Hack98.Key Performance Indicators for Customer Support Sites


Hack 98. Key Performance Indicators for Customer Support Sites

The surest path to lowering customer support costs is making sure that people are successful using your online support content. The surest way to do this is to regularly monitor a handful of support key performance indicators, watching for problems and monitoring the effect of changes.

The customer support business model is often overlooked by both vendors and businesses themselves when it comes to web measurement. Often, companies that have a significant support presence and very significant customer support costs fail to actively work to optimize their support web sites. Perhaps it's because so often, support sites are deployed using packaged applications that are difficult to modify or because support sites are so dependent on search. Regardless of the cause, there is a fantastic opportunity for companies to lower operating expenses by ensuring that customers are able to self-service their support needs, rather than making an expensive phone call.

7.9.1. Basic Key Performance Indicators for Customer Support Sites

The following key performance indicators should be considered core to the regular reporting for any customer support site. Each assumes a constant period of time, for example, the previous day or the previous week, unless otherwise noted, and most KPIs assume that you're able to segment or otherwise isolate traffic to your support site if that site is part of a larger site (which is common).

7.9.1.1 Average pages viewed per visit.

Defined simply as the total number of pages viewed divided by the total number of visits to the site, average pages viewed per visit is a strong indicator of the level of confusion and frustration of your customers. In contrast with the same performance indicator for content sites [Hack #97], a higher ratio of page views to visits is not necessarily good news.

The ideal customer support experience is one where visitors arrive, quickly find the solution to their problem, and leave satisfied. The longer a visitor is forced to hunt around for his answer, the more frustrated he is likely to become, often leading directly to a frustrated customer talking to your customer support staff. If yours is a well-designed site, one that has good search and easy-to-read answers, you're likely to see lower average numbers of page views per visit. If your customer support site is confusing or lacking somehow, you'll see a higher average (and your goal should be reducing that average).

7.9.1.2 Percent new and returning visitors.

Usually directly calculated by your measurement application, I recommend watching these percentages as a guide to how your business is growing and whether your customers are satisfied with your support site. Depending on the type of support you offer, the ideal situation is one where customers repeatedly use your site. If your percentage of return visitors is consistently low, you may want to consider checking your call-center volumes and, if they're consistently high, start asking your support customers if they've tried your web site.

Remember that the cost savings associated with having a support site revolves primarily around ongoing call deflection. If a new customer has a question and is able to find a satisfactory answer on your web site she's likely to use the site again; the opposite is also true. Also, as a caution, when you're examining these percentages, be sure to look at a long enough period of time to allow customers to need support more than once.

7.9.1.3 Average time spent on site.

Similar to the percentages of new and returning visitors, the average time visitors spend on your site is calculated by your measurement application. Average time spent should be used the same way as average page views per visitorthe ideal customer support visit is a short one. The longer the "average" visitor spends, the greater the likelihood that he's struggling to find an answer to his question. Especially when coupled with a high average page views per visit ratio, a long time spent on site can indicate frustration.

Really advanced systems are able to combine traffic data with customer support data by assigning a common unique user identifier [Hack #5] to join the two data sets. In this situation, companies are able to validate page view per visit and time spent measurements and begin to identify which types of customer problems are unlikely to be resolved online (and do something about it!)

7.9.2. Advanced Key Performance Indicators for Customer Support Sites

The following metrics are more advanced but should be seriously considered in addition to the basic indicators, especially those covering internal search.

7.9.2.1 Percent "zero results" searches.

While the subtleties of measuring internal search are many [Hack #64], one of the fundamental performance indicators for support sites is the percentage of searches that result in, well, no results (Figure 7-12). Calculated as the total number of "zero" or no results searches divided by the total number of searches executed, because searches that don't have results frustrate customers like nothing else, this metric is very important.

Figure 7-12. Sample search phrases


To make the calculation, you'll need the ability to differentiate "failed" searches from those that generate at least one result for your web measurement application. Often, search engines will return this in the query string or to the results page in a "results count" variable. Make the calculation using a page view count, allowing for the possibility that a visitor may search repeatedly and still get no results.

Considering that when customers search at your support site, they already have a product name and a pretty good idea of what the item is supposed to do, it is inexcusable for a site to return no results. Any increase in this percentage should prompt you to review your internal search logs looking for zero result searches, making adjustments to your search engine to return results whenever possible.

7.9.2.2 Support satisfaction rate.

If you are actively measuring customer satisfaction, either using your own technology or by working with a vendor like Usability Sciences, Foresee Results, or OpinionLab, I highly recommend that you include some measurement of visitor satisfaction (Figure 7-13) as a key performance indicator.

Figure 7-13. Customer satisfaction measurement


7.9.2.3 Percent "zero yield" search results.

When a customer is searching for an answer, presenting her with no results is bad but presenting her with confusing results that she doesn't believe will answer her question is worse. When faced with too many or hard to differentiate results, searchers often back up and search again without exploring any of the results, hence the result set "yields" no clicks.

Measuring zero results searches is complicated but valuable; measuring "zero yield" searches is very complicated but significantly more valuable. Defined as the number of search result sets that didn't generate clicks divided by the total number of search results returned (measured in page views), this KPI strongly depends on your web measurement solution having fairly robust click-tracking abilities.

Again, any increase in the percentage of zero yield searches should prompt you to examine those searches and critically explore the results set, attempting to identify the reason that a customer would fail to click even one link. A good strategy is to search your offline support database for the same type of information to identify the answers that your support agents are providing and comparing that to the online results set.

7.9.2.4 Percent of visits less than 90 seconds.

The percent of visits under 90 seconds is another valuable indicator of the likelihood that customers are successfully self-servicing their support needs. It is extremely rare that a support problem can be answered online in less than 90 seconds. The higher the percentage of short visits, the greater the likelihood that customers are simply looking up your phone number and driving up your phone support costs.

To make the calculation, you'll need to find the "time spent per visit" or "time spent on site" reportthe report that tells you the number of visits broken down by time spent (see [Hack #96] for additional details). Add up the number of visits less than 90 seconds and then divide by the total number of visits to the site.

7.9.2.5 "Stickiness" for key pages.

The stickiness of the first page in a visit should keep visitors interested and encourage them to click more deeply into the site. For support sites, these pages can help retain customers looking for information, preventing them from picking up the telephone.

Built from two page reportsthe entry page report, describing the number of visits that begin at a given page, and the single-access pages reports, describing the number of visits in which the page is the only one a visitor seesstickiness is one minus the ratio of single access to entry pages:

1.00(Single Access Page Visits / Entry Page Visits)

The result of the calculation will be a percentage, and higher percentages are better than lower, indicating greater stick on the page in question.

7.9.2.6 "Information find" conversion rate.

Your "information find" conversion rate is the ratio of unique visitors viewing content you consider an "answer" divided by all unique visitors. While you may end up with a very high conversion rate and still not be sure whether your customers are satisfied, you can be fairly sure that if this rate is low, something is wrong.

Provided you're successful in segmenting customer support visitors, reasonably you should have a nearly 100 percent conversion rate. Anything less than 50 percent should prompt you to look closely at the usability of your site and the likelihood that support-seeking visitors will be successful. The offline analogy would be someone calling a support hot line and then talking about the weather instead of asking for help.

7.9.3. Other Important Measurements

In addition to the KPIs listed above, you should pay careful attention to the top words and phrases that visitors are entering into your internal search engine. Because most customer support sites are often difficult to effectively navigate, research indicates that visitors have an even higher tendency to search. By carefully watching your top searches you will be able to make sure that these searches are yielding the "right" results and, if you're especially smart, modify top-level support pages to point directly to documents that answer the most pressing customer questions.

You should also watch your top entry pages, especially those driven from search engines, as a proxy for your internal search engine. Remember that the majority of searches on the Internet originate at Google. You want to make sure that you watch for an increase in Google searches for support content the same way as you'll watch for an increase in internal searches.

All of the aforementioned measurements, while not technically KPIs, are important enough to be delivered with the key performance indicator report. For a sample key performance indicator worksheet, built in Microsoft Excel from the measurements recommended in this hack, visit the author's web site at http://www.webanalyticsdemystified.com/kpi_worksheet.



    Web Site Measurement Hacks
    Web Site Measurement Hacks: Tips & Tools to Help Optimize Your Online Business
    ISBN: 0596009887
    EAN: 2147483647
    Year: 2005
    Pages: 157

    flylib.com © 2008-2017.
    If you may any questions please contact us: flylib@qtcs.net