Compensation Surveys

by Staff - Original publish date: December 5, 2011

Compensation Surveys

Paying people fairly is good for business. Underpay, and employees will eventually look for a better offer. Overpay, and the payroll budget and profitability will suffer. That's why companies use market data to research the value of their jobs. But what is "market data" anyway?

To determine the prevailing rate for a job, companies can "benchmark" jobs against compensation surveys that are detailed and specific to the companies' industries and regions. A good compensation survey uses standard, proven methods of data gathering and statistical analysis to determine how much companies pay for a specific job in a specific industry. A number of types of organizations conduct salary surveys, including compensation information businesses, compensation consulting firms, industry associations, educational institutions, and state and federal governments.

More than 80 percent of business managers and HR professionals said their companies either participate in or purchase at least one salary survey each year, according to a poll. Companies with fewer than 500 employees spend an average of $2,000 annually on salary surveys, and companies with more than 5,000 employees spend up to $15,000 or more each year on these important data sources.

Companies pay for compensation data because the benefits exceed the costs. The amount companies spend on surveys is just a fraction of a percent of their total payroll costs. For example, although a company with 5,000 employees may spend $12,000 on compensation surveys, its total payroll is probably at least $15 million - in which case, their survey cost would be just eight one-hundredths of one percent of payroll.

Companies that participate in surveys (i.e., provide their own compensation data) customarily receive a discount on the final report. Fees for compensation surveys range considerably depending on the scope of the survey (regional vs. national, number of jobs surveyed, etc.). Participants could pay as little as a few hundred dollars for a small regional survey, or a few thousand dollars for a comprehensive, national survey. Perhaps the most expensive surveys are very specific regional surveys - those that pinpoint a very particular segment of the recruiting marketplace. Regardless of the survey, non-participants typically pay more than participants.

How do researchers conduct compensation surveys?
Surveys are conducted on a semiannual, annual, or biennial basis. Surveys normally fall into one of two categories: Custom and standard. Custom surveys are ones that attempt to answer very specific questions from a narrow selection of peer companies (e.g., What is the prevailing pay rate for salespeople in the pharmaceuticals business in the Northwest?). These custom surveys tend to be available to, and used by, the participants only. Standard surveys, on the other hand, are often published each year and attempt to cover the same range of companies and jobs. These broad surveys are sometimes sold to non-participants and made available to members or customers of the survey sponsor/vendor. This focus of this article is primarily on standard compensation surveys.

The process of collecting data and producing a salary survey takes careful planning and execution that requires economic investment, people resources, and time. Some companies conduct surveys in-house using their own staff and compensation experts. However, most companies contract a third party to collect the data and do the number- crunching. The third-party approach provides a level of independence that most participants want. Some salary surveys are co-sponsored to attract more participants and to add credibility to the numbers. An experienced data provider in survey methods and statistical analysis is expected to put out high-quality, reliable, accurate data.

Conducting a salary survey is a time-consuming task. A traditional survey of 15 companies encompassing 20 positions can take between 6 and 12 weeks from the initial planning to the time the survey is distributed to participants. For a survey that includes more participants and more positions, it could take as long as four to six months. Survey respondents then have two to six weeks to complete the questionnaire. The length of time depends on the number of positions surveyed and the amount of information requested for each incumbent (person in a given job). After the data has been collected, it can take two to three months to analyze the data and make the findings available to the survey participants and others. Therefore the time from initiating a survey and providing results can be up to 7 months or more.

Traditionally, survey questionnaires are mailed out in paper form or on a diskette to participants, namely company managers or executives and human resources professionals, who will then complete and return the survey before a predetermined closing date. When assessing the methodology of a survey, it is important to look for the number of surveys mailed out, the number of participants, and the number of employees in the report summary. These numbers determine whether the survey is representative of the jobs and the industry. Also, there are several ways to collect and summarize data and it is critical that the user understand the underlying assumptions of a particular survey to assure its data is being used properly. It is even more critical when using multiple surveys to be sure that comparisons or compilations are done appropriately - on an "apples-to-apples" basis.

Compensation survey checklist
Here are some considerations to weigh for a company who is deciding whether to purchase a compensation survey.

  • The background of the survey research firm and cosponsors, if any. Look for reputable firms that follow proven methods to gather and analyze compensation data.
  • The scope of the survey. Look for studies that cover industries, jobs, and regions that are most applicable to your purposes; and that provide data on enough jobs to be cost-effective.
  • The survey methodology. Review the summary of the methodology to make sure it's consistent with standards set forth by reputable industry associations such as WorldatWork. Be especially sure the research organization is surveying human resource professionals or other people knowledgeable about compensation information within a company, rather than individuals.
  • The number of participants in the survey. A good survey should cover a representative number of companies for its target population. A survey doesn't have to cover the entire industry or region to be robust; even a few dozen responding employers in some industries can provide enough data for a valid survey.
  • The names of participants. Look for your competitors and peers. For many jobs, you may be competing for candidates with companies in different industries but the same geographic area.
  • The number of incumbents covered by the survey; and the sample size for each salary. A sample size of 30 or more is more statistically significant than a sample size of 10, provided the sample is representative of the statistical population.
  • The relevance of the job descriptions to the positions being benchmarked. Look for a good match between the survey and your company. Be sure to compare job descriptions, not just job titles.
  • The effective date of the survey data. The date a survey is published is always later than the effective date of the data within the survey. If necessary, age the data from the effective date to the current month.

Multiple survey sources. As with any form of research, it is important to use multiple data sources to narrow in on the "true" answer. Relying on a single source can be misleading if that source doesn't perfectly reflect the market in question. WorldatWork suggests that compensation analysts should use multiple data sources wherever possible; consulting firms and academics agree. The exceptions come when there is only one data source, or when there is a spot-on data source, such as a custom survey, that truly describes a precise market.

Number of participants. Make sure the participants are a good sample of the recruiting market. Generally, eight to ten participating companies is a good sample for positions below the management level. The sample size should increase the more senior the positions being surveyed, both to get a good representation and to allow for more job matches, since each company is organized differently. There could be limited pay data in some industries, or the available data might not be representative of the industry because of a low participation rate in the survey.

Some firms reveal a list of participants, or at least those well known within the industry. The surveying company may disclose big-name participants to draw more interest from smaller companies. A list of major employers can also add credibility to the survey.

An important exception to note is that if a compensation analyst or compensation consulting firm is using multiple surveys to produce their own derivative market numbers, they will aggregate the data by combining the surveys, placing differing weight on different sources and sometimes even making a qualitative adjustment. When the data has been aggregated in this manner, it is not customary to report numbers or names of participants.

Participant profiles. The usefulness and relevance of a salary survey depends largely on the survey participants. For a small company, a salary survey of large corporations in the United States will be less helpful in determining what to pay employees than a survey of smaller organizations. Of course, a small company in a "company town" may find itself in a position to have to pay the same wages as the predominant employer in that town.

Survey participants can be quite different, depending on the goal of the survey. If the survey covers pay in large companies in different geographical locations, the surveying company has to make sure that companies participating in the survey are of similar size but from different locations.

To participate or not to participate
Here are some considerations for a company to weigh when deciding whether to participate in a compensation survey

  • The prisoner's dilemma. Every participant improves the quality and the validity of the results. Your company's participation can stimulate others to participate. In other words, if nobody puts data in, nobody gets data out.
  • Granularity. If your chief competitors are participating, it may make sense for you to participate as well so that an industry cut includes more statistically significant numbers.
  • Validity. If your company is one of a few players in a region or an industry, your participation could make a significant difference in the validity of the data.
  • Cost. Participants usually receive a significant discount on the price of the survey.
  • Ease of use. Some surveys are easier and less time-consuming to fill out than others.
  • Security. Gauge your comfort level with the means of collecting the data. For instance, electronic mail is less secure than some forms of Web-based transmission; paper and pencil is less efficient but more comfortable to some respondents.
  • Added value. If you have participated in similar surveys over the same period, or have purchased similar surveys, make sure the survey in which you are considering participating adds value, such as some key jobs that are difficult to benchmark.

Job descriptions. Just as it is important to find surveys that compare companies of a similar stature, it's also important that the jobs being surveyed are comparable to the job being benchmarked. When consulting a compensation survey, match the job descriptions rather than the job titles, even if the survey uses generic or widely used job titles. For example, an associate could be an entry-level position at one consulting firm, or it could be the title for someone with an MBA at another. Companies are structured differently, and different companies use different names for the same jobs, so job descriptions are the best way to match positions. Beware of surveys that use only job titles, as it is unlikely the data will be a reasonable representation of the jobs you're interested in.

A survey job description should list the primary job function in one or two sentences, followed by key responsibilities. While the descriptions should be generic and not specific to any one company, they should contain enough information for participants to match appropriately to ensure the data is accurate. It is also important to match the organizational level of the positions be surveyed. A position that is at the group level at one company may be at the subgroup or the sector level at another.

Job titles are broken down differently in different surveys. Some surveys break them down by levels within the organizations, i.e., senior management, middle management, and entry level. Positions may also be broken down by job families or the types of responsibilities, i.e., business development, marketing, product management, and sales.

Compensation data. There are many things to consider when analyzing the compensation components of a salary survey. Because companies have different pay structures, compensation data is collected in ranges as well as actual pay. Salary surveys can provide employers more information on the marketplace and how to set competitive pay without overpaying or underpaying employees. Surveys should ask for the minimum, midpoint, and maximum for the surveyed positions, in addition to the actual base salary paid.

Usually, the prevailing practice for any one job is to pay a range of incomes. As a result, although the median pay for a job is likely to be a definable number, the range is just as important. Companies pay employees differently for various reasons. It could be the company's pay philosophy; or it could be the geographic location or the industry practice; or it could be the incumbent's length of service or proficiency in the job. Whatever the reason, it is unlikely that two companies will pay an employee doing the same job exactly the same amount.

When reading the base pay figures, it's important to check how the numbers are calculated. The surveying parties can dictate to the participants how the numbers should be reported. Salaries can be on an annual, monthly, or hourly basis. For example, if the incumbent is a contract employee, hourly salaries are more relevant than an annual figure. The survey may request pay data for individual incumbents or averages for all incumbents matching a specific job description, depending on the types of surveys and their objectives.

Incentives/bonuses. Look at both the actual annualized payments and the target level expressed as a percentage of base pay when evaluating incentives or bonuses. This allows for adjustments for atypical incentives and bonuses. Be sure to understand what is included in this figure and how it's collected. Although there is not a right or wrong definition of what is included in this category, it is important to understand how your numbers compare with those reported. In that sense, you need to know what it represents.

Other payments. As compensation changes, salary surveys are changing to include other forms of compensation such as profit sharing and stock grants. For more senior-level positions, long-term incentives are just as important as base salary. For example, an executive's compensation package at a startup company can be made up of mostly stock options rather than cash compensation. For a survey to represent the total compensation, it needs to take into account the cash valuation of stock options.

It is important to spend a little time learning how these stock option numbers are reported. It is also important to note that with stock options, the value may be a number, such as grant value, that is presented as a dollar amount but is not a present value and therefore cannot be added to base pay and incentive pay to provide a total direct compensation number.

Effective date. For those surveys conducted on a regular basis, such as annual surveys, the effective date will be until the next survey is released in the following year. Otherwise, knowing the effective date of the survey can prevent companies from using outdated salary figures and causing error in pay budget forecasts.

If the survey is not current, the person using it should age the salaries to the current date. If a survey was conducted in September, the salaries are likely to be as of September or even August. If you are using the survey in December to benchmark for a new position in the company, you will have to age the number. A simple way to do this is to take the annual rate at which salaries are moving for this job and prorate it, salary increases overall this year are around 3.5% but this may vary by job title.

A similar approach is used in setting pay levels across a company. Sometimes these figures are set at the beginning, middle, or end of the company's payroll year by aging the appropriate compensation data to those dates.

What about nontraditional sources of data, such as individuals?
Traditionally, pay data is collected by sending forms to human resources professionals and sometimes business managers, those most knowledgeable and authoritative when it comes to pay within their companies. Data provided by corporate representatives is more accurate than data reported by individual employees because employers have a strong business incentive to report data accurately and consistently for a wide range of jobs. They also have experience and understanding of the process so that the data they provide tends to be relatively "clean."

The growth of Web-based data collection methods has made it technologically feasible and cost-effective to gather compensation information from individuals. These new sources often report data that is valid and real. It is, of course, not the same as data collected and reported by trained compensation professionals; and like traditional sources, these alternative sources vary in depth, quality, relevance, and other measures of integrity.

Conventional wisdom has always dictated that compensation data from individuals and recruiters is unavoidably, perpetually biased. Yet this hypothesis has been difficult to test because the data has not been prevalent. Although some see an incentive to exaggerate one's own salary or that of one's most recently placed candidate, a significant misstatement could backfire. Further, by assuring confidentiality and providing additional information, individuals could be convinced to provide very accurate data. Data from alternative sources, such as recruiters or individuals, can be good or bad. When good, recruiter data can be used as an accurate indication of what new-hires are being paid, and individual data should approximate the general market. However, both forms of data are still clearly different from company reported information. Combining the different sets of numbers may be deceiving but comparing them side-by-side can be revealing.

Why pay for compensation data when it's available for free on the Web?
Compensation information is becoming widely available on the Internet, either for free or for a fee most individuals can afford. Many of these sources provide accurate, timely information. Yet just as different types of financial services firms fill different consumer needs at different prices, not all data products are alike. A Web-based brokerage business might provide the tools and information a user needs to make investment decisions, but without the one-on-one, custom consulting to shape a personal investment strategy. While some investors are satisfied with data only, others are willing to pay for the added value these consulting services provide.

Similarly, companies are willing to pay more for compensation data that provides the granularity of detail they need or the information that's hardest to find. They look for the name and reputation of the organization conducting the survey; the number of companies surveyed; the number of "incumbents," or employees covered by the study; the names of the companies that participated in the survey; and other measures of data quality and relevance to their industry. A middle-market professional compensation data tool might offer market-pricing information on commonly priced jobs, providing aggregated information based on primary and secondary research and analysis.

The top-of-the-line product in the compensation data business is a custom study of what each individual job within a company should pay based on very specific, targeted market data. A large company could pay hundreds of thousands of dollars, or more, for this type of in-depth companywide analysis.