Benchmark survey: support & service productivity.

In the early days of the telephone industry, every phone call had to be placed through a switchboard manned by a live operator. As the number of phones and connections grew, telephone companies found themselves recruiting and training vast armies of operators--at a rate (one analyst predicted) that eventually would force telephone companies to hire the entire U.S. population to staff all those new switchboards.

These days, the trend in tech support and customer service staffing look strikingly similar. Every shipment that leaves a loading dock triggers a fresh round of support and service inquiries. New technologies and features bring more calls for help from baffled customers; at the same time, the shift to direct response channels has added more order processors, trouble-shooters, and expediters to software company payrolls. If the present trend continues, it's possible that software companies eventually will have to hire the entire U.S. population to handle the flood of support and service calls.

Of course, the trend won't continue. The telephone industry solved its own manpower crisis by replacing most of its human operators with automated switching systems  (in a sense, everyone who now dials a phone has become a switchboard operator, so the prediction wasn't so far off the mark). We're not sure if the software industry will discover an equivalent automation solution, but it's clear that developers can't just keep throwing bodies at the problem. Somehow, support and service departments will have to become dramatically more efficient, more productive, more capable of supporting larger numbers of users without equally large growth in staffing levels.

The trouble is, industry-wide productivity benchmarks are surprisingly hard to find. For that reason, we recently surveyed a broad sample of software companies about some key service productivity issues. In particular, we looked for overall productivity levels (measured in transactions per employee), the percentage of employees who actually work in "line" positions, and the impact of automation solutions on productivity. We also explored several variables--revenue size, product price, and application category--that affect company-to-company comparisons.

Our total survey sample (after discarding incomplete questionnaires) consisted of 148 companies, of which 45 reported annual sales of less than $1 million, 45 had sales in the $1-$5 million range, and 44 exceeded $5 million (14 provided no revenue data). We also asked about the "total number of customer service and tech support employees, including managers, as of March 31, 1992 (full-time equivalents)." Here, the breakdown of the sample was 40% with 3 or fewer service employees, 30% with 3.5 to 9 employees, and 30% with 10 or more; the median number of service employees in our sample universe was 4.

Several conclusions emerged from the data:

* Productivity levels: Companies with revenues over $5 million-- which tend to have specialized service and support staffs-- typically handle about 1170 tech support and 1540 customer service transactions per quarter per employee. Small and midsized firms by comparison achieved significantly lower levels of productivity.

* Overhead ratios: Both in tech support and customer service, we found a good deal of consistency in the ratio of line employees to management/supervisory and administrative/staff positions. Typically, about 80% of tech support staffing consists of line employees, compared to about 77% of customer service staffing.

* Automation trends: Not surprisingly, we found that automation technologies--which our survey defined as "fax-back, bulletin board, automated voice, customer-initiated database search, and other responses that do not involve a live employee"--still haven't achieved wide penetration in service and support departments. But the early adopters seem to be achieving very promising results. Among companies with some form of service automation, 11% of tech support and 12.5% of customer service transactions are now handled with no human intervention. Moreover, these companies achieve dramatically higher levels of employee productivity--often twice as high as their nonautomated counterparts.

ORGANIZATION AND STAFFING RATIOS

Increasingly, software companies these days seem to be drawing a distinction between product-related "technical support" and salesrelated "customer service." Most of our respondents  (124 companies out of 148) were able to break out staffing levels for each of these functions, and we found some significant differences between the two groups in such areas as call volume and automation.

However, tech support and customer service groups were very similar in one area: the ratio of line employees to management and administrative staff. This is an important percentage for analyzing group productivity; clearly, a support department that suffers from too much overhead will be inefficient, even if its individual technicians perform adequately.

Predictably, overhead ratios are not especially meaningful for service and support departments with just a few employees. The smallest companies in our sample reported that 100% (median) of their tech support and customer service employees performed line functions. However, once a company passed the $1 million level in revenues, or employed more than two support technicians, the ratios changed considerably:

  Tech support staffing ratios (medians)*
                                         Manage-  Adminis
-                                    Line  ment    trative
All respondents  (148 companies)    80%    14%     0%
By annual revenues:
   Under $1 MM (45 companies)       100%   0%      0%
   $1-$5 MM (45 companies)          75%    17%     0%
   $5+ MM (44 companies)            80.5%  14%     5%
By tech support staffing:
   Less than 2 emps  (45 companies) 100%   0%      0%
   2-5 emps  (55 companies)         75%    25%     0%
   5+ emps  (48 companies)          81.5%  13.5%   5.5%

* Totals of medians do not always add up to 100%.

A similar pattern emerged in from our customer service data, although the typical customer service headcount tended to be lower than in tech support departments:

  Customer service staffing ratios (medians)*
                             Line   Management   Administrative
All respondents (124 companies)77%   13.8%          0%
By annual revenues:
   Under $1 MM (35 companies)100%     0%             0%
   $1-$5 MM (45 companies)   67%      17%            0%
   $5+ MM (44 companies)     75%      16.2%        3.6%
By customer service staffing:
  1 or fewer emps
              (46 companies)100%      0%             0%
  1.2-4 emps  (44 companies) 67%     25%             0%
  5+ emps  (34 companies)    77.4%   15%             9%

* Totals of medians do not always add up to 100%.

PRODUCTIVITY LEVELS

"Productivity" is never an easy quality to measure, and we've seen some heated arguments over what yardstick is most appropriate. (One survey respondent sent back a questionnaire with an angry comment: "This survey is 100% meaningless because it does not ask for installed base or units sold/month.") Ultimately, the most efficient companies are going to be the ones that eliminate the reasons for service and support calls, leaving their technicians as idle as the famous Maytag repairman.

But until the telephones stop ringing, probably the most useful way to monitor support productivity is simply total calls (and related inquiries) divided by total support employees. The value of this standard is that it measures the short-term impact of the real-world productivity investments in training, automation, and management skills. If the average technician ends up handling more transactions (with no burnout or loss of customer satisfaction), then certainly service productivity has improved in a measurable way.

To develop a common baseline for calculation transaction volume, we asked our respondents for the total number of tech support and customer service transactions their departments handled between January and March 1992. We also asked them to break out how many of these transactions they handled by telephone, by automated systems, and by "other" methods (including "in-bound faxes, letters, and sales orders, including upgrade purchases").

Surprisingly, even many large companies don't seem to collect this kind of data in any systematic fashion; 87% of our respondents said the numbers they supplied were based on estimates rather than actual transaction counts. But our statistical sample was large enough to provide a reasonably accurate picture of overall industry performance standards and trends.

Thus, we found that larger software companies typically reported the highest transaction volumes per employee in both tech support and customer service. Small and mid-sized firms were dramatically less productive; however, we found in earlier surveys (Soft-letter, 2/18/90 and 7/31/91) that service employees in smaller companies typically perform a broader range of tasks than technicians in large companies. Thus, smaller companies aren't necessarily less efficient, but it is more difficult to measure their efficiency in terms of standard performance benchmarks.

  Tech support transactions per employee, Q1 1992
                                  Median     50% Range
By annual revenues:
   Under $1 MM (45 companies)      142           60-440
   $1-$5 MM (45 companies)         664         335-1175
   $5+ MM (44 companies)           1173        667-1839
By tech support staffing:
   Less than 2 emps  (45 companies) 458         116-912
   2-5 emps  (55 companies)         360         100-825
   5+ emps  (48 companies)         1210        658-1831

Customer service transactions tend to be relatively brief, so service reps are typically able to achieve higher productivity levels:

Customer service transactions per employee, Q1 1992
                                    Median      50% Range*
By annual revenues:
   Under $1 MM (35 companies)       192         50-650
   $1-$5 MM (45 companies)          850         473-1700
   $5+ MM (44 companies)           1543        440-3450
By customer service staffing:
   1 or fewer emps  (46 companies)  447         30-938
   1.2-4 emps  (44 companies)       487         140-1167
   5+ emps  (34 companies)          1867        536-3543

Support and service performance is also affected by such factors as product complexity and customer expectations within individual market niches. We asked our respondents to identify the product category for their "most popular title," and that breakout shows major differences in productivity across various market segments. (In the case of accounting and financial software companies, the data also shows how a seasonal spike in transaction volume can temporarily skew the numbers.)

Productivity by product category*
                       Tech Support         Customer Service
Accounting/
finance/tax           1979  (11 companies)  3333  (9 companies)
publishing/
presentation/fonts    1311  (5 companies)   3250  (4 companies)
Communications/
networking            933  (10 companies)   271  (7 companies)
General business      886 (33 companies)   1000  (27 companies)
Utility/add-on/desk
accessory             800  (15 companies)   800  (13 companies)
CAD/graphics          700  (5 companies)   1168  (4 companies)
Programming tool/
language              540 (15 companies)    850  (11 companies)
Vertical/industry- or
job-specific          458  (29 companies)   487  (20 companies)
System/environment    325  (1 company)          n/a
Consumer/entertainment246  (14 companies)   360  (12 companies)
Education             158  (10 companies)   142  (9 companies)
* Based on total transactions per employee,  Qi 1992.

Finally, we also found some interesting correlations between service productivity and the list price of a company's flagship title. Here, the most productive companies turned out to be those that supported software in the $151-$500 price range, while the least productive were firms with titles priced at the $1000+ level (whose customers apparently expect more extensive hand-holding). But low productivity was also characteristic of companies with the least expensive titles--who, because of their low revenue per sale, presumably should be most aggressive about support efficiency.

Productivity by price category*
                         Tech Support         Customer Service
Under $150  (46 companies)  568                      659
$151-$500  (52 companies)   898                      900
$501-$995  (28 companies)   654                      775
$1000+ (21 companies)       325                      175
* Median transactions per employee,  QI 1992.

THE IMPACT OF AUTOMATION

Recently, we noted that automation technologies in other service industries--such as retail, banking, travel, and hospitality--have enabled employees to process much higher numbers of transactions without compromising the quality of "personal" service (Soft-letter, 5/11/92). Software companies, the perennial shoemaker's children of productivity technologies, have only begun to explore the potential of service automation. Customer tracking systems, knowledge bases, automated call distribution systems, fax-back systems, and electronic bulletin boards are certainly penetrating support and service organizations, but only at

To be sure, it's not always easy to measure the payback from support automation. Rapid company growth and increasing product complexity make year-to-year comparisons almost meaningless, and many commercial automation products have been evolving at an even faster rate. Large corporations are still debating whether desktop computing has had any effect on the productivity of white-collar office workers, so it's perhaps premature to expect much hard data about service automation.

Still, we did try to create a rough measure of the impact of automation on software service and support organizations. We asked our respondents to break out the number of "automated" transactions they handled during the first three months of the year, and compared these numbers to overall transaction volumes and to productivity levels. (Automated transactions included "fax-back, bulletin board, automated voice, customer-initiated database search, and other responses that do not involve a live employee.")

When we tabulated the responses, we found that 54 companies (36%) reported at least some automated tech support transactions, and 29 companies (23%) reported some automated customer service transactions. Within this group of automation pioneers, we found a surprisingly high percentage of transactions have been offloaded from live technicians and service reps--11% (median) for tech support transactions, and 12.5% (median) for customer service transactions. To the extent that these percentages translate into decreased demand for new employees, there's clearly going to be an attractive role for automated support systems.

When we compare the overall productivity level of automated companies against those that reported no automated transactions, moreover, the results are even more dramatic. Depending on company size, automated respondents in our survey sample were able to handle between 7% and 165% more transactions per employee than their non-automated counterparts:

  Automated vs. non-automated productivity levels
by company size
                      Automated   Non
-                                Automated  Improvement
Under $1 million
   Tech support*        300       113          165%
   Customer service*    350       151          132%
$1-$5 million
   Tech support        700        655            7%
   Customer service   1700        700          143%
$5+ million
   Tech support       1311        974           35%
   Customer service   2333       1230           90%
* Median transactions per employee,  QI 1992.

In a sense, these numbers demonstrate an obvious point: In almost any industry, companies that actively pursue productivity gains almost invariably end up the most efficient competitors. Automation by itself doesn't necessarily produce big productivity improvements, but even a toe-in-the-water automation investment is usually a sign that a company's management is looking for smarter ways to work.

And the bottom line seems to be that software companies can improve their service and support productivity in very substantial ways. No single automation or management solution is likely to transform the economics of handling thousands of customer transactions, but we suspect that the industry could achieve an overall doubling of support and service productivity in the next few years. Certainly, that scenario makes a lot more sense than the open-ended hiring binge of the last decade.

A FEW OTHER NUMBERS...

As part of our data collection process, we compiled a few other statistics about current service and support productivity trends:

* The paperwork shuffle: Software companies handle customer inquiries primarily by telephone, but many requests come in by mail or by fax. Since these transactions tend to be more time-consuming  to handle, we asked how many total transactions fell in this "other" category. The results suggest that large companies handle a relatively smaller share of paper transactions than small and mid-sized firms:

  In-bound faxes, letters, and sales orders
     by company size
                                        Median       50% Range
 All companies
    Tech support*  (67 companies)          7%           3%-13%
    Customer service*  (62 companies)      27%          9%-47%
 Under $1 million
    Tech support  (15 companies)           9%           5%-20%
    Customer service  (16 companies)       29%          20%-47%
 $1-$5 million
    Tech support  (17 companies)           9%           4%-17%
    Customer service  (18 companies)       36%          12%-62%
 $5+ million
    Tech support (30 companies)            4%           2%-9%
    Customer service  (24 companies)       10%          4%-38%
      * As a percentage of total customer transactions,  Qi 1992.

* Upgrade processing: Large upgrade campaigns tend to disrupt even the most efficient customer service department, so we've found that even companies with strong in-house order processing capabilities are beginning to explore the use of outside fulfillment services. This approach seems to be most common among large companies:

  How do you process upgrade orders?
All companies
        "Mostly through in-house customer service group"83%
        "Mostly through outside vendor or fulfillment service"12
        "Outside service for peak periods,  otherwise in-house"5
Under $1 million
   "Mostly through in-house customer service group"        92%
   "Mostly through outside vendor or fulfillment service"  4
   "Outside service for peak periods, otherwise in-house"  4
$1-$5 million
   "Mostly through in-house customer service group"        89%
   "Mostly through outside vendor or fulfillment service"  7
   "Outside service for peak periods, otherwise in-house"  5
$5+ million
   "Mostly through in-house customer service group"        60%
   "Mostly through outside vendor or fulfillment service"  28
   "Outside service for peak periods, otherwise in-house"  12