Benchmark report: technical support cost ratios.
At the same time, however, there seems to be remarkably sketchy information about the underlying cost structure of tech support. Even the largest publishers often lack reliable data about industry-wide benchmarks or their own internal support costs, and this lack of data makes forecasting and productivity measurement notoriously unreliable.
To help fill this gap, we've put together a new Benchmark Survey on Technical Support Cost Ratios. Drawing on data supplied by 148 personal computer software developers and publishers, the survey is designed to provide a detailed picture of industry-wide cost ratios, productivity levels, and fee-paid trends among PC-based tech support organizations.
For those who like an overview of the survey results, here's a quick summary of what our data shows:
* The median cost of support for PC software companies currently amounts to 6% (median) of total revenues. This ratio varies considerably by application category: The most expensive products to support include vertical-market and job-specific titles, programming tools, and accounting and finance applications (which typically cost about 10% of revenues), while the least expensive include utilities and consumer titles (which cost 3%-4% of revenues).
* The fully-loaded median cost of answering a single call is $23.33. About two-thirds of this amount represents the cost of direct labor; the balance includes departmental operating costs and other overhead. Based on a median call length of eight minutes, the average cost of answering a tech support call is thus about $3.00 per minute.
* Employee productivity is highest in organizations with four or more employees, which average 200-300 calls monthly per employee (equal to 11-14 calls per day). Small support groups are significantly less productive; average monthly call volume for these organizations runs about 80 calls per employee.
* Typically, users place about three-quarters of all tech support calls (73%) while products are still under warranty. This ratio also varies considerably by application category: Only 11% of calls for accounting and finance titles are covered by warranty, compared to 95% for education titles.
* Especially in high-end product categories, paid support programs already widely accepted: A quarter of the companies in our sample say their end users currently pay for 20% or more of all tech support calls.
Like earlier Soft (*) letter Benchmark Surveys, this report is based on data drawn almost entirely from personal computer software companies, rather than from companies that develop mid-range and mainframe applications. This distinction affects the economics of software support in a number of significant ways: PC software companies typically generate less revenue per sale than large-system vendors, are less likely to sell ongoing maintenance contracts, and more often provide direct support for individual end users.
In addition, the PC software market is dominated by smaller companies, whose size has an impact on the resources and manpower available for tech support services. Although a quarter of the companies in our sample report annual revenues in excess of $10 million, more than half (59%) have revenues of $5 million or less.
When we look specifically at the size of the tech support group in most PC software firms, we see similar demographics. In fact, half the companies in our sample operate with three or fewer tech support employees, and it's apparent from the productivity statistics in these companies that many of the employees in small support organizations don't work exclusively on support tasks.
We also asked our survey respondents to identify the application category and list price for their "most popular title," and the answers point to a remarkably broad range of categories and price points that is typical of the PC software marketplace:
Application categories and median prices Vertical/industry- or job-specific (28 responses) $1,495 Programming tool/language (23 responses) $430 Accounting/finance/tax (12 responses) $397 Publishing/presentations/fonts (6 responses) $199 General business (26 responses) $295 Communications/networking (11 responses) $199 Utility/add-on/desk accessory (13 responses) $99 Education (10 responses) $50 Consumer/entertainment.(14 responses) $50 All respondents $249 Note: Two categories, CAD/graphics and system/environment, received fewer than s ix responses and are not reported as separate application categories.
WHAT'S THE REAL COST OF TECH SUPPORT?
When we look at our survey data, it's clear that no single industry-wide benchmark adequately measures support costs and employee productivity. However, there are three ratios that tend to be reliable (and easily computed) indicators of overall performance: tech support costs as a percentage of total company revenues, the cost per call for direct labor, and the average monthly call volume per support employee.
In analyzing these ratios, we also calculated three additional variables - company revenues, total employees in the support organization, and primary application category - that have an impact on support economics:
* Company size: In general, the survey data suggests that economies of scale take effect at a fairly early stage in company growth - about $5 million in annual sales. At this point, call volumes become large enough to justify a full-time support staff, and there are usually other efficiencies in equipment cost, administrative support, and facilities overhead. As a result, technicians in larger companies handle more calls per month, at a lower cost per call, than their counterparts in smaller companies. But the data also suggests that the largest companies - those with sales over $10 million - are actually slightly less efficient than firms in the $5-$10 million range:
Productivity and annual revenues Cost % Cost/Call Calls/Emp. Calls/Emp Revenues (dir.labor) (monthly) (monthly) $10 million $6.0% $10.41 233 5-$10 million 5.8% $10.11 300 1-$5 million 5.0% $14.18 152 Under $1 million 10.0% $24.00 50 All respondents 6.0% $15.00 139 (*)All values are medians.
* Organization size: We see a similar pattern when we compare cost and productivity ratios for support groups ranked by total headcount. Again, the largest departments - those with ten or more employees - turn out to be somewhat less efficient than support groups with 4-9 employees. But the most costly organizations are still the very small groups, especially one-person support departments, that often end up spending a lot of time on tasks that aren't directly related to telephone support.
Productivity and organization size cost % Cost/Call Calls/Emp. Calls/Emp. Revenues (dir.labor) (monthly) (monthly) 10.+ TS employees 6.4% $15.00 182 4-9 TS employees 7.0% $9.22 233 2-3 TS employees 8.0% $19.50 78 1 TS employee 4.0% $16.25 80 All respondents 6.0% $15.00 139 (*)All values are medians.
* Product category: Perhaps the most revealing way to slice the survey data is by individual application categories. Not surprisingly, software developers who support high-end products tend to handle fewer (and more expensive) calls per technician, and they invest a larger share of total revenues in support. By comparison, companies that support mass-market titles spend a good deal less of total revenues on support and often achieve much higher levels of employee productivity:
[TABULAR DATA OMITTED]
These numbers also show an explicit relationship between call volume and call costs: The higher the number of calls that technicians handle per month, the lower the cost per call. Moreover, this relationship often has very little to do with the price of the product. Thus, titles in the consumer/entertainment category (with a $50 median price) generate a monthly average of 560 calls per technician, at a cost of $4.23 per call; education titles (median price $47) generate just 30 calls per technician, at a cost of $35.71 per call.
* Call statistics: Another useful way to look at the relative cost of support in different application categories is in terms of the number of calls per copy and the average length of those calls. The numbers' help explain the relatively high support costs of complex products, which typically generate two or more (long) calls per copy sold. At the same time, the call statistics also demonstrate how low-end software companies can "afford" to support inexpensive titles: Only a small fraction of their users ever call for support, and these calls tend to be shorter and, presumably, easier to answer.
Call statistics by application category* Cost % Call Lgth. Calls/ Revenues (mins.) Copy Vertical/industry- or job-specific 10.0% 10 2.0 Programming tool/language 10.0% 10 2.0 Accounting/finance/tax 9.9% 9 3.0 Publishing/presentations/fonts 7.5% 7 0.7 General business 5.3% 8 0.4 Communications/networking 5.0% 10 0.4 Utility/add-on/desk accessory 4.3% 6 0.3 Education 3.8% 7 0.1 Consumer/entertainment 3.5% 4 0.1 All respondents 6.0% 8 0.4 (*)All values are medians.
HOW MUCH SHOULD PAYROLL COST?
Not surprisingly, the largest single cost component of technical support is labor. Even in large organizations with relatively high levels of automation, payroll and related benefits usually account for well over 50% of total operating costs; in small support departments, payroll often represents 75%-85% of costs. When we analyze payroll costs by job function, moreover, we find that most of this spending is for direct labor -- that is, technicians whose primary job involves answering telephone questions or otherwise interacting directly with users. We were able to collect job classification data on a total of 2,724 support employees; of these, 79.2% are direct labor, 12.7% management or supervisory, and 8.1% administrative or clerical. (We found a nearly identical ratio a year ago, when we explored support and service productivity; Soft * letter, 7/7/92.)
Payroll costs vary somewhat from one region to another, so several large software companies -- including Microsoft, Lotus, and Symantec -- have set up support departments in parts of the country with lower labor costs. But it turns out that a more critical cost variable is the size of the support organization. Mid-sized support departments pay salaries for direct labor that are roughly comparable to salaries in larger companies, but mid-sized departments seem to pay a good deal more for the pro-rated cost of supervisory and administrative staff:
Payroll costs by organization size (monthly)* Payroll Payroll (dir.labor) (other) 10+ TS employees $3,038 $749 4-9 TS employees $3,433 $1,000 2-3 TS employees $3,000 $1,417 1 TS employee $2,300 n/a All respondents $3,000 $949 (*) Payroll (direct labor)" reflects total salary and benefits per employee divided by the number of direct labor employees. "Payroll (other)" reflects total salary and benefits for management, supervisory, administrative, and clerical staff divided by the number of direct employees (to arrive at a pro-rated value). (*) All values are medians.
HOW MUCH SHOULD OPERATIONS COST?
In addition to payroll, support organizations spend money on a wide variety of other direct and indirect operating costs -- telephone charges, rent, equipment, training, and the like. For forecasting and budget purposes, it's often helpful to calculate these expenses on a per-employee basis. Our survey data suggests that monthly non-payroll costs typically add up to about $1,100 per employee. Moreover, the largest and most efficient organizations -- those with ten or more employees -- report spending patterns that are generally in line with all other support departments:
Non-payroll operating costs* All TS Organizations 10+ Employees General & administrative 28% $300 19% $219 Telephone/telecomm 23% $250 27% $300 Facilities/rent 19% $205 22% $252 Equipment & depreciation 12% $127 16% $177 MIS services 10% $107 9% $100 Travel & seminars 4% $46 4% $47 External training 4% $43 3% $35 Total 100% $1,078 100% $1,130 (*) Median spending per tech support employee, per month.
In looking at these numbers, it's particularly interesting that three categories -- general & administrative, facilities/rent, and MIS services -- account for more than half of all non-payroll operating costs. These costs tend to be allocated to the support department, often according to formulas that don't realistically reflect expenditures. Thus, one way for support managers to reduce costs may be, simply to renegotiate the allocation formulas their company uses in these areas.
WHAT'S THE STANDARD COST PER CALL?
In addition to the direct labor cost of handling a tech support call, we calculated a cost benchmark based on the total cost of a single call. This figure incorporates direct labor costs, "other" labor (management, supervisory, administrative, and clerical payroll), and all non-payroll operating costs. When these costs are rolled together, the results suggest that a typical tech support call costs between $17 and $25 to handle:
[TABULAR DATA OMITTED]
WHAT'S THE MARKET POTENTIAL FOR PAID SUPPORT?
Finally, our survey results shed some light on the much-debated issue of paid support. The data suggests that customers in specific application categories are more likely to be receptive to fee-paid offerings -- and more likely to reject (or ignore) paid support in other categories.
Here, one key metric is the number of calls that fall outside the publisher's normal warranty period (usually, 90 days from the user's first call). For most publishers, the majority of warranty calls deal with installation, printer set-up, new features, compatibility with existing applications, and other questions that most users expect should be answered for free. For our survey sample as a whole, 73% of calls fall within this warranty period, but there are significant variations across product categories. Accounting and finance products generate the largest percentage of post-warranty calls; vertical market titles and programming tools also generate a good deal of extended support activity. The potential market for fee-paid offerings in these categories is certainly greater than in the general business, consumer and utility categories, where most calls (75%) are generated during the warranty period.
Similarly, there are large differences across product categories in the current acceptance of fee-paid programs. In some high-end markets, paid support is accepted as a normal business practice; in other categories, it's sufficiently exotic as a concept that the cost of marketing these programs to users is probably prohibitive. Since there's no direct benchmark for measuring this relative level of acceptance, we looked at the percentage of companies in each product category where fee-paid calls make up more than 20% of total call volume:
[TABULAR DATA OMITTED]
WHERE IS THE INDUSTRY GOING?
Our data provides what is essentially a snapshot of support economics, taken at a moment in time when almost every major developer has started to rethink and reengineer basic tech support policies. So any predictions about the future -- even the near future -- are likely to be sheer guesswork. Still, we'd like to offer a few observations:
* The myth of cost escalation: We looked at tech support costs more than three years ago (Soft-letter, 2/18/90) and found that the median spending level at that time -- before the explosion of new Windows titles and the recent price wars -- was the same 6% of revenues that our latest data shows. Some companies certainly face escalating support costs, but we see little evidence that rising costs are truly an industry-wide phenomenon. When we look at support costs among consumer software companies, moreover, it's clear that declining product prices won't automatically translate into higher support cost ratios.
* The productivity question: We feel strongly that the real problem with tech support costs is the industry's failure to grapple with fundamental productivity issues. Expensive manpower is too often used to answer routine and repetitious calls; too much time is spent on training, research, and other tasks that drastically reduce the amount of time technicians actually spend on the phone. Rather than rely on paid support plans (which may be desirable for other reasons) to reduce support costs, we think it makes far more sense for software companies to focus on automation and other productivity-enhancing techniques. when the industry figures out how to handle more customer questions with fewer people, a genuine revolution in support economics will finally get under way.
[TABULAR DATA OMITTED]