Home > Exclusive Articles, Leadership > A Supply Chain Approach to Workforce Planning

A Supply Chain Approach to Workforce Planning

planningPeter Cappelli

George W. Taylor Professor of Management
The Wharton School – University of Pennsylvania


Workforce planning wasn’t always an afterthought. ‘‘Manpower plans,’’ as they were known, had long been a crucial component of overall business planning. Their roots were in the World War II War Manpower Commission, which required businesses to report on expected staffing levels and requirements to prevent shortfalls in skilled workers that could derail production and the war effort. By the mid-1960s, a study of personnel departments found that 96% of corporations had a dedicated manpower planning function. The assumption in these models was that the supply of talent was within the control of the company, an internal function. The plans began by estimating what the internal supply of candidates would be in the future for each position and then matched that to assumptions about company growth in the demand for talent. Because the supply for all but entry-level jobs came only from within, managing that supply involved some hiring but focused on internal advancement and the rate at which candidates progressed from one job to the other
The peak of workforce planning was probably a late 1960s model called MANPLAN, which attempted to model the movement of individuals within a career system by including in the forecasts individual behavior and psychological variables such as worker attitudes and aspirations, the practices of supervisors (appraisals, compensation arrangements, practices for employee transfers), the group norms in place in the workplace, and the situation in labor markets. These individual level estimates were then aggregated up to the company level to produce overall estimates. Arguably nothing more sophisticated has been created since.


Workforce planning and the related practices of employee development began to fall apart when the ability to forecast the overall level of demand in the economy eroded following the oil shocks in the mid- 1970s. Gross National Product, which had been forecast to grow at a rate of about 5–6% in real terms as it had during most of the 1960s, actually declined in 1974, 1975, and again in 1980. The 1970s became known as the decade of ‘‘stagflation,’’ low economic growth despite inflation. The recession that followed in 1981 was the worst downturn in business activity since the Great Depression. Gross National Product fell by two percentage points in 1982 alone. The manpower forecasts turned out to be incredibly wrong, as were the development plans based on them, because they were based on those forecasts of growth. The talent pipelines in the great corporations, which had lead times of 10 or more years, turned out about 6% too many managers every year.
The layoffs that followed were brutal and well known. But companies also slashed their own workforce planning and talent management functions. A study in 1984 documented the decline of these practices: About 30% of employers used elaborate statistical regression models to forecast talent needs in 1978, but that figure fell to only 9% by 1984; sophisticated Markov Chain vacancy models fell from 22% to 6%; and operations research tools in general declined from use in 23% of employers to only 4.5%. A Conference Board study based on companies that were interested in talent management reported that even in this group, fully half reported that their workforce planning efforts were ‘‘ad hoc’’ (the idea of ad hoc planning is something of an oxymoron), and only 19% of these companies, mainly the oldest and largest corporations, conducted workforce planning of any kind in the mid-1990s. This is compared to the 96% of companies that had a dedicated workforce planning department in earlier decades.
How could companies survive without workforce forecasts? For more than a decade following the 1981 recession, the priority in most companies was to ‘‘restructure’’ and get rid of talent. Because cuts could be made instantly, forecasts were unnecessary. When employers needed talent, the glut of experienced talent thrown on the market by downsizing meant that it easy to hire just what was needed when you needed it. No planning was required. Most companies abandoned their planning processes, and new companies never learned how.
An absence of planning worked okay as long as the problem was just having too much talent. Employers no longer had to worry about the risk of carrying excess employees because they just laid them off. Shortfalls of talent happened less often but were also not much of a problem as long as employers could hire whatever they wanted whenever they needed it. After the longest period of economic expansion in US history in the 1990s, that excess supply of talent evaporated. Outside hiring then became uncertain and expensive. Gerard Brossard, then vice-president for workforce planning at Hewlett-Packard Co., was one of a handful of innovators who built support for a new talent management process with senior executives by comparing what the businesses got in terms of talent with what they ultimately needed. The fact that the estimates were so often wrong, and the costs of being wrong became so big, created support for a different approach.


Workforce planning involves two types of forecasts. The first is internal, what our workforce will look like in the future if we do nothing new: how many incumbents will we have with relevant competencies in each area in the future? The growing rate of attrition has complicated this forecast because attrition is driven by factors largely outside the control of the organization, mainly poaching by competitors. The other complication is that job requirements are broader now, and teamwork and other systems have made employees more interchangeable, offering a great many more options for meeting the demands of the organization.
An example of how to deal with the latter complication comes from Electronic Data Systems Corp. (EDS), the information technology consulting and outsourcing firm. Like other professional service businesses, people are the product at EDS. Individual employees can work on a variety of projects and can team with others to produce a huge range of organizational competencies. The task of line managers is to put people together to create those competencies and then match them to projects. Mary Young at the Conference Board describes EDS before 2004 as having no central planning for talent. Business was growing fast enough that they simply hired people as fast as they could. The belief among executives before then (and in many other companies) was that the technology in information systems changed too quickly to bother attempting to forecast talent needs.
The real challenge for EDS and companies like it is in understanding their own supply: What competencies do we have, and in what ways can they be deployed to meet which customer demands? Answering this question requires understanding the differences across individuals with similar job titles. The planning asset for EDS in this regard is its skills inventory, which keeps track of the competencies that each employee has as measured by their capabilities with different software and programming languages, previous tasks they have performed, roles they have held in the company, etc. To keep this current, the company requires that the individual employees update it after each assignment. The line managers then use the skills inventory to make assignments. The inventory also includes salary information, which affects billing rates and charges to the clients and helps make the most cost-effective assignments. When there are gaps projected between the current supply of talent and projected demand, managers have the option of hiring to fill them, the usual approach, or rearranging some of their current employees in different ways, based on the skills inventory, or using contractors.
Corning Inc. also developed a different approach to workforce planning. Matt Brush, then director of global staffing and planning for Corning, noted that the need for a new model began during the boom years— when the company could not anticipate or even keep up with hiring demands, a sign that workforce planning wasn’t working. When the telecommunications bust hit in 2001, the company lost 90% of its market capitalization value, and the goal shifted quickly to cutting staff. ‘‘One of the things we learned was that we had to get better at anticipating demand, and to do that in staffing we had to move away from being ‘‘order takers’’ to helping the business units figure out what they truly needed.’’
The supply side of the talent-planning process at Corning presses line managers not just to plan but also to think through where the value is truly coming from in their operations. Rather than simply extrapolating from past staffing levels, the idea here is to rethink how valuable each of the current roles or positions in the organization is, essentially taking a fresh look at the status quo. If done right, the company understands which roles are important to reinforce as well as which competencies – and the individuals who have them – need to be redeployed.What is perhaps most impressive about the Corning approach is that it manages to get business leaders to think seriously not just about the future but also about their current supply of talent: how many of their current staff truly have the skills they need to do the job now, how many could be expected to advance into more senior roles, etc.
Cynics will notice that this is what performance appraisals are supposed to do, but they rarely work. What is different here, as Brush points out, is that it avoids the part that supervisors find difficult to do, and that is confront the employee with the assessment. Because the general discussion is just about ‘‘roles needed,’’ it is a little easier to be objective about talent. And because some of the options include retraining and moving individuals whose skills will no longer needed in their current roles, rather than dismissals, the business leaders are much more likely to be provide an accurate assessment of their workforce. While the results of dialogue with business unit leaders rarely lead to perfect estimates, at least they offer a general direction—we need more of these competencies and fewer of these. That at least allows the staffing function to begin the process of identifying candidates and finding talent. It is relatively easy to add or subtract the number of positions to be filled at a later date when the forecasts become clearer. In the absence of this process, Brush found that line management would consistently underestimate the talent they needed in the future.


The second and more difficult forecast in the planning process is to predict the demand for talent, what the organization will need to meet its business objectives. As noted earlier, the competitive environment for businesses is so changeable, and firms adjust their own strategies and practices so frequently that these estimates are rarely accurate. And they getmuchworse the farther outone goes.The error rate in theU.S.ona 1-year forecast of demand at the stock keeping unit (SKU) code or individual product level, for example, is over 30%. Long-term forecasts are essentially worthless in all but the most predictable operations. Even in stable industries like public utilities, an unexpected chief executive officer (CEO) change or a restructuring program can throw all plans up in the air.
Dealing with this uncertainty begins by borrowing techniques from supply chain management, where the overall task is similar to workforce planning: How do we ensure that we have just the right supply of parts or components to meet demand when that demand is uncertain? Among the newest developments in workforce planning is the shift from forecasting to simulations. Dow Chemical Co. is one of the first employers to make this shift. Dow moved to a system that exploited standardized data from its enterprise resource planning system to produce estimates for each location that could be aggregated for the company as a whole. Then it sought a university partner to develop an even more elaborate model, one that used the standardized data to generate estimates for each business unit. The forecasts incorporate a wide range of site-specific factors such as estimates of the political and business climate in each of its countries of operation, changes in labor and employment legislation, and business plans for the operating unit, which include targets for operating productivity. Those forecasts were then aggregated into company-wide estimates.
The advantage of modern computing power with a model like this is that estimates are generated instantly, which allows one to vary the assumptions to see what happens. Playing around with the assumptions basically turns a forecasting model into a simulation: What happens to our forecasted headcount, e.g., if the economy slides below our assumption or if new competitors enter a market? The ability to simulate allows business leaders to see the implications of different strategies for talent, to anticipate how talent constraints could impact those strategies, and in some case, to adjust their business plans if the talent requirements are too extreme.
Arguably the most sophisticated workforce planning is being done by Capital One, the innovator in the credit card business. Capital One had just under 20,000 employees in 2001, it then made outsourcing and other decisions that brought headcount down to 14,000 by 2005, and later with a series of acquisitions saw employment rise to 30,000 by 2007. The need for better talent planning was driven by the costs incurred when employment changed so rapidly.
Capital One made its name in the product market through sophisticated analysis of customer data. Pat Cataldo, a vice-president in the human resources (HR) area, describes the culture of the company as one of ‘‘test and learn,’’ where analysis and application is encouraged. Prasant Setty came into that HR group from Wharton and then McKinsey & Co., with the challenge of improving workforce planning. He assembled a team from fields like marketing and operations research (no traditional HR experts). They used data mining techniques, systems dynamics models from manufacturing, and information from their PeopleSoft system to generate talent-planning models for each business unit in the company. They modeled outcomes like attrition rates, employee morale, and rates of promotion and outside hires. Among the factors they consider in their models are aspects of the organizational chart—span of control, levels of hierarchy, which affects promotion rates, and ‘‘stretch roles,’’ positions that are reserved for developmental assignments.
The HR planning team works with the business unit leaders to develop models around their particular business plans and goals in an effort to align the talent management practices with those business goals. As with Dow Chemical, the forecasting models were easily turned into simulation models, and this is where the power of the analyses comes in. The models allow the line managers to see the options involved in achieving any business plan: If you are planning to grow at 10% this next year, here are the talent requirements needed to achieve that growth. If you do it by accentuating outside hiring, here’s the likely effect on reducing prospects for internal promotion and the associated effects on morale and then on attrition. If you change the span of control, here is the effect on talent needs at the management level, but also the effect on promotion rates. Most important for the Chief Financial Office, the models also allow the managers to see the total compensation implications of all their choices. The line managers see the talent management issues as a system and also see how their choices concerning any single outcome in that system affect the other aspects.
Prasant, now at Google, reports that, rather than resisted by the business managers, this approach has been embraced by them as a useful tool because it takes what are often very general statements of business goals and shows their concrete implications. And it has raised the status of the human resources function in the process by bringing these issues into the beginning of the planning process. Other companies, like Citibank, have also embraced the simulation approach. Its talent-planning models allow them to calculate, for example, the human capital requirements and costs of opening a new branch office as opposed to expanding others in the area.


Simulations help us manage uncertainty by recognizing it, providing a sense of how outcomes could vary based on changes in assumptions. The best outcomes from the simulation process occur when we can identify implications that are robust to several different assumptions, suggesting applications we should definitely pursue, versus implications that are highly variable to changes in context. This information is most useful, though, when we have one other piece of data that has not previously been part of the workforce planning process.
The additional piece of information we need to make the best decisions is to know the costs of being wrong with our forecasts. What is important in answering this question is first to recognize that there are two ways we can be wrong: Not enough talent or too much. In the language of operations research and supply chain management, these problems of undersupply and oversupply are collectively known as ‘‘mismatch costs.’’ The insight comes from the fact that the costs of being wrong in each of these two directions are almost never the same. The goal in supply chains and in workforce planning is the same, to deliver just the right amount of supply to meet demand, neither falling short nor going over. That goal proves almost impossible to do. The next best outcome, however, is well within our grasp, and that is to minimize the costs of being wrong, the mismatch costs.
It might seem reasonable to assume that the two types of mismatch costs will balance out because we will either undershoot or overshoot the estimate: we can’t do both at the same time. This is what most workforce planners implicitly assume by worrying only about what we call the ‘‘point estimate’’ of demand, that is, the precise number our forecasts generate. In other words, if the forecast predicts that we will need 100 computer programs in our division next year, then we try to deliver exactly 100 programmers and stop there. This approach either assumes that we are absolutely certain about what we need, that there is no error, which is almost never true, or that the probability of overshooting the forecast is equal to the probability of undershooting it (i.e., that the probability distribution of the forecast is normal), so that even if we are wrong, focusing on the point estimate is the way to go. But this approach also assumes that the costs involved in overshooting will be the same as the costs of undershooting the forecast: If we are short 10 programmers, it is as big a problem as if we have 10 too many programmers. And that is almost never the case.
In previous generations where there was no real alternative to internal development, having a shortfall of talent was a much, much bigger concern than overshooting. Having too much talent simply led to a deeper ‘‘bench,’’ candidates who had to wait before stepping into a role that made use of their abilities. As a result, most planners, if they recognized uncertainty as an issue at all, tried to set their plans to make sure that they didn’t fall short of talent, even if it meant creating an excess supply in most circumstances.
Outside hiring and the related issue of attrition have now reversed those costs. The cost of undershooting a talent forecast are much less than theywere in the past because they can more easily be offset by outside hiring to fill in any talent gaps that occur if actual demand overshoots forecasts. The costs of overshooting and having an inventory of talent, on the other hand, can be much greater. One reason is because of retention problems. Asking qualified employees to ‘‘sit on the bench’’ and wait until an opportunity comes open for them is a recipe for disaster when the best of those employees routinely get calls headhunters offering themopportunity rightnow,nowaiting,nouncertainty. Another drawback to deep benches is the periodic pressure from operating managers to ‘‘restructure’’ and look for ‘‘fat’’ to cut in order to lower short-term costs. Underused talent can look a great deal like fat to a restructuring agent, and the difference between ‘‘fat’’ and investment in a ‘‘deep bench’’ can all be in the eye of the beholder. In this context, overshooting demand is the bigger problem.


Fortunately, the information needed to produce these more refined workforce plans is not all that hard to come by, and the analyses are straightforward. At present, most employers don’t even recognize that their forecasts are uncertain and make no attempt to estimate mismatch costs. So our estimates don’t have to be perfect to represent an enormous improvement over the status quo.
Let’s begin with the hardest task, estimating the uncertainty around our workforce forecasts. We start with whatever forecast we already have for talent needs. Spending a lot more time and resources trying to generate a better point estimate of our talent needs is unlikely to do us nearly as much good as learning about the uncertainty of that forecast and the costs of being wrong. The reason is that business demand for most employers is subject to so many factors that are effectively unpredictable (e.g., will the economy be up or down in 3 years, will a competitor invent a new product, will the board of directors appoint an outside CEO with a different vision?) that additional resources are unlikely to add much precision to our forecasts.We can learn a lot more by getting a quick sense of how uncertain that best estimate is.
The easiestway to do this is to look at the forecasts in previous years and see how accurate they were: What did they predict, what did actual demand turn out to be? For those employers that have no prior workforce plans to assess, a very good substitute is the overall business forecast for the organization as compared to actual demand. We then calculate the ratio of what actual demand turned out to be to the forecast of demand. If, for example, actual demand last year turned out to be 900 units while we had forecast 1000, then that ratio is 90%. A reasonable assumption, therefore, is that the error rate this year will be about the same, about 10% of this year’s forecast.
If we have forecasts for several years, we can calculate the average error rate of our forecasts in the past and use that to generate our estimate of demand this year. If we don’t have the error rate for prior forecasts and don’t have prior workforce plans against which to work, we can at least estimate the standard deviation of actual demand: How much did the level of business vary year-by-year over the past decade?
Say that our forecast for next year calls for adding 1000 middle managers to the workforce. The average error rate on prior business forecasts is 10%—actual demand tended to differ fromour estimates by 10%. So we conclude that we might need from 900 to 1100 new middle managers. We can also use the standard deviation to calculate the odds that the actualdemand will be any particular number, say, 900 middle managers, or any range, say between 900 and 1200 middle managers. (Doing this last step requires a few assumptions, especially that the distribution of the actual demand over time will follow a normal distribution, and the use of standard normal distribution tables from statistics.)


My colleague and decision theorist Paul Schoemaker suggests engaging the line and operating managers in a process to recognize the uncertainty of their own forecasts for the business with the following exercise:

• Ask them first to identify the four most important assumptions behind their forecasts.
• Ask them to array those assumptions on a twoby- two graph—with importance on one axis and how certain they are about the assumption on the other.
• Then ask them to establish a 90% confidence interval for their forecast. For example, ‘‘our forecast is that demand in this division will grow by 7% next year, and I am 90% certain that the actual demand will lie between 5% and 8%.’’

The purpose of this useful exercise is to force participants to acknowledge that their forecasts are not very accurate and to recognize that the efforts to map human capital onto those forecasts will have to address the underlying uncertainty in demand.


The next step in the process is to estimate the costs of falling short or going long in our estimates, what happens if we have too many middle managers or too few? Few employers have data like this handy, so we will probably need to generate some rough-and-ready estimates simply by asking the line and human resource experts the following questions: What do we do if we need more talent than we have planned for? Do our projects stop, can we hire on the outside, subcontract, or outsource? In other words, what are the costs per vacancy of our next best alternative to staffing these positions internally? Similarly, what happens if we have more staff than demand requires? Can we find something else useful for them to do, and how does this affect the risk of turnover? More generally, we need to calculate the cost per excess worker. These questions are hard to answer with precision, but for the purpose at hand, the important issue is simply which direction is more costly for us to err and by roughly how much is the difference. A useful answer would be something like, ‘‘the cost of having too many employees is roughly twice as great as the cost of having too few.’’
These mismatch costs are likely to differ not just across organizations but also by job. Because the cost of falling short is typically the additional cost of outside hiring to make up the gap, the mismatch costs associated with hiring too few employees can be reasonably small for lower level jobs where the competencies are readily available on the outside market and internal wages are close to that market rate. For jobs with skills that are harder to find and more unique to the organization, the costs of undershooting are much higher because it is more difficult to find qualified candidates on the outside market. Those costs include the cost of outside search as well as having to pay a market premium for compensation and the possible costs associated with ‘‘on boarding,’’ getting the new hire up to speed with the culture and tacit knowledge needed to operate in the new organization. Finally, we need to estimate the risk that outsiders will not do as well as internal promotions—because we are less sure about their capabilities, because questions of culture and fit are more difficult to assess, etc. These may also differ by job.
We use this information on mismatch costs to make a further adjustment to our workforce plans. If the costs of falling short in talent or in going long are about equal, then we might want to stick with our best point estimate of demand, even if that estimate is uncertain. If the costs are unequal but our estimate is highly accurate, with little anticipated error, we might also want to stick with our point estimate. But in all other cases, which represent the vast majority, we need to make adjustments. The bigger the error rate and the bigger the cost difference between going long and falling short, the more we have to adjust our estimates.
Say that our original forecast for middle managers of 1000 was adjusted to be between 900 and 1100, based on a 10% estimate of forecasting error. Say we also conclude that the costs of having too many managers is twice as great as the cost of having too few because the extra cost of hiring some replacements if we fall short is nowhere near as big as the risk of losing the investment in surplus middle managers who are inclined to quit. Now we have the information needed to adjust our estimates of demand. We should develop less than 1000 middle managers to reduce the chance that actual demand will fall short and that we will have a surplus.
Exactly how many fewer we should develop could be estimated with some precision from our estimate of demand if we know the standard deviation of demand, which again requires some assumptions and the use of statistical tables. But a ‘‘seat of the pants’’ estimate will often do. If we think our revised estimate is pretty accurate, we may want to undershoot it only by a little, say develop 900 managers internally because the risk that actual demand will fall short of that is small. If, on the other hand, we think our 1000 estimate is really closer to a guess in which we don’t have a lot of confidence, we might want to go well below—develop only 500–600 managers. The reason, again, is that we want to avoid overshooting actual demand. While such estimates are certainly rough, they clearly beat the alternative of ignoring the problems caused by mismatch costs.
Vivek Gupta, senior vice-president at the Indian software company Zensar Technologies, reports that his company went through something like this thought process in deciding how many computer programmers to train in order to meet their anticipated demand. Fearing retention problems, they initially undershot their best estimate and filled in the gap with outside hiring. As the labor market began to tighten and the ability to hire from the outside grew more difficult, the mismatch costs changed. Now it was more costly to undershoot demand, because shortfalls could not be filled on the outside—which meant that they likely could not get their projects done if their forecasts fell short of actual demand. So then they switched their plan and increased the proportion of talent they developed internally, exceeding their estimate of demand to ensure that they do not fall short of any unexpected increase in business.
This approach to workforce planning gives us a guide not only for how many candidates to develop internally but also how many we should expect to hire from the outside. Where the costs of overshooting are greater, as they typically are now, we aim to undershoot actual demand, which means that there is a good chance that we will need to do some outside hiring to make up the gap. How much hiring we will need is clearly uncertain, but we can get a rough idea by comparing our adjusted forecast of demand (1000 middle managers in the example above) to the number we decide to develop internally based on our assessment of mismatch costs (900 or perhaps 500 for highly uncertain estimates). The difference between those estimates gives us a good guide as to the amount of outside hiring we might end up having to do.
What about those jobs for which there is no internal development, where all vacancies are filled from the outside, such as associates in professional service firms, which hire a new class of associates from college every year. The process is the same: Actual demand is uncertain, the forecasts are measured with error, and there are costs to being wrong. One important difference is that the costs of falling short of the forecasts could be greater because the alternatives may be more limited. It is possible for an employer to address the problem of falling short by going back into the market at a later point in the year and hiring again. Some consulting companies have essentially moved to hiring two classes per year precisely to address that problem. If additional hiring is not an option, temporary help or contracting might be.
Leased or temporary employees and contract work can lower mismatch costs further because they can be deployed at the last minute should actual demand overshoot forecasts. They can also be more easily taken out should demand fall in the next period and fall short of estimates. As a result, we might think of a second decision framework within the general make versus buy choice for getting work done in organizations. An employer might decide to undershoot their expected demand for talent in terms of the amount of internal development they are willing to do. Some proportion 13 of the remainder of the work, the component that is most predictable, might be met through outside hiring. The remaining component that is even less certain might be met by outsourcing and using leased employees. The most flexible components cost the most per unit of work completed, and the employer is paying a premium for the more flexible components to avoid either undershooting or overshooting the actual demand for talent. The more accurate the forecast, then less risk there is of mistakes and the fewer of the more flexible arrangements the employer needs.


Now we have a means for beginning to manage the uncertainty in workforce planning. More accurate estimates on the supply side, especially of competencies at the level of the individual worker, allow us to begin with a better sense of where we stand in terms of talent. Simulations of demand allow us to see the talent challenges associated with business plans and adjust business strategies to them. They also give us a clearer sense of how actual demand might change if any of the assumptions built into the models turn out to be wrong. From that point on, our approach looks at the uncertainty of prior estimates and the mismatch costs of being wrong about demand to adjust our forecasts to the risks of uncertainty. One very important aspect of this approach is that it does not require that one construct elaborate forecasting models of the demand for human capital. Because such forecasts are derived from the business environment, which is highly complex and extraordinarily difficult to anticipate, it is hard to make them better. Rather than assuming that we have certainty about the future, which is how most forecasting models are used, this approach recognizes and then comes to grips with the uncertainty that is inherent in business forecasts.
A final thought about this process is to remember that it is not that difficult to get significantly better at workforce planning. At the moment, most firms do no planning, which essentially means that every development comes as a surprise and the only response is to react to those responses after the fact. The means for responding – outside hiring for shortfalls, layoffs for surpluses – has become expensive and difficult. Traditional planning, which relied entirely on forecasts, has proved equally unsatisfactory because those forecasts have proved to be so unreliable. In this context, a simple and straightforward approach to workforce planning that takes on the central challenge of uncertainty is highly useful.

Organizational Dynamics, Vol. 38, No. 1, pp. 8–15, 2009 ISSN 0090-2616
2008 Elsevier Inc. All rights reserved. doi:10.1016/j.orgdyn.2008.10.004 http://www.elsevier.com/locate/orgdyn

  1. No comments yet.
  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: