequented by TreasuryDirect customers.

• Better understand why some customers have inactive TreasuryDirect accounts; i.e., why they have had no purchases in the past two years or ever.

• Gauge customer interest in new or prospective features and services that are designed to enhance the TreasuryDirect investment process.

• Better understand the financial investment behaviors and practices of our customers.

• Support our goal to have the best business practices.

• Fulfill Public Debt’s commitment under Executive Order 12862 to regularly survey our customers.

Research Methodology

Population

The TreasuryDirect system has just over 619,800 accounts from which we’ll draw our random sample.

Survey Sample

The survey methodology will be a stratified random sample of TreasuryDirect registrants who have logged into their accounts at least once. We’ll divide our population into four subgroups, based on whether customers have made TreasuryDirect purchases and accessed their accounts in the past two years. Group 1 consists of customers who have made purchases within the past two years; 99 percent of these customers have logged in to their accounts in the past two years. Groups 2 and 3 consist of customers with a history of purchasing but not within the past two years. Group 2 customers have accessed their accounts within the past two years, whereas Group 3 customers have not. Group 4 customers are those who have never made a TreasuryDirect purchase but at some point in time logged into their accounts. We view these customers as inactive, and while they are not the primary group of study, we hope to learn from them the reasons for their lack of current interest in TreasuryDirect.

Each subgroup represents a distinct subpopulation of customers. Statistics developed on each of these strata will prove useful on their own, without reference to the other strata. Therefore, we’ll design our sample to provide a 95/5 percent level of confidence and precision for each subpopulation for yes/no questions related to customer satisfaction. This means we expect our measured satisfaction values will fall within five percentage points of the actual values for 95 percent of the samples we might draw at random. The overall level of confidence and precision should exceed 95/5 percent for the sample when considered as a whole. Statistics for the overall population will be obtained as a weighted average of the subgroup statistics based on the measured precision of the subgroup estimates.

With experience from only one previous TreasuryDirect survey to draw upon, we designed the sample based on a conservative customer satisfaction estimate of 50 percent. (Satisfaction actually exceeded 70 percent for all subgroups in the first survey.) This will provide additional assurance that our sample design will achieve required levels of confidence and precision.

We assume response rates will mirror those in the first survey, which ranged from 31 percent for non-purchasers (Group 4) to 57 percent for recent purchasers (Group 1). Response rate estimates do not affect overall burden calculations, as those who don’t undertake the survey don’t return it. As was the case with the first survey, we assume most respondents will choose web-based responses, and that the cost per survey response is constant across the four subgroups.

Table 1 summarizes the sampling frames, survey samples, and projected number of respondents. Table 2 shows the estimated burden per respondent and for the project overall.

Table 1.

	Purchased	Accessed	Respondent Universe	Survey Sample	Estimated Response Rate	Projected Completed Surveys	Margin of Error
Group 1	Within 2 yrs	At some point	170,863	703	57.0%	400	4.9%
Group 2	Over 2 yrs ago	Within 2 yrs	247,843	808	49.5%	400	4.9%
Group 3	Over 2 yrs ago	Over 2 yrs ago	13,713	1,213	32.1%	389	5.0%
Group 4	Never	At some point	187,430	1,275	31.4%	400	4.9%
			619,849	3,999		1,589	2.5%

Table 2.

Number of Respondents	Total Burden per Respondent (Minutes)	Total Annual Burden (Minutes)	Total Annual Burden (Hours)
1,589	9	14,301	238

Collection Method

We believe that the web-based relationship we share with our customers makes an online collection method appropriate. To become a TreasuryDirect account holder, customers apply online by providing personal information (e.g., name, social security number, address, e-mail address, and bank information). We then verify their identity before we allow them to purchase and hold Treasury securities in their accounts. This process gives us a high quality sampling frame for our target population. Therefore, we can reduce the expense to contact customers and leverage the built-in edits and prompts of online surveys to increase data integrity.

Procedures to Deal with Non-Response

Consistent with the research literature on survey data collection, we have planned at least four communications with the survey sample. First, we’ll send a letter on letterhead through surface mail, informing the customer sample of the upcoming Internet survey and asking for their participation. A week later our vendor will e-mail the sample an invitation to the survey with a link to the site and a password for access. Three follow-up reminders are planned, one by surface mail and two by e-mail, to further encourage responses. Third and fourth e-mail reminders are also in our budget, and we will judge the need to send them once we see the response rate.

All survey correspondence will carry our vendor’s e-mail address and toll free 1-800 number, with encouragement to contact them with questions about the survey. We’ll also instruct respondents to avoid sharing personally identifiable information in the survey and to contact Public Debt directly with questions about their TreasuryDirect accounts.

If a potential respondent prefers to respond to the survey questionnaire by some means other than the Internet, our vendor will offer a variety of options: fax, telephone interview, or hard copy with postage pre-paid return envelope. Experienced interviewers who have knowledge of the study will be used to explain its purpose and importance.

Response Rate

Although the information being gathered does not fall under the influential category and will be used for internal purposes only, we’ll take the following approaches to strive for an 80 percent response rate:

• Work with the vendor to create clear, concise questions and a user-friendly online format. The vendor will pretest the survey questionnaire on nine or fewer customers.

• Mail a letter to respondents in advance of the survey (as mentioned above). The letter will be personally addressed to the respondent and will include a) a pledge of confidentiality, b) a senior agency official’s signature, c) the approximate time needed to complete the survey, d) acknowledgment of the vendor conducting the survey, and e) a contact number and e-mail address so respondents can verify the survey’s legitimacy.

• E-mail an invitation to take the survey (as mentioned above). The invitation, sent by the vendor, will include a link to the survey. Respondents will use a password to gain access to the survey site; this should give them confidence that unauthorized people cannot meddle with their responses.

• Assure customers that the online survey is safe and secure by using Secure Sockets Layer (SSL) software to encrypt all respondents' information so that it cannot be read as the information travels over the Internet.

• Send follow-up letter and e-mail(s) to non-respondents (as mentioned above). We’ll assign a unique tracking number to each survey. The tracking number ensures each customer polled may only submit one survey, and it allows us to easily identify non-respondents for follow-up.

• Work with our vendor to take advantage of other reasonable avenues to encourage respondent participation. The vendor will follow all appropriate guidelines to maximize response rates listed in Questions and Answers When Designing Surveys for Information Collection.

Research suggests that response rates to web-based surveys have declined in recent years, most likely because of the preponderance of junk e-mail, filters, Internet surveys, and, perhaps, the Internet’s loss of novelty as a data collection vehicle. Nevertheless, we believe the online relationship we have with our customers has created trust and has allowed us to assemble reasonably accurate contact information—key factors that should allow us to get a good response rate.

The vendor will also supply a professional statistician to assure that non-response is not an issue in our analysis. We don’t believe the survey variables being measured will discourage participation in the survey. Because we’re not including sensitive issues in this survey, we don’t believe that refusals to answer survey items will be likely. This is a simple survey about satisfaction with an online method of buying Treasury products.

We’ll use the survey data mainly to gauge customer satisfaction and to improve customer service. Respondents’ feedback could affect product decisions.