Hewlett-Packard’s (HP) strategy is to provide high-tech, low-cost solutions with the best possible customer experience. The company strives to do that by leveraging its strengths in technology and innovation. Following this strategy, HP found that among HP business customers who use eSupport, more than 75% prefer it to the phone.

Our eSupport framework identifies three types of eSupport: preventive support (where monitoring tools identify and prevent potential critical problems by alerting customers or HP to take action proactively); self-support (where customers are given an array of web tools to help resolve their issues); and e-Assisted support (where customers can connect to HP over the web for assistance, whether by email, online chat, or through case submission). HP has tools, programs, and metrics in each of these areas, but for this article, we’re focused on self-support, and how we measure our effectiveness.

Our objective with self-support is to empower customers to quickly and accurately help themselves, and get the benefit of a rich web experience available 24/7, while increasing efficiency for HP.

This type of metric is often called “call avoidance” or “call deflection”, and is something that managers traditionally look for to help justify their eSupport investments . While our metric gives us a measure on call avoidance, we prefer the more positive name “self-solve success rate.” We use the self-solve success rate (expressed as a percentage) to derive another key metric for HP – the eSupport Delivery Mix. See the section on eSupport Delivery Mix below for more on this metric. We expect the customer mindset of reaching for the phone to change in the next few years as the majority of support is delivered through the Internet . Customers will expect and rely on the Internet for support – making the idea of call deflection not a meaningful paradigm.

To measure our self-solve success rate, we set up a survey four years ago, beginning with a six-month test basis, and moving that to a monthly schedule for more granular trending.

The survey asks a series of questions used to determine our successful self-solve rate. This summarized example comes from the HP IT Resource Center (ITRC), a technical support site for IT Professionals:

1.What was your primary reason for visiting?

1. Troubleshoot a problem or download a patch
2. Check the status of a call/case
3. Research an IT topic
4. Take a training course and/or find IT related education material
5. Contribute to a forums discussion
6. Other (please specify)

2. From your perspective, were you successful in resolving your problem or inquiry?

• Include only those who were successful.

3. Did you contact HP to find your solution?

• Include only those who did not contact HP.

4. Did you still contact HP, even after finding your solution?

• Include only those who did not contact HP.

5. If you had not been successful in finding your solution, would you have contacted HP?

• Include only those who would have contacted HP if not successful.

These questions are summarized in a decision tree (see fig. 1).

Figure 1: The decision tree for HP’s self-solve success survey.

The final number gives us a true measure of the percentage of customer problems that are successfully self-solved on the web.

In the survey information, we also capture the feature of the website that was most useful in solving the problem. This data has helped us make better decisions about retiring certain features and investing in others. For example, although the “Browse Hardware Products” feature in the ITRC was being accessed, it was consistently rated as “viewed but not useful” so we retired the feature.

At one time, we ran two surveys, one for self-solve success rate, and one for more in-depth customer satisfaction with different attributes of the site. These have been combined into a single monthly web-based survey. The survey pops up to a random selection of customers each month. A javascript element on the site checks for a cookie on the customer’s machine so we can be sure that the customer hasn’t been surveyed in the last six months. With the customer’s permission, the email address is gathered and the survey is emailed to the customer. We decided on this approach, rather than completing the entire survey online, because it’s more likely that the customer will have completed their online task, and will be able to judge whether their visit was successful. It’s also less intrusive to the customer in problem-solving mode without the time for an online survey.

The survey is localized in 10 languages - English, French, German, Italian, Spanish, Portuguese, Korean, simplified Chinese, traditional Chinese, and Japanese. The results are analyzed and tracked monthly. Over the years that we have tracked the self-solve success rate, we’ve seen an encouraging improvement in the results as our eSupport websites have been improved and enhanced over time.

While we’ve implemented the survey for both of our eSupport sites (ITRC for IT professionals, and Business Support Center for business professionals ), we have some variations by audience.

For example, in the ITRC, customers need to be entitled to use some areas of the site , so a question is included in the survey to see if entitled customers are more successful than those that are non-entitled. In the case of the Business Support Center, we found that a very high percentage of successful self-solves were due to driver downloads. Customers had come to solve a problem, found a driver, and successfully downloaded it. While this is good to know, we also wanted to see what our success rate was without taking drivers into account, and so developed two metrics – with and without drivers. Tracking the metric without drivers allows us to focus on the non-driver contributors to self-solve success – knowledge base, searching, community forums, etc.

eSupport delivery mix
Once we have our self-solve success rate percentage, we calculate the eSupport Delivery Mix. The eSupport Delivery Mix tells us what percentage of all support incidents coming into HP are resolved via the web, whether self-solve, eAssisted solve, or prevented cases. The mix we measure is between phone and web.

To derive the eSupport Delivery Mix, we first apply the self-solve success rate percentage to the total number of site visits, to show us what percentage of those visits were successful self-solves. HP’s support sites receive millions of visits every month, but not all of these are counted as successful self-solves. For example, if the self-solve success rate was 75%, and the total visits were 10 million, we could say that 7,500,000 of those visits were successful self-solves. We add this to the number of e-Assisted and prevented incidents in the month, and then divide by the total support incidents, including both phone and web (see fig. 2). This measure tells us what proportion of all support incidents were solved via the web.

Figure 2: The final delivery mix balancing phone and online solution methods.

We’re pleased with the upward trend in our eSupport Delivery Mix – now at approximately 80% worldwide. Web self-solve is a large contributor to this effort.

Pascale Vandenbroucke is responsible for the Voice of the Customer in HP Services eService Delivery organization. With 10 years of experience of gathering customer input via web surveys, she has developed an in-depth experience of customer wants and needs in eSupport, as well as expertise in web survey methodologies and vendors worldwide. Pascale's role is worldwide, covering the Americas, EMEA and Asia Pacific. Her office is based in Grenoble, France.

Lucy Owen is responsible for Industry Benchmarking and Competitive Analysis in HP Services eService Delivery organization. Her brief is to follow what the industry and the competition is doing in eSupport and make recommendations for HP. Her nine years experience at HP includes program management and marketing and strategy roles in eBusiness, eSupport, and online software delivery. She moved to HP's California offices from her previous job in the travel reservation industry in the UK.